CN120410168A

CN120410168A - Business process visualization early warning and policy response method, device, equipment and medium

Info

Publication number: CN120410168A
Application number: CN202510508142.7A
Authority: CN
Inventors: 谢宏龙
Original assignee: Ping An Health Insurance Company of China Ltd
Current assignee: Ping An Health Insurance Company of China Ltd
Priority date: 2025-04-22
Filing date: 2025-04-22
Publication date: 2025-08-01

Abstract

The invention relates to the technical field of data analysis, and can be applied to business scenes such as financial science and technology, medical health and the like, and discloses a business process visualization early warning and strategy response method, which comprises the following steps: the method comprises the steps of setting a physical interrupt node to collect interrupt operation track data, generating multi-stage logic interrupt type structured data based on the operation track data, mapping the multi-stage logic interrupt type structured data into a visual topological graph, identifying a preset service policy check failure type, counting interrupt frequency and triggering a hierarchical early warning instruction, and generating a page element weight adjustment parameter and a fault tolerance threshold value set. According to the method, the visual topological graph is constructed, the high-frequency interrupt behavior is identified, the real-time identification and the hierarchical early warning of the failure of the service policy verification are realized, the self-adaptive adjustment of the page elements and the policy threshold value is further driven, and therefore the user insurance conversion rate is effectively improved, and the front-end operation loss risk is reduced.

Description

Service flow visual early warning and strategy response method, device, equipment and medium

Technical Field

The present invention relates to the field of data analysis technologies, and in particular, to a method, an apparatus, a device, and a storage medium for service flow visualization early warning and policy response.

Background

In the field of financial and scientific business, along with the rapid expansion of insurance products and the wide layout of online channels, more and more users apply for self-help insurance through mobile terminal H5 pages. However, in this process, a large number of users have entered the application flow, but eventually have not completed the application act, resulting in a higher proportion of the lost user population. Such users often have shown explicit intent to apply, and their behavioral paths and information carried by the breaking nodes have extremely high conversion potential and business value. However, in the prior art, enterprises generally rely on the traditional approach of embedding points on pages, only limited user operation data can be collected at preset positions, and fine granularity behaviors of the application flow interruption cannot be effectively captured and restored. The passive and static data acquisition modes are not only difficult to identify user behavior deviation in time, but also can not accurately position and visually model the intermediate nodes, so that enterprises lack active response means in the aspects of user conversion rate improvement and process optimization.

In the field of medical health business, the process of selecting health insurance products by users through an online platform is also complex and sensitive, and generally involves a plurality of key nodes such as medical history filling, physical examination information submission, health notification and the like. The current system generally cannot distinguish whether the user interruption is caused by specific factors such as failure of matching health information, deviation of understanding content, poor page response experience and the like. User churn behavior in medical settings not only affects conversion rates, but also may embed potential risk potential. Because of the diversity of causes behind the interruption behavior, the real intention of a user and decision blocking points are difficult to comprehensively reflect by depending on fixed rule embedded points, so that the support based on behavior logic is lacking in product strategies, page interaction optimization and risk model parameter adjustment.

Under the background of rapid development of big data and intelligent visual analysis technology, partial enterprises try to restore and analyze the application flow by introducing a mode of combining log embedded points with manual analysis, but the current means still highly depend on data developers to perform temporary data extraction and query operation, and lack standardized structured behavior modeling and cross-role cooperation mechanisms. In addition, the data processing flow is long, the accuracy is easily subjectively influenced by an analysis mode, the requirements of real-time early warning and flow intervention are difficult to support, and the method is particularly prominent in complex flows of multi-strategy parallel and multi-page skip. Therefore, in the prior art, a large technical blank and application blind area still exist in the aspects of user behavior path restoration, interrupt behavior visual identification and strategy level response capability.

Disclosure of Invention

The invention mainly aims to provide a business process visual early warning and strategy response method, device, equipment and storage medium, and aims to solve the technical problems that the prior art cannot carry out structural classification and visual modeling on business process interruption behaviors and realize hierarchical early warning and strategy optimization response based on interruption frequency.

In order to achieve the above purpose, the present invention provides a method for service flow visualization early warning and policy response, comprising:

Setting a plurality of physical interrupt nodes on a front-end interface of a business process;

Acquiring operation track data of an interrupt service flow through the physical interrupt node;

classifying the operation track data into multi-stage logic interrupt type structured data according to the sequence of the business flow;

mapping the multi-level logic interrupt type structured data into a visual topological graph;

Screening an entry with a logic interrupt type being a preset service policy check failure type from the multi-stage logic interrupt type structured data;

counting the triggering times of the preset service policy check failure types according to the interrupt timestamp fields in the screened items, and generating interrupt frequency data according to the interrupt timestamp fields, the triggering times and the preset service policy check failure types;

triggering a grading early warning instruction containing a preset business strategy version identifier when the interruption frequency data exceeds the upper limit of the baseline frequency;

And according to the preset business strategy verification failure type data associated with the hierarchical early warning instruction, generating a page element weight adjustment parameter and a fault tolerance threshold set corresponding to the preset business strategy version.

Further, in order to achieve the above object, the present invention provides a service flow visualization early warning and policy response device, including:

the interrupt node configuration module is used for setting a plurality of physical interrupt nodes on a front end interface of the business process;

the operation track acquisition module is used for acquiring operation track data of the interrupt service flow through the physical interrupt node;

the logic interrupt classification module is used for classifying the operation track data into multi-stage logic interrupt type structured data according to the service flow sequence;

The topology diagram construction module is used for mapping the multi-level logic interrupt type structured data into a visual topology diagram;

The interrupt screening module is used for screening entries with the logic interrupt type being a preset service policy check failure type from the multi-stage logic interrupt type structured data;

The interruption frequency counting module is used for counting the triggering times of the preset service policy check failure types according to the interruption time stamp fields in the screened items, and generating interruption frequency data according to the interruption time stamp fields, the triggering times and the preset service policy check failure types;

the early warning triggering module is used for triggering a grading early warning instruction containing a preset business strategy version identifier when the interruption frequency data exceeds the upper limit of the base line frequency;

and the policy optimization parameter generation module is used for generating page element weight adjustment parameters and fault tolerance threshold value sets corresponding to the preset business strategy versions according to the preset business strategy verification failure type data associated with the hierarchical early warning instructions.

Further, in order to achieve the above object, the present invention also provides a computer device, where the computer device includes a memory, a processor, and a business process visualization early warning and policy response program stored in the memory and capable of running on the processor, where the business process visualization early warning and policy response program implements the steps of the business process visualization early warning and policy response method described above when executed by the processor.

Further, in order to achieve the above objective, the present invention further provides a computer readable storage medium, where a service flow visualization early warning and policy response program is stored on the storage medium, where the steps of the service flow visualization early warning and policy response method described above are implemented when the service flow visualization early warning and policy response program is executed by a processor.

The invention has the beneficial effects that the invention relates to the technical field of data analysis, and can be applied to business scenes such as financial science and technology, medical health and the like, and discloses a business process visualization early warning and strategy response method, which comprises the steps of setting a physical interrupt node to collect interrupt operation track data, and classifying the data into multi-level logic interrupt type structured data based on business process sequence; and when the frequency exceeds the upper limit of the base line frequency, triggering a hierarchical early warning instruction containing a preset service strategy version identifier, and generating a page element weight adjustment parameter and a fault tolerance threshold value set of a corresponding strategy version according to the instruction. According to the method, the visual topological graph is constructed, the high-frequency interrupt behavior is identified, the real-time identification and the hierarchical early warning of the failure of the service policy verification are realized, the self-adaptive adjustment of the page elements and the policy threshold value is further driven, and therefore the user insurance conversion rate is effectively improved, and the front-end operation loss risk is reduced.

Drawings

The invention will be further described with reference to the accompanying drawings and examples, in which:

FIG. 1 is a schematic diagram of an application environment of a business process visualization early warning and policy response method according to an embodiment of the present invention;

FIG. 2 is a flow chart of an embodiment of a business flow visualization early warning and policy response method of the present invention;

FIG. 3 is a schematic diagram of functional modules of a preferred embodiment of the business process visualization early warning and policy response device of the present invention;

FIG. 4 is a schematic diagram of a computer device according to an embodiment of the invention;

fig. 5 is a schematic diagram of another structure of a computer device according to an embodiment of the invention.

Detailed Description

It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

The business process visualization early warning and strategy response method provided by the embodiment of the invention can be applied to an application environment as shown in fig. 1, wherein a user side communicates with a server side through a network. The method comprises the steps that a server side can set physical interrupt nodes through a user side to collect interrupt operation track data, classify the data into multi-level logic interrupt type structured data based on service flow sequences, map the structured data into a visual topological graph, screen entries with logic interrupt types being preset service strategy check failure types, count interrupt timestamp fields to generate interrupt frequency data, trigger a hierarchical early warning instruction containing preset service strategy version identifiers when the frequency exceeds the upper limit of a base line frequency, and generate page element weight adjustment parameters and fault tolerance threshold value sets of corresponding strategy versions according to the instruction. According to the method, the visual topological graph is constructed, the high-frequency interrupt behavior is identified, the real-time identification and the hierarchical early warning of the failure of the service policy verification are realized, the self-adaptive adjustment of the page elements and the policy threshold value is further driven, and therefore the user insurance conversion rate is effectively improved, and the front-end operation loss risk is reduced. The user terminal may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server may be implemented by a stand-alone server or a server cluster formed by a plurality of servers. The present invention will be described in detail with reference to specific examples.

Referring to fig. 2, fig. 2 is a flow chart of an embodiment of a business flow visualization early warning and policy response method provided by the present invention. It should be noted that although a logical order is depicted in the flowchart, in some cases the steps depicted or described may be performed in a different order than presented herein.

As shown in fig. 2, the service flow visualization early warning and policy response method provided by the invention comprises the following steps:

s10, setting a plurality of physical interrupt nodes on a front end interface of a business flow;

In this embodiment, the purpose of setting a plurality of physical interrupt nodes on the front end interface of the business flow is to analyze the precise location and type of the user behavior interrupt, so as to provide a structural basis for the subsequent interrupt attribution and early warning. The business process front-end interface refers to a collection of page elements that a user can interact with, including but not limited to form fill pages, confirmation pages, payment pages, or notification pages. The page structures correspond to specific flow steps, and controls, buttons, input boxes and the like on each page form key positions which can trigger the interrupt behavior of a user.

Physical interrupt nodes refer to explicit interface controls or behavior trigger points in the system that are capable of capturing operation termination behavior, which do not necessarily represent browser shutdown or process termination, but include behavior that can be perceived by the front end of the system, such as input termination, click-through, page pop-out, etc., such as clicking a back button, dwell time exceeding a threshold, incomplete input, active clicking to close a prompt pop-up, pop-out to other links, etc. Setting these physical break nodes requires the incorporation of structured identifications of page elements, e.g., based on unique identification paths possessed by each control element in a DOM (document object model) structure, such as precisely describing the corresponding control by XPath or CSS Selector.

The nodes are usually accessed in a configuration mode instead of hard-coded writing, the implementation process can be realized through automatic instrumentation of a front-end monitoring SDK, and also can be realized through analyzing and registering an interactive event monitor through a configuration file, and node identification logic is injected in a system rendering process. Each node needs to bind a unique identifier, and the identifier can comprise information such as a control ID, a page path, a control position sequence, a service type code and the like, and establishes a mapping relation with an interrupt event acquisition module in a back-end system.

In the implementation process, the system also needs to define a behavior trigger rule for each registered physical interrupt node, including trigger conditions (such as clicking, input defocusing, page jumping), data reporting content (such as user identification, device identification, time stamp, control identifier, and page context) and abnormality judgment logic. The trigger rule can be deduced through a front-end behavior modeling rule, and the enhancement configuration can be performed after the background automatic statistics of abnormal high-frequency nodes.

In one embodiment, a JavaScript SDK may be embedded in the front-end page, which scans the node identifiers in the configuration manifest during the page load phase, and binds the listening function to the corresponding DOM element event, e.g., click, blur, unload, etc. The configuration list may be in JSON format, supporting fields such as control path, trigger event, exception type flag, etc. The system can integrate the embedded point management background, drag the selection control in the page visual structure, automatically generate the node path and release the configuration, and realize development-free deployment of the operation and maintenance level.

The other mode is to use the non-invasive agent monitoring technology, combine with the browser MutationObserver or IntersectionObserver interface to monitor the control state change, and then analyze in real time according to the user behavior and judge whether to form the physical interrupt event. This approach is more applicable to scenarios where it is difficult for existing front-end systems to modify source code.

The interrupt node registration logic can be encapsulated in a common component life cycle function based on standardized front-end frameworks (such as practice and Vue) to ensure that interrupt nodes are deployed in standardized, inheritable and reusable modes. For a low-code platform, an interrupt control library component can be embedded, and a developer automatically registers interrupt identification and acquisition rules when dragging the component, so that the landing difficulty is further reduced.

In different business environments, such as medical health insurance systems, complex health information notification procedures may be involved, where the control interrupt node includes heart rate, blood pressure, past history fill-in areas. In the financial wind control decision system, the key points of the procedures such as risk assessment questionnaire, identity verification and the like are possible. For a multi-terminal deployment scene, unified control identification naming rules and platform difference adaptation logic can be designed, so that the mobile terminal and the desktop terminal can access the same interrupt perception system.

The method and the system have the advantages that in the insurance application business process, a user needs to complete a plurality of steps on line to realize policy generation. Through analysis of the buried point structure of the existing front-end interface, a large number of control nodes and operation behaviors which are strongly related to the interrupt behaviors exist in the whole insuring path naturally, and the control nodes and operation behaviors can be defined and captured as physical interrupt nodes. For example, in "terminate after open page" the system may embed a physical interrupt node in the page load event, which may be recorded as a non-interactive termination when the user exits the browser without any valid clicks after the page initialization is complete. In the "front-end-of-lead" link, the user may leave without clicking the "immediate-application" button after the marketing page or product introduction page is viewed, and the system forms a physical node for determining interruption by the exposed and non-triggered state of the button. Further, in the steps of "quote fill out page terminate", "health advise page terminate", etc., the page contains multiple high risk fields and strong interactive controls, such as age input, disease history selection, BMI fill out, etc. After the corresponding physical interrupt nodes are configured on the controls, the system can sense whether the user interrupts the flow in a certain field operation in real time or abandons the submission after verification fails. In addition, the actions such as uploading an identity card photo in the 'real name authentication page termination', prompting a button for OCR recognition failure, selecting a card to be inactive in a payment mode in the 'before payment page termination', and the like can be realized through physical interrupt node binding through an actual control event. Whenever the user performs actions such as jumping out, closing a page, staying overtime or sliding quickly at the controls, the system automatically triggers the interrupt event acquisition, and fields such as a control identifier, an interrupt timestamp, a front-back interaction track and the like are attached. The setting of the physical interrupt nodes not only covers the complete path from the guidance to the payment, but also has the characteristics of being measurable, traceable and retrospective, and provides basic support for the subsequent construction of interrupt type structured data and the hierarchical early warning. The nodes are interaction components which truly exist in the front-end interface, have definite event monitoring points, and can realize interrupt sensing in a non-invasive mode, so that the closed-loop landing of the whole visual early warning method is supported.

By disposing the physical interrupt nodes on the front end interface of the business flow, the system can accurately sense the interrupt behaviors of the user on each control and each path in the flow, so as to provide an original event basis with clear structure and clear identification for subsequent data acquisition, path analysis and early warning modeling. The accuracy and the controllability of interrupt positioning are improved, so that interrupt cause analysis does not depend on subjective feedback of a user or fuzzy inference at the rear end any more, and technical support is provided for business process optimization.

S20, acquiring operation track data of an interrupt service flow through the physical interrupt node;

In this embodiment, in order to realize full-flow observation and reason identification of the interrupt behavior, the deployed physical interrupt node needs to be used to perform real-time sensing and data acquisition on behaviors such as unexpected exit, flow disconnection or page jump and the like, which occur in the page operation process of the user. The physical interrupt nodes are generally embedded into key interaction positions in the form of a front-end control monitor, a page event catcher or a user behavior recording script, such as button clicking, field input box defocusing, page loading failure, window closing or interception before refreshing, and the like, and the interaction behavior triggered by a user is captured through an embedded monitoring mechanism and is in linkage judgment with a flow control state, so that whether an interrupt event is formed or not is marked.

In the acquisition process, the system can carry out association record on the interrupt event and the occurrence context thereof, wherein the association record comprises a current page URL, a device type, a user identifier, a network state, an interrupt timestamp, a last operation node identifier, a control element ID for triggering interrupt, an operation type (clicking, jumping out, refreshing and the like), browser environment parameters and the like. The data are pushed to a server data receiving interface by the front-end interrupt node through the embedded point script, the server system carries out structural encapsulation on the received event information, and field names, time formats and event classification standards are mapped uniformly according to configuration rules so as to ensure consistency and traceability of multi-terminal data.

In addition, in order to enhance the integrity of the operation track and the accuracy of interrupt identification, the acquisition mechanism also introduces state retention logic for recording the behavior sequence before interrupt, including information such as an access start point, a historical page circulation link, a filled word segment value, operation delay behavior, focal point stay time and the like. These additional content will support subsequent behavioral chain modeling and flow disruption position fine positioning analysis as part of the track sequence.

The whole process of data acquisition needs to have low delay and high robustness, and the complete data delivery is ensured while the operation experience of a user is not influenced, so that the design of a front-end script needs to adopt an asynchronous non-blocking communication mode, and a buffer queue and a retransmission mechanism are combined to cope with a weak network scene, so that the interrupt event of a key node is ensured not to be missed. The acquisition module is usually decoupled and deployed with the front-end page rendering service, so that unified access and strategy updating are facilitated, meanwhile, the acquisition granularity and the starting state are supported to be controlled remotely through the configuration center, and the dynamic adjustment requirements of different stages of business on the acquisition precision are met.

In actual deployment, the operation track data acquisition function completes automatic loading and event registration by the front-end buried point SDK. The development team will identify all critical physical interrupt nodes based on the page prototype graph and the service path definition, and insert a unified data binding instruction in the front-end template, such as configuring an attribute data-interrupt-node-ID for each control to mark its node ID. After the SDK is started, all the bound interrupt nodes are scanned, corresponding monitoring functions are registered based on a preset event type, for example, click events are registered for button controls, the blast events and change events are registered for input controls, and the beforeunload events are registered for the whole page to sense forced exit.

When the occurrence of an event is monitored, the SDK reads fields such as a control identification, an event type, an occurrence time, a page URL, a user SessionID and the like, and simultaneously adds all the completed form field abstracts in the current page to form an original track event object. The event object is sent to the interrupt track collection service at the back end, and the collection service judges whether the event object belongs to an interrupt event or not and writes the interrupt event into the track analysis database.

For the adaptation of different devices and service ends, an end-to-end data acquisition channel can also be introduced. For example, when RN or Flutter is used in the mobile terminal APP, related components can perform interrupt monitoring in a form of active bridging, a WeChat applet terminal can use onUnload to cooperate with route monitoring, and a PC browser can use visibilitychange to cooperate with performance detection to perform departure judgment.

By precisely binding the perception entrance of the interrupt behavior to a specific physical interrupt node and combining event monitoring and context information expansion, the high-precision and high-coverage acquisition of the interrupt operation is realized, and the problem of fuzzy path identification based on the traditional back-end log speculation is avoided. The real-time reporting of the interrupt behavior is realized, and a basic data source with complete structure and definite granularity is provided for subsequent data classification, frequency statistics and interrupt visualization.

S30, classifying the operation track data into multi-stage logic interrupt type structured data according to the service flow sequence;

In this embodiment, the operation track data refers to a series of operation event sets generated by a user through a front page or a control in a business process, and generally includes event types such as page stay, control clicking, content inputting, request submitting, and the like, and is accompanied by field information such as event triggering time, interrupt time stamp, control identifier, business process node number, and the like. The core of classifying the operation track data according to the service flow sequence is to construct a logic execution path with time consistency and node sequence consistency, so that links where source interruption occurs and the cause of user interruption behavior can be traced. The determination of the service flow sequence may be based on a preset flow definition file that expresses the logical sequence of the stages of the service in the form of a graph structure, a flow linked list, or a node tree, and assigns a unique identifier to each flow node.

When classifying the operation track data, the last valid node operation is firstly extracted, and the last valid node operation is combined with the corresponding interrupt timestamp field to be positioned, wherein the positioning operation is used for determining the flow node where the user stays before leaving or stopping the flow. The occurrence phase of the interruption and its possible cause can be further deduced from the context position of the node in the flow definition file. When the interrupt corresponds to the failure of judging the strategy of the server, if the health notification data does not meet the conditions, the wind control model is insufficient in grading, the system interface is failed in verification, the interrupt can be classified into a preset service strategy verification failure type, and the interrupt is classified into a user active termination type because of the action triggering events such as user page closing, clicking return, response within a limited time and the like.

After classification is completed, the classification result is required to be represented in a structured manner, and the structured fields include, but are not limited to, a business process stage field (such as filling information, health notification, payment confirmation, etc.), an interrupt type field (such as policy verification failure, active termination), a control identification field, an interrupt timestamp field, an associated policy version field, etc. The structured result is used as a basis for mapping the follow-up image into a visual topological graph, performing frequency statistics and generating an early warning instruction.

In an actual system, event tracking middleware can be adopted to cooperate with a browser end embedded point SDK to collect operation tracks, and the operation tracks are transmitted to a back-end analysis module in real time through a message queue. The back-end analysis module utilizes the flow definition metadata and the interrupt classification rule engine to aggregate operation tracks in the single-time application flow of the same user into an event chain, determines the last effective operation node and extracts relevant fields. For the interrupt event which is judged to be the failure of the strategy verification, the strategy verification module can backtrack the hit strategy item and the judgment result, and for the active termination event, the behavior intention is judged by combining with a user behavior event feature library (such as page jump, window closing and the like). Different systems can expand and refine the structured fields according to the depth of the flow, the complexity of the control and the granularity of the policy rules.

An example illustrates that in a medical health scenario, a user applies a heavy illness product through a self-camping H5 page. The application process comprises a plurality of front-end page nodes for home page display, applicant information filling, insured person information filling, health notification, guarantee responsibility confirmation, payment submission and the like. The user inputs the height of 160cm and the weight of 95kg in the health notification page, and the system judges that the BMI index exceeds the upper limit of the configuration of the current underwriting version and does not meet the requirement of health notification. When the user clicks the button for submitting information, the server-side strategy judging module is triggered to return a popup window which does not pass through the prompt and displays the 'temporary support of the application'. The user then closes the page and exits the process. In the process, the operation track acquisition module records the clicking action of the user on the health notification page, the interruption time stamp, the control identifier "# btn _submit_ ghg 3.1.1" and the returned strategy hit item "BMI overrun", and also records the strategy version "health notification strategy V3.1" used by the event. The back-end system analyzes the operation track chain, confirms that the user is interrupted at the health notification node, and the interruption accords with the 'preset service policy check failure type' defined by the system. The system further extracts a control identifier, a policy version identifier, a node name, a trigger timestamp, etc. in the event, and generates a structured data field as follows:

The business flow stage comprises health notification, an interruption type, a control identifier, an interruption timestamp, a policy version identifier and a policy hit type, wherein the interruption type comprises a policy verification failure, the control identifier comprises a # btn _subset_ ghg3.1, the interruption timestamp comprises a 2025-03-29T14:23:11Z, the policy version identifier comprises a health notification policy V3.1, and the policy hit type comprises a BMI overrun.

In the financial and technological business, another user tries to apply for a small accident risk through the H5 end, stays for more than 10 minutes in filling in the contact information page, does not operate, and then directly closes the page. In the acquired operation track, the last event is an input frame focusing timestamp of '2025-03-29T 10:45:02Z', a page closing timestamp of '2025-03-29T 10:56:11Z', and a control identifier of 'input_contact_name'. The back end judges that the behavior belongs to typical user active termination behavior after analysis, maps to 'filling information flow' stage interruption, and generates the following structured data:

The business flow stage comprises the steps of filling contact information, interrupting types, control identifiers, interruption time stamps, policy version identifiers and policy hit types, wherein the user is actively terminated, the control identifiers are in a form of # input_contact_name, the interruption time stamps are 2025-03-29T10:56:11Z, and the policy hit types are null.

Through the structural expression, different types of process interruption can be accurately captured and classified, so that the following steps of data analysis, frequency statistics, early warning identification and the like have clear starting points and logic sources, and visual support and strategy updating basis are provided for process optimization. The structured expression also supports expansion and multiplexing under different service lines and different control structures, and has good engineering suitability and practical value of products.

By classifying the user operation track data according to the service flow sequence and generating the structured interrupt type data, the interrupt behavior in the application flow can be effectively mapped to the logic node type which can be quantitatively analyzed, the flow bottleneck and the strategy blocking position are further revealed, and the behavior attribution and the reason deduction with more context awareness can be realized.

S40, mapping the multi-level logic interrupt type structured data into a visual topological graph;

In this embodiment, after the generation of the multi-level logical interrupt type structured data is completed, in order to achieve the visual tracking and multidimensional data presentation of the interrupt type, the structured data needs to be topologically mapped. In the mapping process, firstly, core classification fields in the structured data, namely a primary logic interrupt type field and a secondary logic interrupt subclass field, need to be identified, and the two fields form a backbone structure of data aggregation. In the topology structure, a primary logic interrupt type field is used for generating a topology master node as an attribution node of macroscopic business process interrupt, such as "health notification failure", "payment giving up", "information missing", and the like, and a secondary logic interrupt subclass field is used for generating a subordinate child node mounted under the master node for expressing more specific reason classifications, such as "BMI overrun", "bank card unverified", "contact mobile phone number missing", and the like.

After node generation, to encode interrupt liveness in different time dimensions, time series fields in the structured data need to be extracted, and a thermal value for each interrupt event is calculated in combination with the current time and the interrupt timestamp, where the thermal value is used to represent the "freshness" or "urgency" of the event. A common calculation method is to divide the difference between the "current time" and the "interrupt timestamp" by a preset time window (e.g. 24 hours), and perform standard inversion processing, where a higher thermal value indicates that the interrupt occurrence time is closer to the current time. And regenerating a color phase gradient value according to the calculated thermal value, and establishing a mapping model between the thermal value and the color, wherein for example, red is used for representing a high thermal value, and blue is used for representing a low thermal value.

In addition, the service policy version identification field contained in the structured data is also required to be bound and displayed in the graph in the modes of icons, labels and the like, and the policy version information is not only used for locating the source of the interrupt policy, but also can be used as an auxiliary clue for subsequent policy evaluation and policy rolling management. And finally, constructing a connecting edge according to the logic subordinate relation between the first-level node and the second-level node, wherein the line width of the connecting edge can be positively correlated with the interrupt frequency corresponding to the first-level logic interrupt type, and the color of the edge is consistent with the hue gradient of the main node so as to enhance visual alignment and hierarchical perception.

The mapping process not only builds a structural node map, but also encodes the subordinate, frequency and thermal data among the nodes into the map, thereby realizing the fusion expression of the interrupt data in the graphic space.

In an implementation, a topology map may be dynamically generated using a front-end visualization engine (e.g., D3.js, ECharts, etc.). The back end of the system firstly establishes a data model according to the structured interrupt data, takes the primary logic interrupt type as a main node of the graph, renders the primary logic interrupt type into a circular node through a front end canvas, and hooks the node size and the accumulated interrupt frequency of the node. Each master node downloads its corresponding child node, which is generated by a secondary logical interrupt child field in the structured data, also presented in a circle or rectangle.

In order to realize the expression of time heating power, the server calculates heating power values in advance in a data processing stage, converts the values into HSV or RGB color values, and dynamically draws the values by a front-end engine. The strategy version identification is attached beside the node in the form of a visual label, and different colors or frame patterns are adopted to distinguish the strategy version identification from the interruption frequency information. The data structures of all nodes and edges will be organized into standard graph data formats (such as nodes and links fields in JSON) to achieve efficient interactions between front and back ends.

In the layout algorithm of the graph, force-directed layout can be adopted to enable the nodes to present natural structure expansion states, and hierarchical layout can also be selected to ensure that the main nodes are always on the main axes of the graph to facilitate focusing. The rendering logic of the edge needs to dynamically adjust the line width parameter according to the frequency of the main node, and allows the click node to check specific strategy hit information or jump to a strategy tuning inlet.

An example illustrates that in a medical health application business, a health risk platform is configured with a health notification verification policy that encompasses multiple products and people. With the development of business, the situation that users cannot finish the application due to the health notification link in a certain quarter is obviously increased. And the background system performs topology mapping on the structured interrupt data of all the insuring failure users in the near period to generate a visual topological graph. In the figure, the "health notification failure" is taken as a primary node, and a plurality of child nodes, such as "BMI overrun", "hypertension unrendered", "physical examination report missing", and the like, are aggregated. The node color is distributed in a gradient from blue to red, the red node is significantly focused on "BMI overrun", and the policy version V3.2 corresponding to the node is identified as a policy tag in the figure. The edge connection is dense, the line width is thick, and the frequency is high, which indicates that the interruption intensively bursts in the near term.

In a financial service application scene, in an application flow of a small personal accident insurance product, an operation team discovers that payment is incomplete through a topological graph as a main node frequency rising trend is obvious, and further tracks that a sub node discovers that the interruption of credit card verification failure is most obvious. The hue of the child node is red, the thermal value is high, the version label is 'payment strategy V1.4', and the strategy version is prompted to interrupt event surge within one week after being online. Based on the method, the team timely adjusts policy parameters and optimizes front-end payment verification logic, so that the subsequent trigger rate of the node is effectively reduced.

By mapping the structured interrupt data into a visual topological graph, the goal of imaging and structuring abstract interrupt classification and policy events is realized. The process integrates information of multiple dimensions of service classification, time heating power, strategy version and data frequency, so that service operators can quickly identify interruption hot spots, trend changes and strategy triggering structures on a visual interface. Compared with traditional log screening or form analysis, the topological graph remarkably improves decision efficiency and strategy response speed.

S50, screening an entry with a logic interrupt type being a preset service policy check failure type from the multi-stage logic interrupt type structured data;

In this embodiment, in the completed multi-level logic interrupt type structured data, the logic interrupt type field usually exists in the form of enumeration or multi-level classification, which indicates the attribution of the cause of the interruption of the service flow. Such structured data not only contains context information (e.g., control identifier, timestamp, policy version identifier) for each interrupt event, but also records the logical interrupt type corresponding to the event in field form. In order to further focus on monitoring and analysis of the execution link of the preset service policy, the structured data entries need to be screened, and a record set with the logic interruption type equal to the verification failure type of the preset service policy is extracted.

This operation relies on a predefined enumeration set of service policy check failure types, which is maintained dynamically by the policy management service or configuration system from an online policy inventory, containing all check classifications with service policy trigger logic, such as "health notification check failure", "payment logic check failure", "user qualification verification failed", etc. The screening process needs to use a primary logic interrupt type field in the structured data as a judgment basis to match whether the field value belongs to a set of preset service policy check failure types.

When the screening action is executed, in order to ensure the accuracy and timeliness of the result, the filtering conditions such as dynamic enumeration pulling, version matching verification, data time interval limitation and the like can be supported, so that the screening result can accurately reflect the real influence range of the current version strategy on interruption. And finally, the item set obtained by screening is used as a core input data set for subsequent statistics, analysis and early warning, so that the data processing logic is ensured to have target focusing property and strategy traceability.

An example illustrates that in a medical health application platform, a health notification policy V3.2 is online for a quarter, the policy containing multiple parameter thresholds and control verification logic. After completion of the health information filling, if the user fails the rule due to the BMI, past medical history, age discrepancy, and the like, the user is recorded as "health notification verification failure". After the platform accumulates interrupt data for a period of time, the system executes screening operation, and all items with the primary logic interrupt type of 'health notification check failure' are screened out from the complete logic interrupt type structured data set, and irrelevant records such as 'payment incomplete', 'user active exit', and the like are filtered out. The screening result contains the real triggering data of the strategy under the appointed time window, including specific control identification, failure time stamp, user source channel and the like, and provides a solid data base for the follow-up generation of frequency data, hierarchical early warning and strategy optimization.

In the financial insurance business scenario, an unexpected insurance product platform is online with a payment verification optimization strategy V1.8 in the fourth quarter of 2023. Through the screening function of the interrupt structured data, the platform extracts all the entries with the logical interrupt type of 'payment verification failure', and confirms that all the entries are triggered by the policy rule verification logic instead of being manually cancelled by a user. The screening action ensures that the follow-up frequency statistics are not interfered by non-policy interruption factors, so that a grading early warning mechanism is more accurate and reliable.

Through accurately screening items corresponding to the preset service policy verification failure types from the multi-stage logic interrupt type structured data, uncontrollable or non-policy related interrupt factors such as active termination of a user, page stay overtime and the like are effectively isolated, and subsequent statistics and early warning analysis are focused on real service blocking caused by policy execution. The screening operation not only improves the target consistency of interrupt data analysis, but also builds a causal data path bound with a strategy version, and provides a data basis for strategy effect evaluation and anomaly monitoring.

S60, counting the triggering times of the preset service policy check failure types according to the interrupt time stamp fields in the screened items, and generating interrupt frequency data according to the interrupt time stamp fields, the triggering times and the preset service policy check failure types;

In this embodiment, the interrupt timestamp field is typically a base field recorded in each structured interrupt entry in a standard time format that serves to identify the exact point in time at which a particular interrupt event occurs in the business process. After the entry screening of the policy check failure type is completed, the time dimension analysis of the policy trigger frequency can be performed by using the field as a key statistical benchmark.

To achieve frequency statistics, a reasonable time aggregate granularity is defined, which is usually expressed as a preset time window, and may include hours, days, weeks, natural months, business quarters, and the like. For example, when dividing by day as a time window, the system needs to determine to which natural day the timestamp of each interrupt event belongs and assign it to the counting bucket under that day. And the system performs grouping statistics on all the screened interrupt events according to time windows, so that the number of times of strategy verification failure in each time window can be obtained, and time sequence structured interrupt frequency data is generated.

The structure of the interrupt frequency data generally comprises three basic fields, namely a time window, a policy check failure type and a triggering number. The structure can be used for supporting the subsequent operations of baseline model calculation, trend analysis, ring ratio change rate calculation and the like, and has high expandability and model compatibility. Parallel statistics of multiple policy types should be supported in the statistics process so as to adapt to actual application scenarios of concurrent operation of multiple service lines and multiple policy versions.

By carrying out aggregation statistics based on the interrupt timestamp field, not only is the trigger frequency modeling of the strategy verification failure type under each time window realized, but also a quantifiable index base is provided for subsequent fluctuation detection, abnormal early warning and dynamic fault tolerance adjustment.

S70, triggering a grading early warning instruction containing a preset business strategy version identifier when the interruption frequency data exceeds the upper limit of the baseline frequency;

In this embodiment, the interruption frequency data is a record of the trigger times of policy verification failure in a time window obtained by statistics according to historical track information, and the upper limit of the baseline times is a conventional fluctuation range of the policy interruption in a similar historical period, which is used for defining an abnormal boundary. When the interruption frequency data reaches or exceeds the baseline boundary, classification judgment and early warning triggering processing are needed immediately.

Firstly, the system reads the triggering times field and the corresponding strategy version identification field in the interruption frequency data object in the current time window, and locates the current active strategy version and the performance fluctuation index thereof. Then, the system calls the historical interrupt frequency data buffer or analysis result, matches the trigger record set of the historical synchronization according to the periodic attribute of the current time window (such as according to the quarters Q1, Q2, etc.), and extracts the trigger frequency sequence, such as the frequency value of the same quarter in the last three years.

Next, the system calculates an upper baseline number of times for the policy over the time window using the historical sequence. The upper limit is usually calculated by a statistical interval method, such as a mode of adding three times of standard deviation (mu+3σ) to the mean value or adding an upper limit of a 95% confidence interval, and the like, so that abnormal growth conditions outside a normal fluctuation range can be covered. The baseline value may be dynamically updated by the policy analysis module, or an upper threshold limit may be set in the configuration system to limit its growth range.

After the upper limit and lower limit comparison is completed, if the current trigger frequency field is greater than the baseline frequency upper limit, the system enters an early warning generation flow. At this time, the system firstly calculates an overrun amplitude value, namely the current trigger frequency minus the base line upper limit, and matches the corresponding hierarchical early warning rule table according to the interval level where the value is located. The mapping relation between the overrun amplitude range and the early warning level is defined in the rule table, for example, the overrun amplitude range and the early warning level exceed 0-10 times corresponding to one-level early warning, the overrun amplitude range and the early warning level exceed 11-30 times corresponding to two-level early warning, and the overrun amplitude range and the early warning level exceed more than 30 times corresponding to three-level early warning.

After successful matching, the system generates a hierarchical early warning instruction which is a structured object and comprises information such as a current time window identifier, actual trigger times, a baseline times upper limit, overrun amplitude, an early warning level of a matching result, a strategy version identifier, a current strategy type field and the like. The instruction can be used as a trigger action push notification and also can be used as a decision parameter for a subsequent strategy weight adjusting module to read.

The generated hierarchical early warning instructions can be pushed to appointed early warning channels, such as mail servers, short message gateways, instant messaging systems, operation large screens and the like. The format supports JSON, XML or an internal preset protocol structure, and has high compatibility and lightweight transmission characteristics. In part of the system, the early warning instruction can also be used as a pre-trigger condition for strategy automatic adjustment, and the strategy fault tolerance setting or front-end significance optimizing component is linked.

By comparing the numerical relation between the interruption frequency data and the upper limit of the baseline frequency, a scene of abnormal increase of the service policy verification failure frequency can be timely identified, and a hierarchical early warning instruction containing a preset service policy version identifier is triggered, so that a system can rapidly locate a policy version and a service control with potential abnormality, further subsequent page adjustment and policy correction operations are driven, and stability of a service flow and response efficiency of policy operation and maintenance are improved.

And S80, verifying failure type data according to the preset business strategy associated with the hierarchical early warning instruction, and generating page element weight adjustment parameters and fault tolerance threshold value sets corresponding to the preset business strategy version.

In the embodiment, after the system receives the hierarchical early warning instruction, the system carries out structural analysis on the preset service policy verification failure type data contained in the hierarchical early warning instruction, and extracts two core fields, namely a unique identifier of the page control and a corresponding service policy version identifier. The unique identifier of the page control typically points to a business interaction node that is bound in the front-end page, such as a form input box, drop-down selector, or a confirm button, which may be generated and cached by the rendering engine when the page is loaded, or preset by the developer as a component parameter. The service policy version identifier is used as an index key of the service side policy checking module and is used for accurately pointing to the currently running policy logic set, and different service interruption behaviors are triggered between different versions possibly due to rule differences.

By taking the control identifier as an index, the system searches interrupt frequency data related to the historical triggering behavior of the control, wherein the data sources comprise a local log file, a real-time monitoring stream or a back-end interrupt record table. The interrupt frequency value is mapped into a priority coefficient after standardized processing, and the coefficient is used for regulating and controlling the significance degree when the page is rendered. The priority coefficient can be linearly and positively correlated with the frequency, and can be dynamically adjusted through a piecewise function, an exponential function or a weight model. For example, when the frequency is greater than a preset threshold, the priority coefficient will be quickly enlarged to ensure that the page key controls can be noticed by the user and correct the input behavior in a short time.

At the same time, the system extracts a historical fault-tolerant configuration record in the policy management database based on the policy version identifier and locates the fault-tolerant trigger number field of the target version. This field represents an upper limit on the tolerance of a policy version to verification failures in the business process, such as the maximum number of allowed failures, the false commit threshold, or the number of waiting retries. The system updates the fault tolerance threshold according to the current trigger level, and the updated value can be set statically, in a same proportion and in a growth strategy or adjusted according to the event response level (for example, the fault tolerance threshold is adjusted to be 150% of the default value when the early warning level is one level). The update action should keep a history threshold record to support backtracking analysis.

After the two subtasks are completed, the system combines the generated page element weight adjustment parameters with the fault tolerance threshold value set to construct a group of strategy optimization configuration data which faces to the bidirectional linkage of the front end and the rear end. The configuration data is expressed in a structured form and comprises a control identifier, a priority coefficient, a strategy version identifier and a maximum fault-tolerant trigger number field, and can be used as a data base for cross-module calling and used for driving front-end dynamic style adjustment and back-end strategy engine tolerance update.

In specific implementation, the hierarchical early warning instruction can be received by constructing an event-driven-based data perception service, and a preset business strategy verification failure type data field carried in the hierarchical early warning instruction is analyzed through a JSON format, wherein the data field comprises a control identifier and a strategy version identifier. The control identifier is predefined by DOM attributes of the front-end control when the embedded point is embedded, and the strategy version identifier is ensured to be consistent by the strategy service in the write operation track data when each verification failure event occurs.

The frequency of interruption of the control identifier may be obtained by periodically scanning a historical behavior table in a buried data log, kafka stream log, or database. Frequency statistics support aggregation analysis using sliding windows, such as counting the cumulative number of times in the sliding window in hours, days, and weeks. In an actual deployment, the calculation of the priority coefficient may map the frequency to a fixed interval using a min-max normalization based method, e.g., [1.0,3.0], or may be based on a Z-score calculation, ensuring that controls three standard deviations above the average possess significant visual weight.

The number of fault-tolerant triggers associated with a policy version identification is typically stored in a versioning table in a policy configuration center, each version containing a set of rules and corresponding threshold entries. The system may request the fault tolerance value through MySQL, redis, or policy service API interfaces. The updating mode of the maximum fault-tolerant trigger times supports strategies such as static replacement, interval adjustment, history contrast dynamic setting and the like, and ensures that the maximum fault-tolerant trigger times are matched with the early warning level. For example, the maximum fault tolerance value can be increased from 50 to 75 when the primary warning level is reached, and the secondary warning level can be moderately adjusted up to 60. All updated thresholds need to be synchronized to the policy engine configuration center and trigger a cache refresh mechanism of the policy service.

The optimized strategy configuration data is finally written into a strategy release queue or a configuration distribution middle stage for subscription use of a front-end page rendering system and a back-end check engine. The front end injects priority coefficient into CSS style according to control identifier, and realizes visual enhancement by adjusting control frame thickness, color saturation or flash animation, and the back end adjusts strategy check logic accordingly, thereby improving fault tolerance elasticity of the system.

By analyzing the page control identification and the strategy version identification in the hierarchical early warning instruction, the system can realize bidirectional linkage adjustment of the page guide key point and the strategy tolerance threshold. On one hand, the pertinence and the intervention capability of page interaction are improved, the psychological burden of repeated failure of a user on a key control is reduced, and on the other hand, the adaptability of a back-end strategy to a high-frequency verification failure scene is enhanced, and false killing or experience degradation caused by too tight setting of a fault-tolerant strategy is avoided. By the mode, the system realizes policy dynamic tuning and user behavior guiding on the premise of not interrupting the service flow, and the conversion efficiency and the user satisfaction are integrally improved.

The invention relates to the technical field of data analysis, which can be applied to business scenes such as financial science and technology, medical health and the like, and discloses a business process visualization early warning and strategy response method, comprising the steps of setting a physical interrupt node to collect interrupt operation track data, and classifying the data into multi-level logic interrupt type structured data based on a business process sequence; and when the frequency exceeds the upper limit of the base line frequency, triggering a hierarchical early warning instruction containing a preset service strategy version identifier, and generating a page element weight adjustment parameter and a fault tolerance threshold value set of a corresponding strategy version according to the instruction. According to the method, the visual topological graph is constructed, the high-frequency interrupt behavior is identified, the real-time identification and the hierarchical early warning of the failure of the service policy verification are realized, the self-adaptive adjustment of the page elements and the policy threshold value is further driven, and therefore the user insurance conversion rate is effectively improved, and the front-end operation loss risk is reduced.

In one embodiment, the step S30 includes:

S301, analyzing a last operation node identifier and an interrupt timestamp field in the operation track data;

s302, matching a verification failure code corresponding to the last operation node identifier in a preset service strategy library;

s303, associating the currently effective business strategy version identification with the verification failure code;

S304, generating a primary logic interrupt type and a secondary logic interrupt subclass according to the service flow sequence, wherein the primary logic interrupt type corresponds to the service flow stage classification, and the secondary logic interrupt subclass corresponds to a verification failure scene defined in a preset strategy library;

s305, converting the interrupt time stamp into a time sequence field in a preset time zone format;

S306, combining the verification failure code, the currently effective service policy version identifier, the primary logic interrupt type, the secondary logic interrupt subclass and the time sequence field to generate the multi-stage logic interrupt type structured data.

In this embodiment, the operation track data in the business process generally includes behavior information generated during the interaction process between the user and the front-end interface, and the core field includes the last page operation point before termination and corresponding timestamp information. The "last operation node identification" in the operation track data is the last control or component number, such as the control ID of a button click event or a form submitting action, of the current interrupt operation is extracted from the original data for subsequent tracing. Meanwhile, the analysis of the "interrupt timestamp field" refers to extracting the point in time when the interrupt action occurs, and is used for subsequent time classification, thermal analysis and the like.

By comparing with a preset service strategy library, the system can position the service verification flow corresponding to the control mark. For example, if the policy verification is not passed after a control is triggered, a "verification failure code" corresponding to the control is found in the policy repository. This step requires ensuring that the policy store is stored Is a bi-directional mapping relationship of (a). Based on the mapping relationship, the 'currently effective service policy version identification' is further associated. The strategy version identification can be obtained by inquiring an entry with a state of effective in a strategy library, so that the historical version and the rules to be abandoned are prevented from being confused.

Based on the business process sequence information, the strategy failure behavior is mapped into two dimensions of a primary logic interrupt type and a secondary logic interrupt subclass. The first-level logic interrupt type generally corresponds to a high-level business stage in the process, such as user identity verification, health information verification, compliance confirmation and the like, and the second-level logic interrupt subclass is refined into a specific trigger scene, such as certificate number invalidation, BMI overrun, disease history conflict and the like. In order to facilitate the subsequent time sequence analysis and visual presentation, the system needs to uniformly format the time stamp fields into a time sequence of 'preset time zone format', for example, uniformly convert into YYYY-MM-DDHH: MM: ss form of the east eight region.

Finally, the user behavior interrupt event is converted into a structured interrupt entry by combining the verification failure code, the policy version identification, the logical interrupt level information and the time sequence field. The structured item has unified field expression and classification identification, and can be directly called by a visualization component, an early warning model or a rule engine.

In actual deployment, the system firstly receives an interrupt behavior log reported by a business front-end embedded point system, and the log contains fields such as a control number, a page path, operation time and the like in a JSON format. The analysis module at the server side utilizes a regular expression or a preset field extractor to extract a 'last operation node identification' and an 'interrupt timestamp field'. And then, by maintaining a relational database in the policy engine module, the system matches the verification failure code according to the control identification and generates a version identification by combining the effective policy version of the current system label. The mapping of the logical type hierarchy may be derived by a flow structure definition configured in the metadata bin. For example, the page path information may correspond to a first stage and the failure type identification code suffix may map a second sub-class. The standardization of the time field relies on a unified time service module that converts to a standard time zone format by calling a service API. The final output structured data may be generated by multi-field splicing to form the following structure:

{ "check failure code": "S101", "policy version": "V3.2", "primary logical interrupt type": "health check", "secondary subclass": "BMI overrun", "time series": "2025-03-3015:45:22" }

The data is pushed to a downstream Kafka message queue for real-time consumption by a topological graph drawing service and an anomaly analysis engine.

According to the embodiment, the capability of accurately extracting the interrupt root cause, the hierarchical classification and the strategy tracing from the original behavior data is realized by introducing a layer-by-layer analysis and structuring processing mechanism for the operation track data. The method has higher logic reduction capability and stronger flow context awareness capability, so that the interrupt behavior has interpretability and controllability in subsequent analysis, and the recognition accuracy of a system to high-intention interrupt users and the optimization guiding capability are improved.

In one embodiment, the step S40 includes:

s401, extracting a primary logic interrupt type field and a secondary logic interrupt subclass field in the multi-stage logic interrupt type structured data;

s402, mapping the primary logic interrupt type to a master node of a topological graph and mapping the secondary logic interrupt subclass to a slave child node of the master node according to a preset visual coding strategy;

s403, extracting a time sequence field in the multi-stage logic interrupt type structured data, and converting the time sequence field into a time sequence thermal value;

s404, generating a node hue gradient according to the time sequence thermodynamic value;

S405, generating a strategy version label based on a service strategy version identification field in the multistage logic interrupt type structured data, and binding the strategy version label to a corresponding master node and a corresponding slave sub-node;

S406, generating a connection edge of the topological graph based on the hierarchical relationship between the master node and the slave sub-nodes, and applying the node hue gradient to the filling attribute of the corresponding node to form the visual topological graph.

In this embodiment, in the process of converting the structured multi-level logical interrupt type data into the visual topological graph, the preprocessing operation of the field structure needs to be completed first. The structured data originates from an interrupt event classification flow, and its core fields include a primary logical interrupt type field, a secondary logical interrupt subclass field, a time sequence field, and a service policy version identification field. Each field needs to be matched with semantics through data consistency check before mapping, for example, a preset hierarchical relation between a secondary sub-class field and a primary type field needs to be ensured, and mounting abnormality or unexpected hierarchical drift is avoided.

The primary logical break type field is used as a basis for identifying the master node, and is typically displayed in a centered or upper layout at the time of patterning to emphasize its aggregation. Within the system, this field will be converted into a graph node type tag for driving the topology graph generator to identify it as "master type". The secondary logical interrupt subclass field is then mounted as a "slave type" under or around the master type node, organized by logical tree relationships or polar layout. The organization method not only improves the information density, but also reduces the cross edges and redundant connection and improves the topology readability.

The processing of the time sequence field is not just thermal value calculation, but the conversion process involves time zone standardization, time precision clipping (such as unifying to minute granularity or hour granularity) and event occurrence density aggregation. The system may configure different time window policies, such as short windows for real-time heat sensing and long windows for policy backtracking visual analysis. After the conversion is completed, each event node is provided with a standardized time stamp, and the time stamp is used as dynamic input in a thermodynamic value function to participate in hue mapping, so that the direct influence of time evolution on the figure color is ensured.

In the composition rule, the connection edge between the master node and the slave node not only expresses the structure level, but also bears the data-driven information transfer channel. The width, transparency, animation flow direction and other attributes of the connecting edge can be bound with the triggering frequency, time intensity or version label of the interrupt event. For example, between nodes that are very frequently interrupted, animated blinking edge display may be employed to enhance visual focus effects. Meanwhile, the color of the connecting edge is set to be consistent with that of the source node, so that color splitting is reduced.

The binding policy of the policy version identification field supports a bi-directional mapping mechanism, which can track the source of the policy version from the node structure and activate the visible node group under a specific version in the version screening panel. The version label is not only limited to a text labeling form, but also can be embedded into graphic elements (such as upper right corner marks and icon superposition) and linked with an event tracking system, and jumps to a strategy configuration background after clicking to form a cross-system interaction closed loop.

Throughout the visualization topology map generation process, a rule-based graphics template engine may be used, such as binding each level one logic type to a particular graphic (circle, rectangle, hexagon, etc.), to assist the user in identifying different types of anomaly patterns. In addition, the generation of the topological graph supports the capabilities of dynamic redrawing, screening, scaling and the like, and ensures that key data can be efficiently acquired under different screen sizes and roles of business personnel.

Furthermore, the system may support multi-level interrupt type nesting, such as adding three levels of logical subclasses, shown in the form of aggregation circles or collapsed nodes in the graph. The mechanism is particularly important in complex flow interruption scenes (such as medical information filling), and can effectively relieve graphic congestion and maintain interaction efficiency. The updating mechanism of the graph can also dynamically refresh the node state based on a timing task or trigger mechanism to reflect the real-time interrupt situation.

According to the embodiment, the structured interrupt type data are converted into the visual topological graph, so that the hierarchical structure and specific distribution of interrupt events can be visually presented, the time dimension characteristics of the interrupt can be expressed through the hue of the node, and policy attribution and version tracing are assisted by means of policy version identification. By integrating originally scattered and invisible business interruption information into an interactive map, an efficient problem identification and decision support means is provided for product operators and policy making teams.

In one embodiment, after the step S40, the method further includes:

s407, acquiring time interval parameters and commodity classification codes configured through an interactive interface;

S408, extracting a logic interrupt type subset which can be matched with the time interval parameter and the commodity classification code from the visual topological graph;

S409, performing visual hiding operation on the logic interrupt types which are not matched with the time interval parameters or the commodity classification codes;

s410, determining the ring ratio change rate of each node in the logic interrupt type subset based on historical service period data;

S411, marking the logic interrupt type with the ring ratio change rate exceeding a preset change rate threshold as a high-attention logic interrupt type, and marking physical interrupt nodes belonging to the high-attention logic interrupt type with pulse special effects in the visual topological graph;

And S412, generating an abnormal category optimization list based on the commodity classification codes corresponding to the high-attention logic interrupt types.

In this embodiment, after the topology map is generated, a targeted focus analysis may be further performed on the type of logical interrupt presented therein, so as to improve the correlation between the data readability and the service optimization. Firstly, acquiring time interval parameters and commodity classification codes in analysis dimensions, wherein the two parameters are provided by an interactive interface configuration module, and a user can input the parameters through a time selector and a commodity category screening component. The time interval parameter generally comprises a start-stop time stamp or a natural period range (such as quarterly and month), and the commodity classification code is a commodity category identifier defined in a preset business system, such as health risk, serious disease risk, travel risk and the like.

The system screens the broken nodes in the visual topological graph based on the time interval parameters. Each node internally contains a time sequence field for matching the time dimension and a commodity classification field for matching the business dimension. Nodes meeting the matching of two dimensions are extracted to form a logic interrupt type subset, and the subset is the object of subsequent high-frequency focusing and fluctuation analysis. For nodes which do not meet any dimension matching condition, the system triggers graphic hiding operation, namely visual rejection is realized by means of transparency reduction, edge connection cancellation, label removal and the like, so that visual interference is avoided.

After extracting the subset of logical interrupt types, the system will perform a ring ratio calculation for each node based on historical traffic cycle data. The interruption frequency of the node is derived from interruption frequency data generated in the preamble step, and the system obtains the frequency values of the current period and the last period according to the double dimension aggregation of commodity classification and node identification. The upper period is defined as the last unit time period with the same period as the current time interval, and is 2023-Q4 if 2024-Q1 is present. The calculation formula of the ring ratio change rate may be:

the loop ratio change rate= (current period break frequency-up period break frequency)/up period break frequency x 100%.

The calculation mode is used for measuring trend fluctuation conditions of each logic interrupt type in the time dimension. When the ring ratio change rate of a certain node exceeds the change rate threshold set by the system (for example, +50%), the high-attention logic interrupt type is regarded. The system marks the nodes and gives visual enhancement effects, such as continuous-jumping pulse light rings, color flickering, edge diffusion and other dynamic special effects, to attract the attention of analysts.

After the visual marking is completed, the system can also summarize and form an abnormal category optimization list according to commodity classification codes corresponding to marking nodes. The list is output in sequence according to the ascending amplitude of the interruption frequency, the number of nodes, the dimensions related to the control and the like, and is used for the business department to identify the current insuring category needing important optimization. The list can be directly used for downstream tasks such as product operation policy optimization, user flow reconstruction, front-end interaction adjustment and the like.

In the specific implementation process, the analysis personnel can select the time interval and the commodity classification code through the parameter configuration component of the front-end interface. The selected time interval parameters may support natural language parsing (e.g., "this quarter", "last month") and may also support time stamp range entry accurate to date. The commodity classification codes can be obtained by interfacing with an insurance product information system, building a drop-down selection list and supporting multiple selection operations.

And after receiving the screening conditions, the back-end processing module executes double screening operation in the generated visual topological graph data. The screening logic preferentially searches the nodes conforming to commodity classification codes in the primary logic interrupt type field and the secondary logic interrupt subclass field, and compares whether the nodes are in the current time interval or not in the time sequence field in each node. If the node and the node are matched, the node and the associated structure thereof are reserved, otherwise, the node is marked as an inactive state, the front end sets the transparency of the node to 0.2 during rendering, and the connecting edge of the node and the main node is cancelled.

In order to calculate the ring ratio change rate, the system queries the interruption frequency data of the upper period time window corresponding to the commodity classification codes, which are stored in the historical database, for the same node identification. If the current time interval is the first quarter of 2024, the system will automatically extract the first quarter of 2023 or the last quarter (as defined by the business) as the reference period. And obtaining the change rate of each node after normalization calculation of the difference value between the current frequency data and the historical frequency data. The system may set a global default threshold (e.g., 50%) or set separately for different merchandise classifications.

Once a node is identified as a high interest type, the system will append a "beat" class animation style in its topological graph representation, e.g., using WebGL or Canvas rendering to achieve periodic radiation effects of node edge color, and add a "high interest type" identifier to the legend to facilitate understanding.

Finally, an abnormal category optimization list is generated at the back end in a structured JSON format, and comprises fields such as commodity classification codes, trigger node quantity, average change rate, main control identification and the like. The list is pushed to an operation system or a decision support platform through a message queue for subsequent automatic policy adjustment or manual intervention and audit.

According to the embodiment, by introducing a double-dimension screening mechanism of time interval parameters and commodity classification codes, the display result of the logic interrupt type has more service pertinence and timeliness, and information redundancy and interference caused by full visualization are avoided. The ring ratio change rate is further calculated based on the historical service period data, and the interrupt nodes with outstanding change trend are dynamically marked in a visual form, so that potential abnormal links in the service flow can be effectively found. The automatic generation mechanism of the abnormal category optimization list constructs a technical closed loop from interrupt identification to operation intervention, and the efficiency of product optimization and strategy response is remarkably improved.

In one embodiment, the step S60 includes:

S601, classifying each item into a corresponding preset time window based on an interruption time stamp field in the screened item, wherein the preset time window is a predefined statistical period interval;

S602, grouping and aggregating the items classified into the same preset time window based on the preset service policy check failure type, and generating a policy check failure type item set arranged according to the time window sequence;

s603, counting the triggering times of interrupt events of a strategy verification failure type item set in each preset time window, wherein the triggering times are the accumulated number of all items in the corresponding time window, and each item corresponds to an interrupt event of one preset service strategy verification failure;

S604, combining the preset time window, the triggering times of the interrupt event and the verification failure type of the preset service strategy to generate the interrupt frequency data.

In this embodiment, when processing an entry of a preset service policy verification failure type screened from the structured data, in order to analyze the concentration, time distribution and policy problem density of the interrupt event, a time aggregation model of the interrupt event needs to be built based on a timestamp field. The model firstly requires standardization processing of the interruption timestamp field in each piece of data, namely, the interruption timestamp field is converted into a time format under a unified time zone, such as ISO 8601 time standard, so that the follow-up clustering operation is ensured not to cause distribution deviation due to time zone difference of data sources.

The time window is a core analysis unit of the model, and is not only a simple time period division, but also needs to support flexible definition of service driving, for example, insurance application peak time periods can be concentrated in 19:00 to 23:00 of weekday evenings, so that non-uniform time windows such as a 'weekday evening window', 'holiday day front and back window', and the like can be constructed. The time windows can be dynamically generated by business rules based on fixed time length, and have higher interpretability and pertinence in statistical logic.

Each interrupt entry needs to be logically clustered in combination with a policy check failure type after being included in its associated time window. The clustering can be further expanded to multi-level fields such as strategy category, applicable commodity, strategy effective time period and the like according to the strategy identification field so as to construct a multidimensional cross statistical model. For example, the same policy identifies that the frequency of possible triggers for interrupts in different commodities is quite different, so the aggregation logic should support multi-key joint grouping and provide field level weight adjustment capability.

The frequency statistics themselves should not be just a summary of the single counts. To ensure the ability to identify sustained trigger events, a rolling time window mechanism may be introduced, i.e., for each policy failure event, its impact is propagated forward and backward for a certain time range while accumulating its belonging time window. For example, when multiple users are continuously interrupted at the same policy point for 5 minutes throughout the day, the statistical system should identify it as a "local high frequency anomaly" rather than just as an independent event.

The finally generated interrupt frequency data not only comprises basic fields such as time windows, strategy types and interrupt event quantity, but also can be expanded to generate analysis index fields such as trend vectors, abnormality indexes or fluctuation coefficients. These fields will serve as inputs to the subsequent hierarchical early warning mechanism and the policy optimization calculation model, providing a data base for refined control.

The key to this statistical process is not the static summary of the data, but the cross-mapping and aggregation integration between the time dimension and the policy logic. The system should support a high-frequency refresh mechanism and asynchronous statistical task scheduling capability, so that the frequency data has real-time feedback capability in key scenes such as strategy change, sales promotion, holidays and the like.

The method includes the steps that in a self-supply H5 application flow provided by a certain insurance platform for a medical health scene, a user sequentially completes interactive input of a plurality of control fields such as basic information filling, health question and answer selection, occupation classification confirmation, payment mode setting and the like, the system performs real-time verification on each input content through a preset strategy, and a formal application link can be entered only after all verification passes. The recent platform monitoring team observes that the stay time of a large number of users in the period of filling the health information is obviously prolonged in the night peak period, the application completion rate is obviously reduced, and the condition that the strategy verification triggering frequency is abnormal is suspected.

For this reason, the system processes all interaction records that did not successfully complete the application flow in the past 7 days, extracts the structured entry in which the logical interrupt type is the "preset business policy check failure type", and the focus policy version identifier is HEALTH-GUOCHENG-3.2 failure record. The strategy is mainly used for restricting the answer integrity and compliance of the user in the healthy question-answering link, and once the behaviors of missing of the necessary-filling word segments, inconsistent answer and rule, missing of additional data and the like occur, the strategy is judged to be failed in verification.

The system first extracts the break timestamp fields in these entries and converts all time information uniformly into Beijing time (east eight zone) format for subsequent building of the time window. Taking a record of 21 hours, for example, 2024, 5, 12, and 21, the system falls into the statistical window "2024-05-1221:00-22:00" and aggregates all records belonging to the window and to HEALTH-GUOCHENG-3.2 policy versions.

The system counts the number of triggers of the policy within each window according to a predefined time window granularity (in units of hours). In the period, the strategy is triggered 58 times in a cumulative way, and the strategy is only triggered 23 times in the same period of the previous day, so that obvious fluctuation trend is formed. The system automatically writes the aggregate result into the structured output, and includes a preset time window field (e.g. "2024-05-1221:00-22:00"), a policy version identification field (HEALTH-GUOCHENG-3.2), and an interrupt event trigger number field (58), to form interrupt frequency data of the current cycle.

The analysis flow ensures the accuracy and comparability of the triggering times calculation process through an automatic classifying and window-by-window counting mode, provides a quantization basis for the subsequent triggering early warning rule, and simultaneously supports the data infrastructure of front-end interaction logic optimization and strategy fault-tolerant configuration adjustment. The system avoids the traditional hysteresis mechanism which relies on manual judgment and off-line extraction, and realizes the active discovery and high-frequency response of the risk of the interruption of the strategy level.

According to the embodiment, the data analysis basis with the structured time dimension can be obtained on the premise of not interrupting the service flow by mapping the interruption time stamp to the preset time window and carrying out frequency statistics by combining the strategy type. The platform can be helped to find out the trend of interrupting the proliferation of a certain strategy type in a short time, and meanwhile, the stability of different time periods or different strategy types can be compared transversely. By constructing the frequency data structure in advance, high-efficiency and uniform bottom layer support is provided for follow-up trend identification, early warning triggering, strategy optimization and other operations.

In one embodiment, the step S70 includes:

S701, determining the baseline frequency upper limit of each preset time window based on a triggering frequency field and a history contemporaneous time window field in the history interrupt frequency data;

S702, configuring a hierarchical early warning strategy to define early warning levels corresponding to different overrun amplitudes of which the triggering times exceed the upper limit of the baseline times;

s703, when the triggering times of the interruption events in the interruption frequency data exceed the upper limit of the baseline times of the current preset time window, matching the grading early warning strategies according to the overrun amplitude, and generating grading early warning instructions corresponding to the early warning levels;

s704, sending a grading early warning instruction to the configured early warning notification receiving end through a preset notification channel.

In this embodiment, in order to respond timely to abnormal fluctuation of the policy check interruption frequency, a hierarchical early warning instruction containing a service policy version identifier needs to be automatically generated according to the deviation degree between interruption frequency data and a history level. In the process, firstly, with a time window as a unit, whether the interruption times in each statistical period reach a threshold value for triggering early warning is determined. For this purpose, a historical statistical baseline with time comparability needs to be established, and a historical contemporaneous time window field provides a reference of a time dimension for the baseline, wherein the field represents a historical time period which is the same as a current preset time window period classification, for example, when '2024, Q3, 2 nd week workday, midday period' is taken as a current window, the historical contemporaneous time may include '2023, Q3, 2 nd week workday, midday period'. And extracting a corresponding trigger frequency field by traversing the time periods belonging to the historical contemporaneous classification in the history interruption frequency data to form a sample set for statistics.

After the sample data is acquired, the system calculates the baseline frequency upper limit of the preset time window in various modes such as mean, standard deviation, polar error, quantile and the like. The most common way is to set a dynamic upper limit by means of an average weighted standard deviation, e.g. baseine = μ +2σ, where μ is the historical contemporaneous average number of interruptions and σ is the standard deviation. The method can adapt to the baseline deviation caused by the change of the business magnitude, and effectively control the false alarm risk.

The hierarchical early warning strategy is used for setting early warning response grades of different grades according to the overrun amplitude, and the strategy can be configured in a combined way based on strategy types, service priorities and risk grades. For example, if the magnitude of a certain strategy exceeding the upper limit of the baseline frequency exceeds 20%, the first-stage early warning is judged, and if the magnitude exceeds 50%, the second-stage early warning is increased. The policy may be dynamically adjusted by a rule configuration service and support automatic binding of risk classification criteria based on policy version identification.

After the early warning level matching is completed, the system generates a hierarchical early warning instruction which comprises a current preset time window, calculated out overrun amplitude and a business strategy version identification field recorded in interruption frequency data. The identification ensures that the instruction content is accurately bound with specific policy configuration, and is convenient for personalized response of follow-up policy fault tolerance parameters, control significance weights or notification frequencies.

And finally, the system pushes the grading early warning instruction to an early warning notification receiving end through a configured notification channel, wherein the receiving end can be mail service, short message service, an internal message system, an abnormal processing center console and the like. The pushing process can be configured with notification priority, notification interval time and repeated sending rules so as to meet the difference requirements of different industry fields on early warning response efficiency.

According to the embodiment, the interruption frequency data is dynamically compared with the upper limit of the baseline frequency obtained based on the historical contemporaneous statistics, and a hierarchical early warning mechanism driven by the overrun amplitude is introduced, so that automatic sensing and response of abnormal fluctuation of the countermeasure verification can be realized. The method and the system not only get rid of the dependence on fixed threshold values or manual subjective judgment and remarkably improve the monitoring sensitivity to the interruption risk of the business process, but also can adaptively configure the early warning level and the notification mode according to the business importance and the risk level of different strategies, thereby enhancing the controllability and the accuracy of the early warning system.

In one embodiment, after the step S80, the method further includes:

s901, analyzing a control identifier and a priority coefficient in the page element weight adjustment parameter;

S902, matching the control identifier with a document object model node in an interface rendering service component;

s903, adjusting the weight value of the page element of the matched document object model node according to the priority coefficient;

S904, injecting the adjusted page element weight value into a browser rendering pipeline to improve the visual significance of the control corresponding to the failure type of the verification of the preset service strategy;

s905, analyzing the strategy version identification and the maximum fault-tolerant trigger times in the fault-tolerant threshold set;

s906, performing version matching on the strategy version identification and effective strategy entries in a strategy verification service database;

s907, updating the maximum fault-tolerant trigger number field in the matched effective strategy entry according to the maximum fault-tolerant trigger number;

s908, synchronizing the updated effective strategy entry to a business process execution service component to adjust the triggering condition of the preset business strategy verification failure type.

In the embodiment, after the system completes the generation of the page element weight adjustment parameters and the fault tolerance threshold value sets corresponding to the preset service policy versions, the processing logic is divided into two directions, namely, the page layer parameters are acted on the front-end visual presentation, and the policy layer parameters are fed back to the policy verification service system, so that the bidirectional dynamic adjustment of the user interaction experience and the policy tolerance is realized.

Firstly, a system analyzes a control identifier and a priority coefficient in a page element weight adjustment parameter, wherein the control identifier is used for uniquely referring to an interactive control which fails to check a strategy, and the priority coefficient is a dynamic index calculated according to the interrupt frequency corresponding to the control and is used for reflecting the attention weight of the control in the current business flow. In a specific implementation, the control identifier can be expressed in a mode of DOM node ID, CSS selector or unique path of the H5 page component, and the priority coefficient can output a floating point value by adopting a standardized weight model, so that consistency is maintained under different devices and resolutions.

The system then matches the control identifier with the document object model node in the interface rendering service component, typically by JavaScript runtime accessing the page DOM structure in modern browsers, to the target node through a selector or path rule. The matching process not only requires field consistency, but also needs to consider the scenes of delayed rendering, nested structure, component multiplexing and the like of the dynamic generation component, so that the system can introduce node tree traversal, delayed binding or MutationObserver monitoring mechanisms to enhance matching robustness.

Once the matching is completed, the system adjusts the weight value of the page element matched to the node according to the priority coefficient, and the weight value controls the visual saliency of the node in page display, such as word size, frame highlighting, background color strengthening, animation frequency and the like. The adjustment of the weight value can enhance the interactive guiding effect by means of CSS variables, style attributes or a responsive binding mechanism of the front-end framework on the premise of not influencing the user operation logic. For example, controls with higher priority coefficients will be given a higher visual hierarchy making them more perceived and operated by the user.

After finishing the setting of the visual enhancement parameters, the system injects the adjusted weight values of the page elements into the browser rendering pipeline. The operation depends on a staged rendering mechanism of DOM- & gt Style- & gt Layout- & gt Paint- & gt Composite in a modern Web rendering architecture, a parameter injection point is positioned in a Style or Layout stage, a rendering task is pushed by a front-end main thread or Web workbench, and the control is ensured to finish redrawing according to the adjusted weight in the next frame refreshing period, so that the quick effect is realized under the condition of not affecting the performance of the main business flow.

Meanwhile, the system analyzes the strategy version identification and the maximum fault-tolerant trigger times in the fault-tolerant threshold set, and performs version matching with the effective strategy entry in the strategy verification service database. Policy version identification is typically marked in the form of a policy type + version number (e.g., HEALTH-GUOCHENG-3.2) for pinpointing policy logic currently executing. The matching process needs to ensure the consistency of version granularity, and avoids the generation of historical legacy errors due to strategy adjustment.

And after the corresponding strategy entry is found, the system updates the maximum fault-tolerant trigger number field in the entry according to the maximum fault-tolerant trigger number, so that the tolerance of the strategy check logic is dynamically adjusted. For example, after a large number of interruptions in the short term, the fill limit, retry number, or tolerance boundary of the partial fields may be temporarily relaxed, enhancing the adaptability of the system to complex user input behaviors.

Finally, the updated policy entry will be synchronized to the business process execution service component. The synchronous operation can be realized through mechanisms such as database write monitoring, cache updating pushing or configuration center distribution, and the like, so that the service system is ensured to immediately adopt updated parameters when the policy verification is executed next time, and a rapid closed loop of early warning driving, parameter adjustment and service response is formed.

The method is characterized in that in a medical health insurance self-service H5 insurance application process of an insurance company, a user can generate a formal insurance order by completing filling of a plurality of front-end forms and checking of business rules. Recently, the operation team finds that the jump-out rate of the page in the health notification field is obviously increased through data analysis, especially in the night peak period, the conversion rate fluctuation is obvious, and policy blocking or page experience abnormality is suspected. Therefore, the system enables the whole processing flow based on the visual early warning of the business flow interruption.

Firstly, physical interrupt nodes are deployed at each key page position of the application flow, wherein the physical interrupt nodes comprise front-end components such as health notification, occupation selection, information filling of an applicant, payment mode selection and the like. The physical interrupt nodes collect user behavior track data in the page operation process through the embedded monitoring script, and record original behavior logs including user input fields, button clicks, jump-out actions, residence time, equipment environment and network states.

When a user fails to submit in a certain page interrupt operation or repeatedly, the system analyzes and classifies the operation track data according to sequential logic preset in a service flow, and multi-stage logic interrupt type data of structural expression is generated. For example, if the "field is not filled completely" or "non-standard health status is selected" type errors occur repeatedly in the health notification page, the behavior is classified as "preset business policy check failure type", and if the user actively closes the page before seeing the "next" button, the behavior is classified as "user active termination type". Each record is assigned a primary logical interrupt type (e.g., "health notification phase") and a secondary logical interrupt subclass (e.g., "notification field miss"), and records an interrupt timestamp.

The system maps the structured data into a visual topological graph, the health notification stage is used as a main node, and different secondary interrupt types are used as subordinate sub-nodes. The node color is determined by the time-thermal value of the interrupt event, the near-occurrence event is displayed in red, and the history event is displayed in blue. For example, if "health questions and answers are not checked out" as child nodes, and frequently occur in approximately 1 hour, the corresponding node is displayed in dark red. Each node also binds the policy version number referenced by the event (e.g., "HEALTH-GUOCHENG-3.2") and the failure control ID (e.g., "health_input_001").

The system further screens the entry with the logic interrupt type of 'preset service policy check failure type', and extracts the interrupt time stamp field in the entry. Classifying according to a set hour level statistics window (such as dividing every 1 hour), and counting the triggering times of interrupt events of each strategy type in each time window. For example, the "HEALTH-GUOCHENG-3.2" strategy accumulated 58 interruptions during the time period of "2024-05-1221:00-22:00" with a significant increase over 23 occurrences of the same time period on the previous day.

The system calculates the upper limit of the baseline times of the current time window (e.g. 42 times after the historical mean value is added with three times of standard deviation) according to the historical frequency data of the same time period in three days. Since the current trigger times 58 exceed the baseline upper limit, the system triggers a hierarchical early warning mechanism. And triggering the first-stage early warning when the configured classification strategy exceeds the upper limit of 25%, and immediately generating a first-stage early warning instruction containing the current time window, the overrun range and the strategy version identifier by the system.

The early warning instruction is pushed to a preset notification channel, wherein the notification channel comprises a mail end and a system message end, and the operation system and a policy auditor receive and process the early warning instruction. Meanwhile, the system analyzes the strategy failure data bound by the early warning instruction, obtains a control identifier (such as health_input_001) and a strategy version (HEALTH-GUOCHENG-3.2) from the strategy failure data, generates a priority coefficient (such as 1.6) according to the interrupt frequency of the control in three days, and generates a page element weight adjustment parameter.

After the front-end engine of the page receives the parameter, the visual saliency of the control is adjusted in real time, such as thickening fonts, highlighting red frames, improving the occurrence frequency of guiding prompts, and the like, so that a user can more easily identify and complete necessary filling items when revisiting the page. Meanwhile, the back-end policy service queries HEALTH-GUOCHENG-3.2 entries marked as effective in the policy database, temporarily adjusts the maximum fault-tolerant trigger number field to be a higher threshold (e.g. from 50 to 80), and reduces the possibility of interception caused by short-term behavior fluctuation.

After the strategy adjustment is synchronized to the service execution service, new fault-tolerant configuration is automatically started in the subsequent user access, so that the user is allowed to obtain more interaction attempt space in the operation process, and the interruption frequency is prevented from being further increased.

Through the flow, the system captures the strategy interruption fluctuation risk at the first time, automatically completes front-end interaction optimization and back-end strategy flexible adjustment, ensures service continuity and user conversion rate to the greatest extent, and simultaneously constructs an application flow monitoring and response system with real-time early warning capability and closed-loop self-adaptation capability.

According to the embodiment, the abnormal risk of the user behavior is converted into executable visual guidance and strategy tolerance control actions through fine analysis of the data fields associated with the early warning instructions, so that linkage optimization of the page end and the strategy end is achieved. The method has the advantages that the significance of the high-risk control is enhanced by injecting the weight adjustment parameters of the page elements, the probability of key operation of a user is improved, meanwhile, the fault tolerance capability of the system is improved by dynamically adjusting the strategy fault tolerance threshold, and the loss risk caused by misjudgment is reduced on the premise of not sacrificing compliance. Finally, the system can actively adjust interface interaction and strategy rules based on real behavior data, so that quick response, continuous optimization and risk control closed loop under a service interruption scene are realized, and the user application conversion rate and overall experience consistency are greatly improved.

In an embodiment, a service flow visual early warning and policy response device is provided, where the service flow visual early warning and policy response device corresponds to the service flow visual early warning and policy response method in the above embodiment one by one. Referring to fig. 3, fig. 3 is a schematic functional block diagram of a service flow visualization early warning and policy response device according to a preferred embodiment of the present invention. The system comprises an interrupt node configuration module 10, an operation track acquisition module 20, a logic interrupt classification module 30, a topology map construction module 40, an interrupt screening module 50, an interrupt frequency statistics module 60, an early warning trigger module 70 and a strategy optimization parameter generation module 80. The functional modules are described in detail as follows:

the interrupt node configuration module 10 is configured to set a plurality of physical interrupt nodes at a front end interface of a service flow;

an operation track collection module 20, configured to collect operation track data of an interrupt service flow through the physical interrupt node;

The logic interrupt classification module 30 is configured to classify the operation track data into multi-level logic interrupt type structured data according to a service flow sequence;

A topology map construction module 40, configured to map the multi-level logical interrupt type structured data into a visual topology map;

The interrupt screening module 50 is configured to screen the multi-level logic interrupt type structured data for an entry whose logic interrupt type is a preset service policy check failure type;

The interruption frequency statistics module 60 is configured to count the trigger times of the preset service policy check failure types according to the interruption timestamp fields in the screened entries, and generate interruption frequency data according to the interruption timestamp fields, the trigger times and the preset service policy check failure types;

The early warning triggering module 70 is configured to trigger a hierarchical early warning instruction including a preset service policy version identifier when the interruption frequency data exceeds the upper limit of the baseline frequency;

The policy optimization parameter generating module 80 is configured to generate a page element weight adjustment parameter and a fault tolerance threshold set corresponding to a preset service policy version according to the preset service policy verification failure type data associated with the hierarchical early warning instruction.

In one embodiment, the logic interrupt classification module 30 is specifically configured to:

analyzing the last operation node identification and the interrupt time stamp field in the operation track data;

matching a verification failure code corresponding to the last operation node identifier in a preset service policy library;

Associating a currently effective service policy version identifier with the verification failure code;

Generating a primary logic interrupt type and a secondary logic interrupt subclass according to the service flow sequence, wherein the primary logic interrupt type corresponds to the service flow stage classification, and the secondary logic interrupt subclass corresponds to a verification failure scene defined in a preset strategy library;

Converting the interrupt time stamp into a time sequence field in a preset time zone format;

And combining the verification failure code, the currently effective service policy version identifier, the primary logic interrupt type, the secondary logic interrupt subclass and the time sequence field to generate the multi-stage logic interrupt type structured data.

In one embodiment, the topology map construction module 40 is specifically configured to:

Extracting a primary logic interrupt type field and a secondary logic interrupt subclass field in the multi-stage logic interrupt type structured data;

mapping the primary logic interrupt type to a master node of a topological graph according to a preset visual coding strategy, and mapping the secondary logic interrupt subclass to a slave child node of the master node;

Extracting a time sequence field in the multi-stage logic interrupt type structured data, and converting the time sequence field into a time sequence thermal value;

Generating a node hue gradient according to the time sequence thermodynamic value;

generating a strategy version label based on a service strategy version identification field in the multistage logic interrupt type structured data, and binding the strategy version label to a corresponding master node and a corresponding slave sub-node;

And generating a connection edge of the topological graph based on the hierarchical relationship between the master node and the slave sub-nodes, and applying the node hue gradient to the filling attribute of the corresponding node to form a visual topological graph.

acquiring time interval parameters and commodity classification codes configured through an interactive interface;

Extracting a logic interrupt type subset which can be matched with the time interval parameter and the commodity classification code from the visual topological graph;

Performing visual hiding operation on the logic interrupt types which are not matched with the time interval parameters or the commodity classification codes;

Determining the ring ratio change rate of each node in the logic interrupt type subset based on historical service period data;

Marking the logic interrupt type with the ring ratio change rate exceeding a preset change rate threshold as a high-attention logic interrupt type, and marking physical interrupt nodes belonging to the high-attention logic interrupt type in the visual topological graph by pulse special effects;

and generating an abnormal category optimization list based on the commodity classification codes corresponding to the high-attention logic interrupt types.

In one embodiment, the interrupt frequency statistics module 60 is specifically configured to:

classifying each item into a corresponding preset time window based on the interrupt timestamp field in the screened item, wherein the preset time window is a predefined statistical period interval;

grouping and aggregating the items classified to the same preset time window based on the preset service policy check failure types to generate a policy check failure type item set arranged according to the time window sequence;

Counting the triggering times of interrupt events of a strategy verification failure type item set in each preset time window, wherein the triggering times are the accumulated number of all items in the corresponding time window, and each item corresponds to an interrupt event of one preset service strategy verification failure;

And combining the preset time window, the triggering times of the interrupt event and the verification failure type of the preset service strategy to generate the interrupt frequency data.

In one embodiment, the early warning triggering module 70 is specifically configured to:

Determining the baseline frequency upper limit of each preset time window based on the triggering frequency field and the historical contemporaneous time window field in the historical interrupt frequency data;

Configuring a grading early warning strategy to define early warning levels corresponding to different overrun amplitudes of which the triggering times exceed the upper limit of the baseline times;

When the triggering times of the interruption events in the interruption frequency data exceed the upper limit of the baseline times of the current preset time window, matching the grading early warning strategy according to the overrun amplitude, and generating a grading early warning instruction corresponding to the early warning level;

and sending a grading early warning instruction to the configured early warning notification receiving end through a preset notification channel.

In one embodiment, the policy optimization parameter generating module 80 is specifically configured to:

analyzing a control identifier and a priority coefficient in the page element weight adjustment parameter;

Matching the control identifier with a document object model node in an interface rendering service component;

According to the priority coefficient, adjusting the weight value of the page element of the matched document object model node;

Injecting the adjusted page element weight value into a browser rendering pipeline to improve the visual significance of the control corresponding to the failure type of the verification of the preset service strategy;

Analyzing the strategy version identification and the maximum fault-tolerant trigger times in the fault-tolerant threshold set;

Performing version matching on the strategy version identifier and an effective strategy entry in a strategy verification service database;

Updating the maximum fault-tolerant trigger number field in the matched effective strategy entry according to the maximum fault-tolerant trigger number;

And synchronizing the updated effective strategy entry to a business process execution service component so as to adjust the triggering condition of the preset business strategy verification failure type.

In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes non-volatile and/or volatile storage media and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external user terminal through network connection. The computer program, when executed by the processor, implements functions or steps of a service side of a service flow visual early warning and policy response method.

In one embodiment, a computer device is provided, which may be a user terminal, and the internal structure of which may be as shown in fig. 5. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is for communicating with an external server via a network connection. The computer program when executed by the processor realizes the functions or steps of a service flow visual early warning and strategy response method on the user side

In one embodiment, a computer device is provided that includes a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program:

In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:

It should be noted that, the functions or steps that can be implemented by the computer readable storage medium or the computer device may correspond to the descriptions of the server side and the client side in the foregoing method embodiments, and are not described herein for avoiding repetition.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (SYNCHLINK) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.

It should be noted that, if a software tool or component other than the company appears in the embodiment of the present application, the embodiment is merely presented by way of example, and does not represent actual use. The foregoing embodiments are merely illustrative of the technical solutions of the present application, and not restrictive, and although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that modifications may still be made to the technical solutions described in the foregoing embodiments or equivalent substitutions of some technical features thereof, and that such modifications or substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A business process visualization early warning and strategy response method is characterized by comprising the following steps:

2. The business process visualization early warning and strategy response method of claim 1, wherein classifying the operation trace data into multi-level logical interrupt type structured data according to a business process sequence comprises:

3. The business process visualization early warning and policy response method of claim 1, wherein mapping the multi-level logical interrupt type structured data into a visualization topology comprises:

4. The business process visualization early warning and policy response method according to claim 1, wherein after mapping the multi-level logical interrupt type structured data into a visualization topological graph, further comprising:

5. The method for visualizing an early warning and a policy response according to claim 1, wherein counting the number of triggers of the preset service policy check failure type according to the interrupt timestamp field in the screened entry, and generating interrupt frequency data according to the interrupt timestamp field, the number of triggers and the preset service policy check failure type, comprises:

6. The method for visualizing an early warning and a policy response in a business process according to claim 1, wherein when the interruption frequency data exceeds the upper limit of the baseline frequency, triggering a hierarchical early warning instruction including a preset business policy version identifier, includes:

7. The business process visualization early warning and strategy response method according to claim 1, wherein after generating the page element weight adjustment parameter and the fault tolerance threshold set corresponding to the preset business strategy version according to the preset business strategy verification failure type data associated with the hierarchical early warning instruction, the method further comprises:

8. The utility model provides a visual early warning of business flow and strategy response device which characterized in that, visual early warning of business flow and strategy response device includes:

The interrupt frequency statistics module is used for counting the trigger times of the preset service policy check failure types according to the interrupt timestamp fields in the screened entries, and generating interrupt frequency data according to the interrupt timestamp fields, the trigger times and the preset service policy check failure types

9. A computer device comprising a memory, a processor, and a business process visualization early warning and policy response program stored on the memory and executable on the processor, the business process visualization early warning and policy response program implementing the steps of the business process visualization early warning and policy response method of any one of claims 1-7 when executed by the processor.

10. A computer readable storage medium, wherein a business process visualization early warning and policy response program is stored on the storage medium, and the business process visualization early warning and policy response program when executed by a processor implements the steps of the business process visualization early warning and policy response method according to any one of claims 1-7.