CN107203529B - Method and device for multi-service correlation analysis based on similarity of metadata graph structure - Google Patents

Abstract

The invention discloses a method and a device for analyzing multi-service relevance based on structural similarity of a metadata graph, wherein the method comprises the following steps: after metadata are obtained from a plurality of services, a relation graph of metadata objects is established; judging whether the same meta-model in the relation graph of the metadata object has the common metadata object and the attribute of the metadata object, if so, obtaining the structural similarity of the metadata object according to the similarity of the vertex, the attribute of the vertex and the edge of the structure in the relation graph of the metadata object; determining an incidence relation among a plurality of businesses based on the structural similarity of the metadata graphs.

Description

Method and device for multi-service correlation analysis based on similarity of metadata graph structure

technical field

The invention relates to correlation analysis technology, in particular to a multi-service correlation analysis method and device based on the similarity of metadata graph structure.

Background technique

Metadata refers to the data that describe the data, mainly describing the concepts, relationships, rules, semantics, etc. of the domain. Metadata is an effective way to manage massive data systems (such as data warehouses, data marts, Hadoop big data platforms, etc.), it can provide a clear and complete catalog for accessing data, so that users can clearly understand the data as a whole and guide users Use data efficiently.

When using the existing technology to perform correlation analysis based on metadata, there are mainly the following defects:

First, a relationship link of metadata is a business process with direct reference relationship or direct data flow relationship from beginning to end. In general, there are many indirect connections between multiple businesses of an enterprise, but the existing The metadata system has no way to determine the key connection points between multiple businesses, so when a business caliber changes, it cannot intuitively evaluate the impact of this business on other businesses, and can only use manual backtracking to find each metadata object Impact on business processes.

Second, the existing metadata correlation analysis only roughly compares the number of overlapping metadata objects in the two relational links. In fact, different businesses often use different attributes of metadata objects and the logical process relationships are often different. Therefore, the results of association analysis that do not consider metadata object attribute information and logical relationships often lack accuracy.

Third, at present, the classification of business mainly relies on the manual classification of business personnel with experience, which can be handled reluctantly when the amount of data is small, but in the face of massive and complex big data, manual classification is obviously not enough, and the existing metadata application system Methods and mechanisms to assist classification are lacking.

SUMMARY OF THE INVENTION

In view of this, the embodiments of the present invention hope to provide a method and device for analyzing multi-service correlation based on the similarity of metadata graph structure, which at least solves the problems existing in the prior art.

The technical solution of the embodiment of the present invention is realized as follows:

According to an embodiment of the present invention, a method for analyzing multi-service correlation based on the similarity of metadata graph structure, the method includes:

After obtaining metadata from multiple services, establish a relationship diagram of metadata objects;

Determine whether there are common metadata objects and metadata object attributes in the same metamodel in the relationship diagram of the metadata object, and if there are, according to the similarity of the vertices and vertex attributes and edges of the structure in the relationship diagram of the metadata object to obtain the structural similarity of the metadata graph;

Based on the structural similarity of the metadata graph, an association relationship between multiple services is determined.

In the above solution, after the metadata is obtained from multiple services, a relationship diagram of metadata objects is established, including:

The metadata is divided into multiple classes according to different granularities, and the description model established for each class is the meta model;

forming the metadata object from instances or entities of the metamodel;

A metadata relationship is established according to the reference or data flow relationship between the metadata objects, and a directed graph of the metadata objects is established with the metadata objects as vertices and the relationship between the metadata objects as edges, and the metadata A directed graph of data objects serves as a relational graph of the metadata objects.

In the above scheme, the method also includes:

The relationship between the resource objects involved in each business and the resource objects supports the use of a directed graph of the metadata objects to represent.

In the above scheme, according to the similarity of vertices and vertex attributes and edges of the structure in the relationship graph of the metadata object, the structural similarity of the metadata graph is obtained, including:

obtaining the similarity between the vertices of the structure in the relational graph of the metadata object and the vertex attributes;

obtaining the similarity of edges in the relational graph of the metadata object;

The structural similarity of the metadata graph is obtained according to the similarity of the vertex combined with the vertex attribute and the similarity of the edge.

In the above solution, the acquisition of the similarity between the vertex and vertex attributes of the structure in the relationship graph of the metadata object includes:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common vertices and their attributes of the two metadata subgraphs in the specified specification graph, and calculate the similarity between the vertices of the metadata subgraph structures corresponding to any two services combined with the vertex attributes according to the proportions.

In the above solution, the obtaining the similarity of edges in the relational graph of the metadata object includes:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common edges of the two metadata sub-graphs in the specified specification graph, and calculate the edge similarity of the metadata sub-graph structures corresponding to any two services according to the proportion.

In the above solution, the association relationship between multiple services is determined based on the structural similarity of the metadata graph, including:

The vertices of the metadata subgraph structure corresponding to any two services are combined with the similarity of the vertex attributes and the similarity of the edges of the metadata subgraph structure corresponding to the any two services, and the association between any different services is measured. sex;

According to the angle that needs to be paid attention to actually, the weight value is adjusted by the adjustment factor to obtain the service correlation degree value, and the correlation relationship between the multiple services is determined by the service correlation degree value.

According to an embodiment of the present invention, a multi-service correlation analysis device based on the similarity of metadata graph structure, the device includes:

The establishment unit is used to establish the relationship diagram of the metadata object after obtaining the metadata from multiple services;

A processing unit for judging whether there are common metadata objects and metadata object attributes in the same metamodel in the relational graph of the metadata objects, and if so, according to the vertices and vertices of the structure in the relational graph of the metadata objects The similarity of attributes and edges to obtain the structural similarity of the metadata graph;

A determining unit, configured to determine the association relationship between multiple services based on the structural similarity of the metadata graph.

In the above scheme, the establishment unit further includes:

a classification subunit, used to divide the metadata into multiple classes according to different granularities, and the description model established for each class is the meta model;

forming subunits for composing the metadata object from instances or entities of the metamodel;

The relationship establishment subunit is used to establish a metadata relationship according to the reference or data flow relationship between the metadata objects, and takes the metadata object as a vertex and the relationship between the metadata objects as an edge. A directed graph, using the directed graph of the metadata object as a relationship graph of the metadata object.

In the above scheme, the device also includes:

The relationship between the resource objects involved in each business and the resource objects supports the use of a directed graph of the metadata objects to represent.

In the above scheme, the processing unit further includes:

a first processing subunit, used to obtain the similarity of the vertex and vertex attributes of the structure in the relationship graph of the metadata object;

a second processing subunit, configured to obtain the similarity of edges in the relational graph of the metadata object;

The third processing subunit is configured to obtain the structural similarity of the metadata graph according to the similarity of the vertex-combined vertex attributes and the similarity of the edge.

In the above scheme, the first processing subunit is further used for:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common vertices and their attributes of the two metadata subgraphs in the specified specification graph, and calculate the similarity between the vertices of the metadata subgraph structures corresponding to any two services combined with the vertex attributes according to the proportions.

In the above scheme, the second processing subunit is further used for:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common edges of the two metadata sub-graphs in the specified specification graph, and calculate the edge similarity of the metadata sub-graph structures corresponding to any two services according to the proportion.

In the above scheme, the determining unit is further used for:

The vertices of the metadata subgraph structure corresponding to any two services are combined with the similarity of the vertex attributes and the similarity of the edges of the metadata subgraph structure corresponding to the any two services, and the association between any different services is measured. sex;

According to the angle that needs to be paid attention to actually, the weight value is adjusted by the adjustment factor to obtain the service correlation degree value, and the correlation relationship between the multiple services is determined by the service correlation degree value.

The multi-service correlation analysis method based on the similarity of the structure of the metadata graph according to the embodiment of the present invention includes: after obtaining the metadata from multiple services, establishing a relational graph of the metadata objects; judging that the relational graphs of the metadata objects are the same Whether there is a common metadata object and metadata object attribute in the metamodel, if so, obtain the structural similarity of the metadata graph according to the similarity of the vertices and vertex attributes and edges of the structure in the relationship graph of the metadata object; Based on the structural similarity of the metadata graph, an association relationship between multiple services is determined. By adopting the embodiments of the present invention, the accuracy and effect of the correlation analysis can be improved.

Description of drawings

1 is a schematic flowchart of a method according to an embodiment of the present invention;

2 is a schematic diagram of a three-tier architecture of a metadata system in an application scenario applying an embodiment of the present invention;

3 is a schematic diagram of a multi-service correlation analysis based on similarity of metadata graph structure in an application scenario applying an embodiment of the present invention;

FIG. 4 is a flowchart of multi-service correlation analysis based on the similarity of metadata graph structures in an application scenario applying an embodiment of the present invention.

Detailed ways

The implementation of the technical solution will be further described in detail below with reference to the accompanying drawings.

A method for analyzing multi-service correlation based on similarity of metadata graph structure according to an embodiment of the present invention, as shown in FIG. 1 , the method includes:

Step 101 , after obtaining metadata from multiple services, establish a relationship diagram of metadata objects.

Step 102: Judge whether there are common metadata objects and metadata object attributes in the same metamodel in the relationship diagram of the metadata object, and if so, then according to the vertices and vertex attributes of the structure in the relationship diagram of the metadata object and Edge similarity to get the structural similarity of the metadata graph.

Step 103: Determine the association relationship between multiple services based on the structural similarity of the metadata graph.

In an implementation manner of the embodiment of the present invention, after obtaining metadata from multiple services, establishing a relationship diagram of metadata objects includes: dividing the metadata into multiple classes according to different granularities, and each class The respectively established description models are the meta-model; the metadata objects are composed of instances or entities of the meta-model; the metadata relationship is established according to the reference or data flow relationship between the metadata objects, and the metadata The objects are vertices, the relationship between the metadata objects is an edge, a directed graph of the metadata objects is established, and the directed graph of the metadata objects is used as the relationship graph of the metadata objects.

In an implementation manner of the embodiment of the present invention, the method further includes: the resource objects involved in each service and the relationship between the resource objects are represented by the directed graph of the metadata object.

In an embodiment of the embodiment of the present invention, obtaining the structural similarity of the metadata graph according to the similarity of the vertices, vertex attributes and edges of the structure in the relationship graph of the metadata object includes: obtaining the metadata The vertices of the structure in the relational graph of the object are combined with the similarity of the vertex attributes; the similarity of the edges in the relational graph of the metadata object is obtained; according to the similarity of the vertexes combined with the similarity of the vertex attributes and the similarity of the edges, the meta-data is obtained. Structural similarity of data graphs.

In an implementation manner of the embodiment of the present invention, the obtaining the similarity between the vertices of the structure in the relationship graph of the metadata object and the vertex attributes includes: each service is performed with a metadata subgraph of a metadata directed graph. Representation; obtain the proportion of the common vertices and their attributes of the two metadata subgraphs in the specified specification graph (such as the minimum graph), and calculate the similarity between the vertices of the metadata subgraph structure corresponding to any two services combined with the vertex attributes according to the proportions Spend.

In an implementation manner of the embodiment of the present invention, the obtaining the similarity of edges in the relationship graph of the metadata object includes: each service is represented by a metadata subgraph of a metadata directed graph; obtaining two The common edge of the metadata subgraph accounts for the proportion of the specified specification graph (such as the minimum graph), and the similarity of the edges of the metadata subgraph structure corresponding to any two services is calculated according to the proportion.

In an implementation manner of the embodiment of the present invention, determining the association relationship between multiple services based on the structural similarity of the metadata graph includes: synthesizing the vertices of the metadata subgraph structure corresponding to any two services Combining the similarity of vertex attributes and the similarity of the edges of the metadata subgraph structure corresponding to any two services, the correlation between any different services is measured; according to the angle that needs to be paid attention to, the weight is adjusted by the adjustment factor to obtain A business correlation degree value, and the business correlation degree value determines the correlation relationship between multiple services.

The multi-service correlation analysis device based on the similarity of the structure of the metadata graph according to the embodiment of the present invention includes: a establishing unit for establishing a relationship graph of metadata objects after obtaining metadata from multiple services; and processing A unit for judging whether there are common metadata objects and metadata object attributes in the same metamodel in the relationship diagram of the metadata object, and if so, according to the vertices and vertex attributes of the structure in the relationship diagram of the metadata object and the similarity of the edges, to obtain the structural similarity of the metadata graph; and a determining unit for determining the association relationship between multiple services based on the structural similarity of the metadata graph.

In an implementation manner of the embodiment of the present invention, the establishing unit further includes:

a classification subunit, used to divide the metadata into multiple classes according to different granularities, and the description model established for each class is the meta model;

forming subunits for composing the metadata object from instances or entities of the metamodel;

The relationship establishment subunit is used to establish a metadata relationship according to the reference or data flow relationship between the metadata objects, and takes the metadata object as a vertex and the relationship between the metadata objects as an edge. A directed graph, using the directed graph of the metadata object as a relationship graph of the metadata object.

In an embodiment of the embodiment of the present invention, the device further includes:

The relationship between the resource objects involved in each business and the resource objects supports the use of a directed graph of the metadata objects to represent.

In an implementation manner of the embodiment of the present invention, the processing unit further includes:

a first processing subunit, used to obtain the similarity of the vertex and vertex attributes of the structure in the relationship graph of the metadata object;

a second processing subunit, configured to obtain the similarity of edges in the relational graph of the metadata object;

The third processing subunit is configured to obtain the structural similarity of the metadata graph according to the similarity of the vertex-combined vertex attributes and the similarity of the edge.

In an implementation manner of the embodiment of the present invention, the first processing subunit is further configured to:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common vertices and their attributes of the two metadata subgraphs in the specified specification graph, and calculate the similarity between the vertices of the metadata subgraph structures corresponding to any two services combined with the vertex attributes according to the proportions.

In an implementation manner of the embodiment of the present invention, the second processing subunit is further configured to:

Each business is represented by a metadata subgraph of a metadata directed graph;

Obtain the proportion of the common edges of the two metadata sub-graphs in the specified specification graph, and calculate the edge similarity of the metadata sub-graph structures corresponding to any two services according to the proportion.

In an implementation manner of the embodiment of the present invention, the determining unit is further configured to:

The vertices of the metadata subgraph structure corresponding to any two services are combined with the similarity of the vertex attributes and the similarity of the edges of the metadata subgraph structure corresponding to the any two services, and the association between any different services is measured. sex;

According to the angle that needs to be paid attention to actually, the weight value is adjusted by the adjustment factor to obtain the service correlation degree value, and the correlation relationship between the multiple services is determined by the service correlation degree value.

Taking a real application scenario as an example, the embodiments of the present invention are described as follows:

First, an application scenario of the embodiment of the present invention is introduced as follows:

In today's era of big data, the successful implementation and application of business intelligence BI depends on effective metadata management and application. Metadata is defined as data that describes other data, mainly including related topics, concepts, terms, structures, processes, relationships, and rules in the fields of business, technology, and management. High-level metadata applications can serve as guideposts for various complex systems and massive data, help users better understand the ins and outs of various businesses, enhance the basic support capabilities of data for businesses, and improve data quality management and control capabilities. Achieve efficient enterprise management. However, the current application of metadata is still in the stage of simple use, lacking high-level and in-depth research and application, and still needs great improvement in the analysis of complex relationships in multiple businesses.

The above application scenario adopts the embodiment of the present invention, which is to measure the correlation between multiple services according to the similarity of the vertices and their attributes and edges of the metadata directed graph corresponding to the services, which can reflect the impact of complex intersections between multiple services. It provides guidance for business personnel to be familiar with the relationship between multiple businesses, and provides decision-making for enterprises to conduct business analysis. The main problems that can be solved are: 1) By establishing a relationship graph of metadata objects and comparing the similarity of graph structure vertices and their attributes and edges to determine the relationship between multiple services, it can intuitively reflect the impact of one service change on other services. degree of influence. 2) Considering the situation of not using the same resource attributes and the logical relationship of the business before and after the business, and making full use of the vertices of the metadata object and their attribute information and the edges corresponding to the logical process relationship before and after to measure the similarity of the metadata graph structure, Make business relevance analysis results more credible. 3) After calculating the business correlation degree based on the similarity of the metadata graph structure, a series of applications can be created, such as: business change impact early warning system, redundant and repetitive processes of sorting and merging business, automatic auxiliary business classification, etc., which can solve large-scale problems. Some complex issues facing data.

The three-tier architecture of the metadata system is shown in Figure 2. The above application scenario adopts the embodiment of the present invention, and a multi-service meta-object analysis function module based on the similarity of the metadata graph structure is added (as shown in A15 in Figure 2), On this basis, high-level extended applications as shown in A16 in Figure 2 are further proposed, such as the business change warning module, the redundant repetitive process module for combing and merging business, and the automatic auxiliary business classification modules. The rest are A11, A12, A13, Modules marked by A14 are existing modules.

The main principle of the multi-service correlation analysis function module based on the similarity of the metadata graph structure is shown in Figure 3. Figure 3 is the principle diagram of the multi-service correlation analysis based on the similarity of the metadata graph structure, and the specific description is as follows:

First, the resource objects used by different businesses have similarities, that is, different businesses use some resource objects or some attributes of these resource objects in common, and the mapping metadata graph has common features for vertices and their attributes, then these There is a relationship between businesses.

For example, in Figure 3, the company has two market share reports corresponding to two businesses. The data of these two reports are summarized through the same table, but some of the same fields of the table are used according to different calibers. Then the two business-related reports are related by the tables involved and the attribute fields of the tables.

At present, structured data (eg, relational databases, OLAP online analysis data, etc.) and description information of unstructured data (eg, log files, XML files, Webservice interfaces, Hadoop platform data, etc.) are the common main bodies for generating metadata , by extracting and entering the description data of these data automatically or manually is the main way for the acquisition layer of the metadata system to obtain the data.

根据元对象之间的引用或数据流向关系建立元数据关系，表示成r_χ,γ即元数据对象

与

之间的关系。以元对象作为顶点，元对象之间的关系作为边，那么可以建立元数据的有向图，表示成G＝〈V,E>，其中顶点表示成集合

边表示成邻接矩阵

这样，每个业务涉及的资源对象和这些资源对象之间的关系都可以用元数据的有向图表示出来。在元数据的功能层上，以对业务进行抽象获得元数据的有向图为基础，比较元数据有向图中同一元模型是否存在共同的元对象以及元对象的属性，根据元数据图结构的顶点和顶点属性的相似性衡量不同业务之间的关联性。In the logic layer of the metadata system, metadata is divided into delta categories according to different granularities, and a description model is established for each category, called a metamodel, so that all metadata can be classified according to the metamodel and represented as a Set M={m ₁ ,m ₂ ,...,m _δ }, where each metamodel m _χ can be described by several attributes, namely m _χ =(a ₁ ,a ₂ ,...,a _κ ) . An instance or entity of a metamodel is called a metaobject and is represented as

The metadata relationship is established according to the reference or data flow relationship between the metadata objects, which is expressed as r _χ,γ , which is the metadata object

and

The relationship between. Taking meta-objects as vertices and the relationship between meta-objects as edges, then a directed graph of metadata can be established, represented as G=<V, E>, where vertices are represented as sets

Edges are represented as adjacency matrices

In this way, the resource objects involved in each business and the relationship between these resource objects can be represented by a directed graph of metadata. On the functional layer of metadata, based on the directed graph of metadata obtained by abstracting business, compare whether there are common meta objects and attributes of meta objects in the same meta model in the directed graph of metadata, according to the structure of the meta data graph The similarity of vertices and vertex attributes measures the relevance between different services.

和

的属性之间的相似度，计算公式(1-1)如下：Since the attribute dimensions of the same meta-model are the same, the dimensions of the attributes of the meta-objects and meta-objects derived from the same meta-model are also the same, but because the services are different, the specific meta-object or its attribute values may be Differently, using the embodiment of the present invention, the cosine similarity is used to measure different meta-objects of the same meta-model m _χ .

and

The similarity between the attributes of , the calculation formula (1-1) is as follows:

属性不为空，则表示为1，否则为0。Among them, if of

If the property is not empty, it is represented as 1, otherwise it is 0.

Each business is represented by a subgraph of the metadata directed graph. Considering the proportion of the common vertices and their attributes in the minimum graph of the two graphs, the vertex combination of the metadata graph structure corresponding to any two services α and β can be calculated. The similarity of vertex attributes, as shown in formula (1-2):

where the subgraphs g _α , g _β ∈ G.

Second, the logic processes of different businesses are similar, that is, these different businesses use the logic flow from some resource objects or their attributes to other resource objects or their corresponding properties, and the mapping metadata graph is a common continuous Directed edges, these services are related.

In this embodiment, from the perspective of the abstract metadata directed graph, if there are common continuous and directed edges between the meta objects, the difference can be measured according to the similarity of the edges of the metadata graph structure. correlation between businesses. For example, in the above example, the two business-related reports of the company are related by the common table and the attribute field of the table, and this table and field are processed by the same stored procedure, so there is a process from the stored procedure to the table. A continuous, directed logical link between its fields.

Considering the proportion of the common edge of the metadata directed graph corresponding to the two services in the smallest graph, the edge similarity of the metadata graph structure corresponding to any two services α and β can be calculated, as shown in formula (1-3):

3. Combine the two aspects of the vertices of the metadata graph structure, the similarity of vertex attributes and the similarity of edges to measure the correlation between multiple different services. According to the angle that needs to be paid attention to, the weights are adjusted by adjusting factors to obtain Business relevance value.

In reality, the resource objects and attributes commonly used by the two services and the business logic process are often considered at the same time to compare the relationship between the two services. Therefore, this embodiment combines the above two formulas (1-2) and (1-3), The correlation formula for calculating any two services α and β is proposed, as shown in (1-4):

rel(α,β)=sim(g _α ,g _β )=θ·sv(g _α ,g _β )+(1-θ)·se(g _α ,g _β ) (1-4)

If one of the two services α and β is a sub-service of the other service, the correlation degree of the two services is 100%, that is, rel(α, β)=1.

Fourth, according to the correlation value between different businesses, a series of applications can be created, such as: business change impact warning system, redundant and repetitive processes for sorting and merging businesses, automatic auxiliary business classification, etc.

On the functional layer of the metadata system, a series of advanced extended applications can be established by using the multi-service correlation analysis results of the similarity of the metadata graph structure.

Specifically, the business change impact early warning system can evaluate the impact of a business change operation on other businesses in advance. If the business correlation is high and the impact exceeds the early warning threshold, an alert will be issued, so as to avoid considering changing only one business and ignoring others Serious adverse effects on business.

The application of sorting out and merging redundant and repetitive processes of businesses can find out the redundant and repetitive processes that may exist in these businesses according to the degree of business relevance, and adjust and merge these processes, thereby saving resources and costs.

The application of automatic auxiliary business classification can perform automatic auxiliary classification according to the relevance of the business and the existing business categories, reducing the workload of manual classification.

Figure 4 is a flow chart of the multi-service correlation analysis based on the similarity of the metadata graph structure. As shown in Figure 4, the entire process of the multi-service correlation analysis method based on the similarity of the metadata graph structure is as follows:

Step 11: Metadata collection is performed on the services that need to be managed.

Here, metadata collection includes business system interfaces, source codes, documents, database tables, views, stored procedures, ETL data extraction, cleaning, transformation, mapping, loading and other rules, data model API, OLAP of modeling tools Collection of resource object descriptions such as online analysis data and collection of relational rule descriptions, in which structured data can be obtained through the data dictionary, and unstructured data including XML files, log files, Webservice interfaces, Hadoop platforms, etc. are parsed by providing standard rules. way to obtain.

Step 12, establish a meta model according to the specified granularity, each meta model m _χ is described by several attributes, that is, m _χ =(a ₁ ,a ₂ ,...,a _κ ), and the metadata is classified according to the meta model .

根据元对象之间的引用/被引用或数据流出/流入等规则建立元数据关系rχ,γ即元数据对象

与

之间的关系。以元对象作为顶点，元对象之间的关系作为边，建立元数据的有向图G＝〈V,E〉，其中顶点为集合边为邻接矩阵Step 13, describe the metadata according to the meta model and create a meta object

Establish a metadata relationship rχ, γ is the metadata object according to the rules of referencing/referencing or data flowing/flowing between meta-objects

and

The relationship between. Using meta-objects as vertices and the relationship between meta-objects as edges, establish a directed graph G=<V, E> of metadata, where vertices are sets edge is an adjacency matrix

Step 14. According to the principle that some resource objects are commonly used by different services or some attributes of these resource objects are associated with these services, calculate any two according to the aforementioned formulas (1-1) and (1-2). The vertices of the metadata graph structure corresponding to each business α and β combine the similarity of the vertex attributes.

Step 15. According to different services, the logic flow from some resource objects or their attributes to other resource objects or their corresponding attributes is used, then these services are related to each other, according to the aforementioned formula (1-3) Calculate the edge similarity of the metadata graph structure corresponding to any two services α and β.

Step 16: Calculate the correlation degree of any two services α and β according to formula (1-4) by integrating the two aspects of the vertices and the similarity of the vertex attributes and the similarity of the edges of the metadata graph structure.

Step 17: Determine whether the two services α and β are related according to the correlation degree threshold of the two services α and β;

Step 18: Create a series of applications according to the correlation degree values of the two services α and β.

Here, in terms of creating a series of applications based on the correlation value, such as: business change impact early warning system, combing and merging redundant and repetitive processes of business, automatic auxiliary business classification, etc.

With the embodiments of the present invention, 1) some resource objects or some attributes thereof are commonly used according to different services, and business logics from some resource objects or their attributes to some other resource objects or their corresponding attributes are commonly used The similarity of the vertices, attributes and edges of the metadata directed graph structure abstracted by the similarity of the process, and the correlation between multiple services is measured according to the similarity; 2) Use the vertices of the metadata directed graph structure and its attributes and edge similarity to analyze the correlation between multiple services and implement the metadata correlation analysis module of the application layer. The embodiment of the present invention makes up for the deficiency that the existing metadata technology cannot handle the complex relationship between multiple services. The proposed method for analyzing the multi-service correlation based on the similarity of the metadata graph structure, through the establishment of the metadata object Relationship graph and compare the similarity of graph structure vertices and their attributes and edges to determine the relationship between multiple businesses, which can intuitively reflect the impact of a business change on other businesses. On this basis, a series of applications can be created to realize business Change early warning, redundant business process merging and automatic auxiliary business classification, etc., solve some problems faced in big data. This scheme has high practicability in practical application.

If the integrated modules described in the embodiments of the present invention are implemented in the form of software function modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention may be embodied in the form of software products in essence or the parts that make contributions to the prior art. The computer software products are stored in a storage medium and include several instructions for A computer device (which may be a personal computer, a server, or a network device, etc.) is caused to execute all or part of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, removable hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes . As such, embodiments of the present invention are not limited to any particular combination of hardware and software.

Correspondingly, an embodiment of the present invention further provides a computer storage medium, in which a computer program is stored, and the computer program is used to execute the multi-service correlation analysis method based on the similarity of the metadata graph structure according to the embodiment of the present invention.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention.

Claims (8)

1. A multi-service relevance analysis method based on metadata graph structure similarity is characterized by comprising the following steps:

after metadata are obtained from a plurality of services, a relation graph of metadata objects is established; the relation graph of the metadata object is a directed graph of the metadata object;

judging whether the same meta-model in the relation graph of the metadata object has a common metadata object and metadata object attribute, if so, obtaining the structural similarity of the relation graph of the metadata object according to the similarity of the vertex, the vertex attribute and the edge of the structure in the relation graph of the metadata object;

determining an incidence relation among a plurality of businesses based on the structural similarity of the relation graph of the metadata object;

wherein, the obtaining the structural similarity of the relationship graph of the metadata object according to the similarity of the vertex, the vertex attribute and the edge of the structure in the relationship graph of the metadata object comprises: acquiring similarity of vertex combination vertex attributes of the structure in the relation graph of the metadata object; acquiring the similarity of edges in the relation graph of the metadata object; according to the similarity of the vertex combined with the vertex attributes and the similarity of the edges, obtaining the structural similarity of the relation graph of the metadata object;

the obtaining of the similarity of the vertex combination vertex attributes of the structure in the relationship graph of the metadata object includes:

each service is represented by a metadata sub-graph of a directed graph of metadata objects; acquiring the proportion of the common vertex of the two metadata subgraphs and the attribute thereof in a specified specification graph, and calculating the similarity of the vertex combination vertex attribute of the metadata subgraph structure corresponding to any two services according to the proportion;

the obtaining of the similarity of the edges in the relationship graph of the metadata object includes: each service is represented by a metadata sub-graph of a directed graph of metadata objects; and acquiring the proportion of the common edge of the two metadata subgraphs in the specified specification graph, and calculating the similarity of the edges of the metadata subgraph structures corresponding to any two services according to the proportion.

2. The method of claim 1, wherein after obtaining metadata from the plurality of services, establishing a relationship graph of metadata objects comprises:

dividing the metadata into a plurality of classes according to different granularities, wherein a description model established by each class is the meta-model;

composing the metadata object from instances or entities of the meta-model;

and establishing a metadata relationship according to the reference or data flow relationship among the metadata objects, establishing a directed graph of the metadata objects by taking the metadata objects as vertexes and the relationship among the metadata objects as edges, and taking the directed graph of the metadata objects as the relationship graph of the metadata objects.

3. The method of claim 2, further comprising:

the resource objects involved by each business and the relationships between the resource objects support representation using directed graphs of the metadata objects.

4. The method of claim 1, wherein determining the associative relationship between the plurality of businesses based on the structural similarity of the relationship graph of the metadata object comprises:

combining the vertex of the metadata sub-graph structure corresponding to any two services with the similarity of the vertex attributes and the similarity of the edges of the metadata sub-graph structure corresponding to any two services, and measuring the relevance between any different services;

and according to the angle which needs attention actually, the weight is adjusted through the adjustment factor to obtain a service relevance value, and the relevance relation among a plurality of services is determined according to the service relevance value.

5. A multi-service relevance analysis apparatus based on metadata graph structure similarity, the apparatus comprising:

the system comprises an establishing unit, a processing unit and a processing unit, wherein the establishing unit is used for establishing a relation graph of a metadata object after acquiring metadata from a plurality of services;

the processing unit is used for judging whether the same meta-model in the relation graph of the metadata object has the common metadata object and the metadata object attribute, and if the same meta-model in the relation graph of the metadata object has the common metadata object and the common metadata object attribute, the structural similarity of the relation graph of the metadata object is obtained according to the similarity of the vertex, the vertex attribute and the edge of the structure in the relation graph of the metadata object;

a determining unit, configured to determine an association relationship between multiple businesses based on structural similarity of the relationship graph of the metadata object;

wherein the processing unit further comprises: the first processing subunit is used for acquiring similarity of vertex combination vertex attributes of a structure in a relational graph of the metadata object;

the second processing subunit is used for acquiring the similarity of edges in the relation graph of the metadata object;

the third processing subunit is used for obtaining the structural similarity of the relation graph of the metadata object according to the similarity of the vertex combined with the vertex attribute and the similarity of the edge;

the first processing subunit further configured to: each service is represented by a metadata sub-graph of a directed graph of metadata objects; acquiring the proportion of the common vertex of the two metadata subgraphs and the attribute thereof in a specified specification graph, and calculating the similarity of the vertex combination vertex attribute of the metadata subgraph structure corresponding to any two services according to the proportion;

the second processing subunit further configured to: each service is represented by a metadata sub-graph of a directed graph of metadata objects; and acquiring the proportion of the common edge of the two metadata subgraphs in the specified specification graph, and calculating the similarity of the edges of the metadata subgraph structures corresponding to any two services according to the proportion.

6. The apparatus of claim 5, wherein the establishing unit further comprises:

the classification subunit is used for dividing the metadata into a plurality of classes according to different granularities, and the description model established by each class is the meta-model;

a composition subunit for composing the metadata object from instances or entities of the meta-model;

and the relationship establishing subunit is used for establishing a metadata relationship according to the reference or data flow direction relationship among the metadata objects, establishing a directed graph of the metadata objects by taking the metadata objects as vertexes and the relationship among the metadata objects as edges, and taking the directed graph of the metadata objects as the relationship graph of the metadata objects.

7. The apparatus of claim 6, further comprising:

the resource objects involved by each business and the relationships between the resource objects support representation using directed graphs of the metadata objects.

8. The apparatus of claim 5, wherein the determining unit is further configured to:

combining the vertex of the metadata sub-graph structure corresponding to any two services with the similarity of the vertex attributes and the similarity of the edges of the metadata sub-graph structure corresponding to any two services, and measuring the relevance between any different services;

and according to the angle which needs attention actually, the weight is adjusted through the adjustment factor to obtain a service relevance value, and the relevance relation among a plurality of services is determined according to the service relevance value.

Images