CN114511122A - Dynamic adjustment method, system, device and storage medium for air ticket reservation - Google Patents

Abstract

The invention provides a method, a system, equipment and a storage medium for dynamically adjusting airline ticket booking, wherein the method comprises the following steps: extracting the characteristics of the air ticket data through a decision tree, and cutting subdivided nodes to obtain N characteristics which influence the orderable booking and serve as important characteristics, wherein N is a preset value; obtaining a non-selling entity based on historical data of the last time window based on an algorithm model; obtaining an upsell duration for each upsell entity based on the number of upsell times in a number of past continuous time windows; entering into the forbidden sale when the passing rate pass of the forbidden sale entity is smaller than the forbidden sale threshold; otherwise, no sale prohibition is entered; a unique selling-prohibited entity is obtained through the de-duplication merging algorithm. The invention can find the ticket management defect of the airline department in time and make dynamic production adjustment in time, thereby improving the flow experience of booking airline tickets for users.

Description

Dynamic adjustment method, system, device and storage medium for air ticket reservation

technical field

The present invention relates to the field of interactive operation, in particular, to a method, system, device and storage medium for dynamic adjustment of air ticket reservation.

Background technique

When users order international air tickets, the pre-booking process mainly goes through five pages, the query page for specifying query conditions, the filling page for entering passenger information, the value-added page for purchasing additional products or services, the payment page for paying fees, and confirming the completion of the order details. Page. After the user specifies the itinerary type, departure destination, date, cabin class, passenger type and other query parameters on the query page, the system initiates a search for each quotation system for the user, merges the results and displays it to the user. After browsing the quotation list, the user selects the desired flight group and starts the booking process based on multiple factors such as route, quotation, refund and modification, baggage, and auxiliary products. On each page, some key verifications will be initiated in the background to help users better confirm the itinerary. For example, when the user enters the filling page, the background will initiate Bookability, including Fare Validation and Availability Validation. The freight rate verification is used to verify whether the freight calculation and tax calculation are correct when the engine is quoted, and the shipping space verification is used to verify whether the shipping space obtained during the quotation is accurate. For another example, when the user is on the value-added page, the background will initiate a reservation operation, and initiate an application for a reserved seat to the Global Distribution System (GDS), and the GDS will also initiate a request to the CRS (Central Reservation System) of the airline reservation system. In this process, the airline conducts the verification of various information such as passengers and itineraries, the most important of which is the verification of the GDS class data and the airline's class consistency. If these key verifications fail, there will be front-end interception prompts such as "price unavailable" and "space sold out", and the booking experience is not smooth, causing user dissatisfaction. Users often cannot effectively distinguish the interception caused by external factors or Trip factors, and even lead to the misunderstanding of the so-called "more expensive" and "big data kill". It can be seen that a reliable quotation system needs to continuously output high-quality and high-stability quotations under the dual constraints of cost and coverage.

Therefore, the present invention provides a dynamic adjustment method, system, device and storage medium for air ticket reservation.

SUMMARY OF THE INVENTION

In view of the problems in the prior art, the purpose of the present invention is to provide a dynamic adjustment method, system, equipment and storage medium for air ticket reservation, which overcomes the difficulties of the prior art, and can timely discover the ticket management defects of airlines and make production in time. Dynamic adjustment to improve the user experience in the process of booking air tickets.

An embodiment of the present invention provides a dynamic adjustment method for air ticket reservation, comprising the following steps:

S110. Perform feature extraction on the air ticket data through a decision tree, and prune subdivided nodes to obtain N features that affect bookability and reservation as important features, where N is a preset value;

S120, obtaining a prohibited entity based on the algorithm model based on the historical data of the previous time window;

S130. For each prohibited sales entity, obtain the prohibited sales duration based on the number of prohibited sales in the past several consecutive time windows;

S140. When the pass rate pass of the prohibited sales entity is less than the prohibited sales threshold, the sales prohibition is entered; otherwise, the prohibited sales are not entered;

S150, obtaining a unique prohibited entity through the algorithm of de-duplication and merging.

Preferably, in the step S110, the important features include at least one of the itinerary type, departure city, departure country, cabin sales place, reservation data warehouse, and ticket issuing desk.

Preferably, the step S120 includes the following steps:

S121, the algorithm learns the historical data of the previous time window, constructs a model, and applies it to the current time window;

S122, select a time window t, each time window has the same length, and the length is r;

S123, select a stage f, an orderable stage or a reservation stage, based on the N feature dimensions as the data columns, and the data of this stage within the preset time range as the data row M to form a data matrix and train the model;

S124, loop i=0:N-1, and execute each round of level i:

个特征，基于上述数据矩阵，聚合N个特征维度，分别计算总量和通过率，生成由(N-i)个明确特征和i个虚化特征构成的禁售实体。S125. In the i-th round, select i features, for

Based on the above data matrix, aggregate N feature dimensions, calculate the total amount and pass rate respectively, and generate a prohibited entity composed of (Ni) clear features and i virtual features.

Preferably, in the step S122, as the length r of the time window decreases, the accuracy of the model increases;

The step S124 includes: i=0 round is the most fine-grained, and when the i=0 round feature is lower than the set threshold, the data object is marked with a prohibited sale.

Preferably, the step S130 includes the following steps:

S131. Obtain the prohibited sales status of each of the prohibited sales entities in the past T consecutive time windows, and define the state st=0, which means that the sales are not prohibited; st=1, then the sales are prohibited, and the value range of t is 1 to T;

计算历史窗口期中该实体被标记为禁售的次数；S132, define penalty factor

Calculate the number of times the entity has been marked as banned during the historical window;

表示历史窗口期中该实体被连续标记为不禁售的最长长度的比例，其中，L表示历史时间窗口从t＝1开始，st＝0可连续的最大长度；S133. Define reward factor

Represents the ratio of the longest length of the entity that is continuously marked as not banned for sale in the historical window period, where L represents the maximum length of the historical time window starting from t=1 and st=0;

S134 , obtaining the prohibition duration d=(1-pass)*r*(1+max{p-b,0}).

Preferably, the step S140 includes the following steps:

S141. Based on the intersection of the real situation and the model predicted situation in the confusion matrix, four situations of true negative, false negative, false positive and true positive are obtained. Among them, the negative corresponding model does not implement the ban, the positive corresponding model implements the ban, and the true corresponds to The prediction results of the model are consistent with the actual situation, and the prediction results of the pseudo-corresponding model are inconsistent with the actual situation;

S142. Define the improvement rate as the proportion that is actually intercepted and banned from sales, namely TN/All;

S143. Define the false ban rate as the proportion that actually passed but entered the ban, namely FN/All;

S144, traverse all prohibition thresholds according to the preset step interval, and calculate the curve of improvement rate and false prohibition rate under the threshold;

S145, calculate the marginal utility of improvement and the marginal cost of false prohibition, find a prohibition threshold with a large improvement rate and a small false prohibition rate, and consider it to be the most valuable prohibition threshold;

S146. Use the ban threshold with a large improvement rate and a small false ban rate as the most fine-grained ban threshold, and the ban thresholds of other levels are equal to the ban threshold/N with a large improvement rate and a small false ban rate Step decay.

Preferably, in the step S150, when the high-level prohibition message includes a low-level prohibition message, the low-level prohibition message is removed, and only the high-level prohibition message is retained.

An embodiment of the present invention also provides a dynamic adjustment system for air ticket reservation, which is used to realize the above-mentioned dynamic adjustment method for air ticket reservation, and the dynamic adjustment system for air ticket reservation includes:

The feature extraction module extracts the features of the air ticket data through the decision tree, and prunes the subdivided nodes to obtain N features that affect the availability and reservation as important features, and N is a preset value;

The banned entity module, based on the historical data of the previous time window, obtains banned entities based on the algorithm model;

The lock-up duration module, for each lock-up entity, the lock-up time is obtained based on the number of lock-ups in several consecutive time windows in the past;

Execute the prohibition module, when the pass rate pass of the prohibited entity is less than the prohibition threshold, it will enter the prohibition; otherwise, it will not enter the prohibition;

De-merge module, obtain the only banned entity through the algorithm of de-merger.

Embodiments of the present invention also provide a dynamic adjustment device for air ticket reservation, including:

processor;

a memory in which executable instructions for the processor are stored;

Wherein, the processor is configured to execute the steps of the dynamic adjustment method for air ticket reservation by executing the executable instructions.

Embodiments of the present invention further provide a computer-readable storage medium for storing a program, when the program is executed, the steps of the above-mentioned dynamic adjustment method for air ticket reservation are implemented.

The purpose of the present invention is to provide a dynamic adjustment method, system, equipment and storage medium for air ticket reservation, which can timely discover the ticket management defect of the airline company and make dynamic production adjustment in time, so as to improve the user's process experience of air ticket reservation.

Description of drawings

Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings.

FIG. 1 is a flow chart of the dynamic adjustment method of air ticket reservation of the present invention.

FIG. 2 is a schematic diagram of a radical model in the implementation process of the dynamic adjustment method for air ticket reservation of the present invention.

FIG. 3 is a schematic diagram of a conservative model in the implementation process of the dynamic adjustment method for air ticket reservation of the present invention.

FIG. 4 is a schematic diagram of a compromise model in the implementation process of the dynamic adjustment method for air ticket reservation of the present invention.

FIG. 5 is a schematic block diagram of the dynamic adjustment system for air ticket reservation of the present invention.

FIG. 6 is a schematic structural diagram of a dynamic adjustment device for air ticket reservation according to the present invention.

FIG. 7 is a schematic structural diagram of a computer-readable storage medium according to an embodiment of the present invention.

Detailed ways

The embodiments of the present application are described below through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the present application. The present application can also be implemented or applied to the system through other different specific embodiments, and various details in the present application can also be modified or changed according to different viewpoints and applied systems without departing from the spirit of the present application. It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other under the condition of no conflict.

The embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that those skilled in the art to which the present application pertains can easily implement. The present application can be embodied in many different forms, and is not limited to the embodiments described herein.

In the representations of this application, references to the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples", etc., are intended to be combined with the specific features represented by the embodiment or example. , structure, material or feature is included in at least one embodiment or example of the present application. Furthermore, the particular features, structures, materials or characteristics shown may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, those skilled in the art may combine and combine the different embodiments or examples presented in this application, as well as the features of the different embodiments or examples, without conflicting each other.

In addition, the terms "first" and "second" are only used for descriptive purposes, and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature delimited with "first", "second" may expressly or implicitly include at least one of that feature. In the expression of this application, "plurality" means two or more, unless expressly and specifically defined otherwise.

In order to clearly describe the present application, components irrelevant to the description are omitted, and the same or similar components are assigned the same reference numerals throughout the specification.

Throughout the specification, when a device is said to be "connected" to another device, this includes not only the case of "direct connection" but also the case of "indirect connection" with other elements interposed therebetween. In addition, when it is said that a certain device "includes" a certain constituent element, unless there is no particular description to the contrary, it does not exclude other constituent elements, but means that other constituent elements may also be included.

When a device is said to be "on" another device, this can be directly on the other device, but it can also be accompanied by other devices in between. When a device is said to be "directly on" another device in contrast, there are no other devices in between.

Although in some instances the terms first, second, etc. are used herein to refer to various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, the first interface and the second interface, etc. are represented. Also, as used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context dictates otherwise. It should be further understood that the terms "comprising" and "comprising" indicate the presence of features, steps, operations, elements, components, items, kinds, and/or groups, but do not exclude one or more other features, steps, operations, elements, The existence, appearance or addition of components, items, categories, and/or groups. The terms "or" and "and/or" as used herein are to be construed to be inclusive or to mean any one or any combination. Thus, "A, B or C" or "A, B and/or C" means "any of the following: A; B; C; A and B; A and C; B and C; A, B and C" . Exceptions to this definition arise only when combinations of elements, functions, steps, or operations are inherently mutually exclusive in some way.

The technical terms used herein are only used to refer to specific embodiments and are not intended to limit the application. The singular form used here also includes the plural form, as long as the sentence does not clearly express the opposite meaning. The meaning of "comprising" as used in the specification is to embody particular characteristics, regions, integers, steps, operations, elements and/or components, but not to exclude other characteristics, regions, integers, steps, operations, elements and/or components exist or append.

Although not defined differently, including technical and scientific terms used herein, all terms have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. Terms defined in commonly used dictionaries are additionally interpreted to have meanings consistent with the content of the relevant technical literature and current tips, and as long as they are not defined, they should not be unduly interpreted as ideal or very formulaic meanings.

At present, quotation regulation can usually be divided into tariff regulation and space regulation. The adjustment of freight rates is mainly reflected in the difference in initial pricing under different conditions, such as different periods (low season and peak season, mid-week and weekends), different refund conditions, and different prices under different baggage conditions. Therefore, the freight rate data is relatively static. At the same time, the cost of obtaining freight rate data by Trip is fixed, and the accuracy is not subject to cost constraints. As the data of the airline's Revenue Management (or Yield Management) system, the cabin is essentially dependent on the regulation of supply and demand. When a certain freight rate is booked, the airline's revenue management system will recalculate to determine whether the space is not. Continue to open; if user behavior can be regarded as a random process, then the cabin data will be randomly disturbed by the user's predetermined behavior all the time, and will change constantly. In addition, users always tend to choose low-priced products among all quotations, and low-priced quotations often change quickly due to large demand. The system has been iteratively optimized so far, whether it is the cabin or the reservation, the current user reservation interception is mainly subject to the unpredictability of external data. Our own engines cover more than 700 airlines around the world, involving most full-service airlines and some low-cost airlines. The sources of class data are various GDSs or airlines. The accuracy of the original data provided to Trip may have random problems of different degrees such as airlines, routes, and regions, which are local, occasional data problems or business problems. For example, during the epidemic, airlines cancelled a large number of flights, and the data of each airline was released to different levels, which affected the pass rate of price inspection, cabin inspection pass rate, and reservation pass rate to varying degrees. The random and long-tail characteristics of this part of the data make it necessary to build a data-driven dynamic model to intelligently handle a class of problems with predetermined interception. After a series of revenue management-related optimizations have been launched for cabin reservations, it is possible to run a data model of “feedback through pass rate” based on the big data of user booking behavior. It can be seen that what affects user interception is basically an external cause, and the long tail cannot be predicted in advance. At present, the practice of each quotation engine is relatively mechanical, or after a single interception occurs, the product is banned for a fixed number of hours according to a series of attributes; or after multiple interceptions, products with low quality are manually banned for a long time, and the ban is lifted regularly by reviewing . And the current ban, all happened after the user "has" been intercepted, and lacked predictability.

FIG. 1 is a flow chart of the dynamic adjustment method of air ticket reservation of the present invention. As shown in FIG. 1, an embodiment of the present invention provides a dynamic adjustment method for air ticket reservation, including the following steps:

S110. Perform feature extraction on the air ticket data through a decision tree, and prune subdivided nodes to obtain N features that affect bookability and reservation as important features, where N is a preset value;

S120, obtaining a prohibited entity based on the algorithm model based on the historical data of the previous time window;

S130. For each prohibited sales entity, obtain the prohibited sales duration based on the number of prohibited sales in the past several consecutive time windows;

S140. When the pass rate pass of the prohibited sales entity is less than the prohibited sales threshold, the sales prohibition is entered; otherwise, the prohibited sales are not entered;

S150, obtaining a unique prohibited entity through the algorithm of de-duplication and merging.

In a preferred embodiment, in step S110, the important features include at least one of itinerary type, departure city, departure country, cabin sales place, reservation data warehouse, and ticket issuing desk.

In a preferred embodiment, step S120 includes the following steps:

S121, the algorithm learns the historical data of the previous time window, constructs a model, and applies it to the current time window;

S122, select a time window t, each time window has the same length, and the length is r;

S123, select a stage f, an orderable stage or a reservation stage, based on the N feature dimensions as the data columns, and the data of this stage within the preset time range as the data row M to form a data matrix and train the model;

S124, loop i=0:N-1, and execute each round of level i:

个特征，基于上述数据矩阵，聚合N个特征维度，分别计算总量和通过率，生成由(N-i)个明确特征和i个虚化特征构成的禁售实体S125. In the i-th round, select i features, for

Based on the above data matrix, aggregate N feature dimensions, calculate the total amount and pass rate respectively, and generate a prohibited entity composed of (Ni) clear features and i virtual features

In a preferred embodiment, in step S122, as the length r of the time window decreases, the accuracy of the model increases;

In step S124 , it includes: i=0 round is the most fine-grained, and when the pass rate is lower than the set threshold under the characteristics of this round, a prohibition mark is added to the data object.

In a preferred embodiment, step S130 includes the following steps:

S131. Obtain the prohibited sales status of each prohibited entity in the past T consecutive time windows, and define the state st=0, which means that the sales are not prohibited; st=1, then the sales are prohibited, and the value range of t is 1 to T ;

计算历史窗口期中该实体被标记为禁售的次数；S132, define penalty factor

Calculate the number of times the entity has been marked as banned during the historical window;

表示历史窗口期中该实体被连续标记为不禁售的最长长度的比例，其中，L表示历史时间窗口从t＝1开始，st＝0可连续的最大长度；S133. Define reward factor

Represents the ratio of the longest length of the entity that is continuously marked as not banned for sale in the historical window period, where L represents the maximum length of the historical time window starting from t=1 and st=0;

S134 , obtaining the prohibition duration d=(1-pass)*r*(1+max{p-b,0}).

In a preferred embodiment, step S140 includes the following steps:

S141. Based on the intersection of the real situation and the model predicted situation in the confusion matrix, four situations of true negative, false negative, false positive and true positive are obtained. Among them, the negative corresponding model does not implement the ban, the positive corresponding model implements the ban, and the true corresponds to The prediction results of the model are consistent with the actual situation, and the prediction results of the pseudo-corresponding model are inconsistent with the actual situation;

S142. The defined improvement rate is the proportion that is actually intercepted and banned from sales, namely TN/All,

S143. Define the false ban rate as the proportion that actually passed but entered the ban, namely FN/All,

S144 , traverse all prohibition thresholds according to the preset step interval, and calculate and obtain a curve of improvement rate and false prohibition rate under the threshold.

S145: Calculate the marginal utility of the improvement and the marginal cost of the false ban, find a ban threshold with a large improvement rate and a small false ban rate, and consider it to be the most valuable ban threshold.

S146. Use the value as the most fine-grained ban threshold, and the ban thresholds of other levels are attenuated by the value/N in a uniform step.

In a preferred embodiment, in step S150, when the high-level prohibition message includes a low-level prohibition message, the low-level prohibition message is removed, and only the high-level prohibition message is retained.

The present invention can proactively intervene in advance when the predetermined quality is known to be unsatisfactory through the constructed prohibition system. In addition, the statically matched prohibited-sale entity is the same as the fixed-period prohibited-sale duration, which also needs to be optimized in a dynamic way, hoping that the model has the ability to “smart” abstraction. The model system obtained by machine learning can prohibit the sale of a series of entities with common problems in advance, so that user interception will no longer occur, and the quality of reservations will be directly improved. More importantly, after the quotation engine consumes the model, it can output the next-lowest quotation with better quality and better conversion for users to choose, instead of simply doing subtraction on the result set.

In recent years, with the advancement of technology, business expansion and globalization strategy, the quotation resources of Trip international air tickets have been gradually enriched, the route coverage has been expanded, and the competitive advantage has been continuously expanded. At the same time, the smoothness of users at each predetermined stage, as part of system reliability, is also critical and needs to be improved and optimized. This dynamic model aims to create a highly smooth booking system that, on the one hand, helps to drive conversions and increase revenue, and on the other hand aligns with Trip's commitment to high-quality service.

The specific implementation process of the present invention is as follows:

(1) Data preparation.

Use the user bookable data on the fill-in page and the user booking data on the value-added page.

(2) Feature selection.

Use CART decision tree for feature extraction, prune over-subdivided nodes, and obtain N important features that affect orderability and reservation. Further, these features extracted based on the classification tree method also need to be deeply consistent with the understanding of the international air ticket business. For example, itinerary type (Trip Type), departure city (Origin City), departure country (OriginCountry) and the POC revenue management method adopted by European airlines are closely related; , Booking GDS (Booking GDS), and Ticketing Agency (Ticketing Agency) are highly relevant. Information on city pairs and country pairs helps capture randomly distributed routing problems.

(3) Algorithm design

(3.1) Prohibited entities

Prohibited entities refer to a class of data objects that are prohibited from being sold, and the features extracted in 2.2 are used as entity attributes.

The algorithm learns the historical data of the previous time window, constructs a model, and applies it to the current time window.

Select the time window time t, each time window has the same length, and the length is r. The model can be refined by continuously shrinking r, which is equivalent to micro-differentiation. The value of r needs to consider the trade-off between thinning and overfitting.

Select a stage phase f, which can be ordered or reserved. Based on N feature dimensions as data columns, the data of this stage within a certain time range is used as data row M to form a data matrix and train the model. The key indicators used here are the total amount (counts) and the pass rate (pass).

Loop i=0:N-1, execute each round of level i:

个特征，基于上述数据矩阵，聚合N个特征维度，分别计算总量和通过率，生成由(N-i)个明确特征和i个虚化特征构成的禁售实体。其中i＝0轮是最细粒度，没有任何虚化特征的实体。如果该轮特征下，通过率低于设定阈值，作禁售标记。In round i, select i features, for

Based on the above data matrix, aggregate N feature dimensions, calculate the total amount and pass rate respectively, and generate a prohibited entity composed of (Ni) clear features and i virtual features. where i=0 round is the most fine-grained entity without any blurring features. If the pass rate is lower than the set threshold under the characteristics of this round, it will be marked as banned.

(3.2) Lock-up period

Linear in the pass rate formula to calculate the current lock-up duration. Based on the data performance of multiple time windows in history, a reward and punishment mechanism is set for the duration of the ban. For products that have performed well recently in history, when the quality of the current product is slightly worse, the duration of the ban is set to be relatively short (rewards); the long-term performance in history is very poor For products, the ban time is set more severely (punishment), and the reward and punishment mechanism makes the ban time more reasonable.

For a prohibited entity,

The historical time window is composed of a series of consecutive time windows in the past, from near to far, t from 1 to T.

Define the state State s _t = 0, then the sale is not banned;

s _t = 1, the sale is prohibited.

计算历史窗口期中该实体被标记为禁售的次数；表示重复拦截惩罚力度。Define penalty factor

Calculate the number of times the entity is marked as banned for sale during the historical window period; it indicates the punishment for repeated interception.

表示历史窗口期中该实体被连续标记为不禁售的最长长度的比例。其中L表示历史时间窗口从t＝1开始，s_t＝0可连续的最大长度，用于衡量“连续”质量较优的次数。连续较优次数与历史窗口大小的占比，作为奖励力度。Define the reward factor

Indicates the proportion of the longest length in the historical window during which the entity has been continuously marked as unbanned. Wherein L represents the maximum continuous length of the historical time window starting from t=1 and s _t =0, which is used to measure the number of times that the “continuous” quality is better. The ratio of the number of consecutive optimizations to the size of the historical window is used as a reward.

Calculate the lock-up period

Duration d=(1-pass)*r*(1+max{p-b,0})

(3.3) Banning Threshold

Given a prohibition threshold, if the pass rate of the prohibited entity is less than the prohibition threshold, it will enter the prohibition; otherwise, it will not enter the prohibition. Based on the idea of confusion matrix (Confusion Matrix), the real situation and the model prediction situation are crossed to obtain true negative (TN, True Negative), false negative (FN, False Negative), false positive (FP, FalsePositive) and true positive (TP, True Positive) 4 possibilities. Yin and Yang correspond to the model not implementing a ban on sales and implementing a ban on sales, respectively. True and false correspond to the consistency or inconsistency between the model prediction results and the actual situation. The improvement rate is defined as the proportion that is actually intercepted and entered into the ban, that is, TN/All, which is reflected in the increase in the production pass rate after the ban model takes effect. The false ban rate is defined as the proportion of those who actually passed the ban but entered the ban, namely FN/All, which can be regarded as the opportunity cost brought by the ban. According to a certain step interval gap g, traverse all possible ban thresholds, and calculate the curve of improvement rate and false ban rate under the threshold. Calculate the marginal utility of the improvement and the marginal cost of false ban, find the ban threshold with a large improvement rate and a small false ban rate, and consider it as the most valuable ban threshold. This value is used as the most fine-grained (Level 0) ban threshold, and the ban thresholds of other levels are attenuated by this value/N in a uniform step.

(3.4) De-recombination

In the actual implementation, the operation of level0 is based on the original data, and the data of level1...levelN should theoretically be based on the data of Level N-1. In order to simplify the calculation, they are all based on Level0. This allows for a containment relationship between all prohibited entities, which is redundant. The same embargoed record may be contributed to different embargoed entities. An algorithm for de-merging is required, showing a clear unique banned entity. When the high-level banned news includes low-level banned news, only the high-level banned news is retained.

Four deduplication and merging algorithms are designed. The essence is to find a path to the level of a banned entity in the constructed multi-fork tree.

The first radical model (Figure 2), which allows arbitrary levels of cross-level mergers, is relatively free. The nodes in the figure represent the banned entities and their levels, and the connection lines indicate that there is an inclusion relationship between the two entities, and the high-level entity contains the low-level entity. The second conservative model (Figure 3) does not allow any leapfrog merging, that is, all reserved prohibited entities meet the level of the entity where the entity is located, and there are corresponding prohibited entities, that is, the depth of the subtree Must be equal to level+1. The dotted line nodes in the figure represent banned entities that failed to generate compared to the aggressive model. The dotted line node does not actually exist, it is only used to illustrate that the reason why L3 cannot be generated is the lack of support for L1 and L2.

The third compromise model (Fig. 4) does not allow arbitrary cross-level merging, and only allows downward merging when there are adjacent entities. Unlike the conservative model, the trade-off model retains prohibited entities as long as there is one path instead of all paths, so that the tree depth is less than or equal to Level+1. The lower the level of the banned entity, the finer and scattered the granularity; the higher the level, the greater the improvement of the banned sale, and the merger needs to be extremely cautious. In contrast, the eclectic model combines aggressive and conservative models with the best balance.

However, in some scenarios, the compromise model still inevitably generates low-level banned entities, which weakens the generalization ability of the model; and then optimizes a composite version on the compromise model. With n chosen, a compromise model is applied between levels n+1...N; an aggressive model is applied between levels 0-n. Iteratively optimize n to find the optimal composite model.

(4) Model evaluation

In addition to using the improvement rate and false rejection rate defined in (3.3) to evaluate the business effect, the model is also evaluated using the classic and general accuracy rate (Precision), recall rate (Recall), and accuracy (Accuracy) of machine learning.

(5) Model application

The model is connected to the message management center as a producer. When a new banned message is generated, it will call the write interface provided by the message management center to write the latest message; the air ticket international engine acts as a consumer to monitor the message management center in real time. The message will be consumed immediately after the message, and the suboptimal solution will be output.

The effect after the actual use of the scheme of the present invention:

50 weeks of bookable (BK) and reservation (RV) data for 20191220-20201210 production, covering pre-pandemic (before 20200201) and post-pandemic.

Iterate to get the optimal model and parameter combination.

Feature selection: Trip Type, POS,

Shopping Engine, Booking GDS,

Validating Airline, Ticketing Agency,

Ori-Dest Country Pair, Ori-Dest City Pair;

Model selection: composite model, n=3;

Parameter optimization: Time t=8h[05,13,21];

Threshold BK=35%, Threshold RV=20%, the step interval is uniform and linear.

Through the construction of the present invention, a dynamic model in which both the banned entity and the banned duration are generated by data is obtained. Before the epidemic (before 20200201), the pass rate of system-orderable BK was maintained at 92%, the model improvement rate was 1.6%, and the false ban rate was 0.3%; before the epidemic (202002-202008), the pass rate was maintained at 86-89%, and the model improvement rate 3.5%-6%, the false ban rate is 0.5%; in the late stage of the epidemic, the pass rate is 90-92%, the improvement rate is 2-2.5%, and the false ban rate is 0.3-0.4%. Reservation RV pass rates performed similarly. The model can pull down or sag the hole pass rate curve, so that the orderable and reserved pass rate can be stabilized at around 92%, achieving "controllable".

It can be further found that for different original pass rates, the degree of improvement of the model is different. If the original pass rate is poor, the improvement will be larger. For example, when the international epidemic broke out in the middle and late March, 70% of international flights were cancelled, and the original BK was 75%. The algorithm can maintain stability to around 89%. If the original pass rate is better, the improvement is small, such as January, the original BK is 92-93%, and the model is only improved to around 93.5%. In the face of unpredictable and significant reductions, the model can improve the system to a reliable level, and when the external data environment performs better, the model is maintained at a stable level and the performance is in line with expectations.

In addition, models can help identify clinically meaningful production problems, both contributing to a timely production response. It can help to discover the revenue management logic behind the phenomenon of airlines' "can check but not book" and "virtual cabin", or the third-party data problem caused by the inconsistency between GDS class and airline class.

FIG. 5 is a schematic block diagram of the dynamic adjustment system for air ticket reservation of the present invention. As shown in 5, the dynamic adjustment system 5 of air ticket reservation of the present invention includes:

The feature extraction module 51 performs feature extraction of the air ticket data through a decision tree, and prunes the subdivided nodes to obtain N features that affect the availability and reservation as important features, where N is a preset value;

The prohibited-sale entity module 52, based on the historical data of the previous time window, obtains the prohibited-sale entity based on the algorithm model;

The lock-up duration module 53, for each lock-up entity, obtains the lock-up duration based on the number of lock-up times in the past several consecutive time windows;

Execute the prohibition module 54, when the pass rate pass of the prohibition entity is less than the prohibition threshold, the prohibition is entered; otherwise, the prohibition is not entered;

The de-recombination module 55 obtains the only banned entity through the de-recombination algorithm.

The dynamic adjustment system for air ticket reservation of the present invention can timely find out the ticket management defect of the airline company and make dynamic production adjustment in time, so as to improve the user's process experience of air ticket reservation.

The above-mentioned embodiments are only preferred examples of the present invention, and are not intended to limit the present invention. Any equivalent substitutions, modifications and changes made within the principles of the present invention are all within the protection scope of the present invention.

The embodiment of the present invention also provides a dynamic adjustment device for air ticket reservation, which includes a processor. A memory in which executable instructions for the processor are stored. Wherein the processor is configured to execute the steps of the method for dynamically adjusting airline reservations via executing the executable instructions.

As shown above, the dynamic adjustment system for air ticket reservation of the present invention in this embodiment can timely discover the ticket management defects of the airline company and make dynamic production adjustment in time, so as to improve the user's process experience of air ticket reservation.

As will be appreciated by one skilled in the art, various aspects of the present invention may be implemented as a system, method or program product. Therefore, various aspects of the present invention can be embodied in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, which may be collectively referred to herein as implementations "Circuit", "Module" or "Platform".

FIG. 6 is a schematic structural diagram of a dynamic adjustment device for air ticket reservation according to the present invention. An electronic device 600 according to this embodiment of the present invention is described below with reference to FIG. 6 . The electronic device 600 shown in FIG. 6 is only an example, and should not impose any limitation on the function and scope of use of the embodiments of the present invention.

As shown in FIG. 6, electronic device 600 takes the form of a general-purpose computing device. Components of the electronic device 600 may include, but are not limited to, at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, and the like.

The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 executes the steps according to various exemplary embodiments of the present invention described in the foregoing electronic prescription flow processing method section of this specification. For example, the processing unit 610 may perform the steps shown in FIG. 1 .

The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 6201 and/or a cache storage unit 6202 , and may further include a read only storage unit (ROM) 6203 .

The storage unit 620 may also include a program/utility 6204 having a set (at least one) of program modules 6205 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, An implementation of a network environment may be included in each or some combination of these examples.

The bus 630 may be representative of one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local area using any of a variety of bus structures bus.

The electronic device 600 may also communicate with one or more external devices 700 (eg, keyboards, pointing devices, Bluetooth devices, etc.), with one or more devices that enable a user to interact with the electronic device 600, and/or with Any device (eg, router, modem, etc.) that enables the electronic device 600 to communicate with one or more other computing devices. Such communication may occur through input/output (I/O) interface 650 . Also, the electronic device 600 may communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network such as the Internet) through a network adapter 660 . Network adapter 660 may communicate with other modules of electronic device 600 through bus 630 . It should be appreciated that, although not shown, other hardware and/or software modules may be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives And data backup storage platform, etc.

Embodiments of the present invention also provide a computer-readable storage medium for storing a program, and the steps of the method for dynamically adjusting air ticket reservations are implemented when the program is executed. In some possible implementations, various aspects of the present invention can also be implemented in the form of a program product, which includes program code, when the program product runs on a terminal device, the program code is used to cause the terminal device to execute the above-mentioned description in this specification. The steps according to various exemplary embodiments of the present invention are described in the section of the electronic prescription flow processing method.

As shown above, the dynamic adjustment system for air ticket reservation of the present invention in this embodiment can timely discover the ticket management defects of the airline company and make dynamic production adjustment in time, so as to improve the user's process experience of air ticket reservation.

FIG. 7 is a schematic structural diagram of a computer-readable storage medium of the present invention. 7, a program product 800 for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read only memory (CD-ROM) and include program codes, and can be stored in a terminal device, For example running on a personal computer. However, the program product of the present invention is not limited thereto, and in this document, a readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples (non-exhaustive list) of readable storage media include: electrical connections with one or more wires, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

A computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave with readable program code embodied thereon. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A readable storage medium can also be any readable medium other than a readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages—such as Java, C++, etc., as well as conventional procedural programming Language - such as the "C" language or similar programming language. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server execute on. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (eg, using an Internet service provider business via an Internet connection).

To sum up, the purpose of the present invention is to provide a dynamic adjustment method, system, device and storage medium for air ticket reservation, which can timely discover the ticket management defects of the airline company and make dynamic production adjustment in time, so as to improve the user's process experience of air ticket reservation.

The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be considered that the specific implementation of the present invention is limited to these descriptions. For those of ordinary skill in the technical field of the present invention, without departing from the concept of the present invention, some simple deductions or substitutions can be made, which should be regarded as belonging to the protection scope of the present invention.

Claims (10)

1. A method for dynamically adjusting airline reservations, comprising the steps of:

s110, extracting the characteristics of the air ticket data through a decision tree, and pruning subdivided nodes to obtain N characteristics which influence the orderable positions as important characteristics, wherein N is a preset value;

s120, obtaining a sale prohibition entity based on the historical data of the previous time window based on an algorithm model;

s130, obtaining the sale prohibition duration for each sale prohibition entity based on the sale prohibition times in a plurality of past continuous time windows;

s140, entering into forbidden sale when the passing rate pass of the forbidden sale entity is smaller than a forbidden sale threshold value; otherwise, no sale prohibition is entered;

and S150, obtaining a unique selling forbidding entity through a de-duplication merging algorithm.

2. The method for dynamically adjusting airline reservation according to claim 1, wherein in the step S110, the important feature includes at least one of a type of travel, a city of departure, a country of departure, a sales space of a cabin, a reservation data warehouse, and a ticket desk.

3. The method for dynamically adjusting airline reservation according to claim 1, wherein the step S120 comprises the steps of:

s121, learning historical data of a previous time window by an algorithm, constructing a model, and applying the model to a current time window;

s122, selecting time windows t, wherein each time window is equal in length, and the length is r;

s123, selecting a stage f, namely, defining a stage or a booking stage, forming a data matrix and training a model based on N characteristic dimensions as data columns and data of the stage in a preset time range as data rows M;

and S124, executing each round of level i:

s125, in the ith round, selecting i characteristics

And the characteristics are aggregated by N characteristic dimensions based on the data matrix, the total amount and the passing rate are respectively calculated, and the sale forbidding entity consisting of (N-i) clear characteristics and i virtualized characteristics is generated.

4. The method for dynamically adjusting airline reservations according to claim 3, characterized in that, in step S122, as the length r of the time window decreases, the accuracy of the model increases;

the step S124 includes: and adding an sale prohibition mark to the data object when the passing rate is lower than a set threshold value under the characteristic that the i is 0 round.

5. The method for dynamically adjusting airline reservation according to claim 1, wherein the step S130 comprises the steps of:

s131, obtaining the sale prohibition status of each sale prohibition entity in the past T consecutive time windows, and if the defined status st is 0, then not prohibiting sale; st is 1, sale is forbidden, and the value range of T is 1 to T;

s132, defining penalty factor

Calculating the number of times that the entity is marked as forbidden in the historical window period;

s133, defining reward factors

Indicating the proportion of the longest length of the historical window in which the entity is continuously marked as not-for-sale, wherein L indicates the maximum length of the historical time window which can be continuously marked from t-1 to st-0;

and S134, obtaining the sale prohibition time d ═ 1-pass × (1+ max { p-b,0 }).

6. The method for dynamically adjusting airline reservation according to claim 4, wherein the step S140 comprises the steps of:

s141, based on the fact that the model prediction situation intersects with the true situation in the confusion matrix, 4 situations of true negative, false positive and true positive are obtained, wherein the negative corresponding model does not execute sale forbidding, the positive corresponding model executes sale forbidding, the true corresponding model prediction result is consistent with the actual situation, and the false corresponding model prediction result is inconsistent with the actual situation;

s142, defining the improvement rate as the actual intercepted ratio entering the forbidden sale, namely TN/All;

s143, defining the false prohibition rate as the actual passing ratio but entering the forbidden sale, namely FN/All;

s144, traversing all sale prohibition thresholds according to a preset step diameter interval, and calculating to obtain a curve of improvement rate and prohibition error rate under the threshold;

s145, calculating improved marginal utility and forbidden marginal cost, and finding out a forbidden sale threshold with high improvement rate and low forbidden rate, wherein the forbidden sale threshold is considered as the most valuable forbidden sale threshold;

and S146, taking the sale prohibition threshold with large improvement rate and small forbidding rate as the finest granularity sale prohibition threshold, and attenuating the sale prohibition thresholds of the rest levels according to the sale prohibition threshold with large improvement rate and small forbidding rate/N uniform step diameter.

7. The method for dynamically adjusting airline reservation according to claim 1, wherein in step S150, when the high-Level sale prohibition message includes a low-Level sale prohibition message, the low-Level sale prohibition message is removed, and only the high-Level sale prohibition message is retained.

8. A system for dynamically adjusting airline reservations, for implementing the method for dynamically adjusting airline reservations of claim 1, comprising:

the characteristic extraction module is used for extracting the characteristics of the air ticket data through the decision tree, and cutting subdivided nodes to obtain N characteristics which influence the orderable position as important characteristics, wherein N is a preset value;

the forbidden entity module is used for obtaining a forbidden entity based on historical data of the last time window and an algorithm model;

the sale prohibition duration module is used for obtaining sale prohibition duration for each sale prohibition entity based on the sale prohibition times in a plurality of past continuous time windows;

executing a sale forbidding module, and entering sale forbidding when the pass rate pass of the sale forbidding entity is less than a sale forbidding threshold; otherwise, no sale prohibition is entered;

and the de-duplication merging module obtains a unique selling prohibition entity through a de-duplication merging algorithm.

9. A device for dynamically adjusting airline reservations, comprising:

a processor;

a memory having stored therein executable instructions of the processor;

wherein the processor is configured to perform the steps of the method of dynamic adjustment of airline reservations of any of claims 1-7 via execution of the executable instructions.

10. A computer-readable storage medium storing a program which, when executed by a processor, performs the steps of the method for dynamic adjustment of airline reservations of any one of claims 1 to 7.

Images