Data distribution method and system, program product, equipment and storage medium

Abstract

The invention provides a data distribution method and system, a program product, equipment and a storage medium, which are applied to online air ticket booking platform product testing based on flight information units, and the method comprises the steps: obtaining a plurality of flight information units, and determining the flight information units based on air routes and air route take-off days; based on a preset grouping rule, dividing the plurality of flight information units into at least two groups so as to carry out a comparison test; and performing different product optimization operations on the flight information units of different groups according to a preset product test strategy. According to the method, the problems of overlong experimental period and low experimental efficiency in the prior art are effectively solved, the experimental efficiency is improved, the evaluation period is shortened, strategy evaluation of various marketing scenes is supported on the premise of ensuring the consistency of user experience, A / B testing can be effectively carried out by grouping ODT, and the experimental efficiency is improved. Therefore, the product effect can be evaluated more quickly.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data diversion method, system, program product, device and storage medium. Background Art

[0002] In the Internet industry, AB experiments are widely used as a common method to evaluate the effectiveness of commercial strategy delivery. Usually, strategy makers will divide the users in the experimental scenario into experimental and control groups according to a certain ratio, implement different marketing strategies or display different page layouts for the experimental group users, and then evaluate the performance of the experimental group relative to the control group in terms of business indicators after the experimental cycle to determine whether the new business strategy should be fully promoted.

[0003] However, in product testing of online air ticket booking platforms, such as ticket price adjustment, iteration of auxiliary marketing products, etc., traditional AB experimental methods are often difficult to apply directly, considering the fairness of user consumption. This is because online air ticket booking platforms must ensure that when users browse flights on the same route, the product types and prices they see remain consistent, and the traditional AB experimental model may result in different users on the same route seeing different prices, which is contrary to the above principle of fairness.

[0004] At present, the online air ticket booking platform has proposed an alternative solution to the alternate-day experiment to address the above problems. This solution uses the entire experimental scenario as the experimental group and the control group in rotation on a daily basis. That is, the new product optimization plan is fully launched on the experimental day, and the original plan is maintained on the control day, and then the effect of the plan is evaluated after aligning the dates. However, the evaluation cycle of the alternate-day experimental plan is doubled compared to the traditional AB experiment, and only one plan can be evaluated in the same period, resulting in low overall experimental efficiency. Therefore, how to shorten the experimental cycle and improve experimental efficiency while ensuring consistency in user experience is a technical problem that needs to be solved in the product testing field of online air ticket booking platforms.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] In view of the problems in the prior art, the present invention provides a data shunting method, which is applied to the product testing of an online flight reservation platform, so as to solve the problem in the prior art that it is impossible to efficiently evaluate the effect of a product optimization strategy on the premise of ensuring the consistency of product types and selling prices when users browse flights on the same route. By grouping flight information units determined based on the route and the departure date of the route, and performing different product optimization operations on different groups, the effective evaluation of the product optimization strategy can be realized while avoiding confusing users.

[0007] An embodiment of the present invention provides a data shunting method, which is applied to the product testing of an online flight reservation platform based on flight information units, and includes the following steps:

[0008] Obtain a plurality of flight information units, where the flight information units are determined based on the route and the departure date of the route;

[0009] Based on a preset grouping rule, divide the plurality of flight information units into at least two groups for comparative testing;

[0010] According to a preset product testing strategy, perform different product optimization operations on the flight information units of different groups.

[0011] In some optional embodiments, the flight information unit includes at least one of the basic attributes of the air ticket and the business scenario requirements.

[0012] In some optional embodiments, the basic attributes of the air ticket include the route, the departure date, the cabin class, and the flight type, and the business scenario requirements include the user's historical behavior, the user's preference settings, and the marketing activity information.

[0013] In some optional embodiments, after obtaining a plurality of flight information units, where the flight information units are determined based on the route and the departure date of the route, the following steps are further included:

[0014] For each flight information unit, extract at least one feature, where the features include: the historical average occupancy rate, the historical average ticket price, and the number of days of advance booking;

[0015] Based on the features, use a clustering algorithm to pre-cluster the flight information units, and divide the similar flight information units into the same cluster;

[0016] Assign each cluster as a whole to different groups.

[0017] In some optional embodiments, the preset grouping rule uses a greedy algorithm to group the plurality of flight information units, and includes the following steps:

[0018] Perform an initial grouping on the plurality of flight information units, and the initial grouping is implemented by means of random assignment or assignment based on the clustering result;

[0019] Based on the initial grouping, iteratively exchange the flight information units between groups to minimize the differences between groups.

[0020] In some alternative embodiments, in the step of iteratively exchanging the flight information units between groups, allow the shunt difference value of at least one metric to increase within a preset range, but at the same time limit the decrease of the shunt difference values of the remaining metrics.

[0021] An embodiment of the present invention further provides a data shunting method for product comparison testing of an online air ticket reservation platform, including the following steps:

[0022] Adopt the data shunting method as described above to divide multiple flight information units into at least two groups;

[0023] For different groups, adopt different product display strategies;

[0024] Collect and analyze the user behavior data of each group to evaluate the effects of different product display strategies;

[0025] According to the evaluation results, adjust the product display strategy of the online air ticket reservation platform.

[0026] An embodiment of the present invention further provides a data shunting system applied to an online business travel reservation platform for implementing the above data shunting method. The system includes:

[0027] An acquisition module for acquiring multiple flight information units, where the flight information units are determined based on the route and the route departure date;

[0028] A grouping module for dividing multiple flight information units into at least two groups based on preset grouping rules for comparative testing;

[0029] An optimization module for performing different product optimization operations on the flight information units of different groups according to preset product testing strategies.

[0030] An embodiment of the present invention further provides a data shunting program product, where the program product includes computer instructions, and when the computer instructions are executed by a processor, the steps of the above method are implemented.

[0031] An embodiment of the present invention further provides a data shunting device, including:

[0032] A processor;

[0033] A memory storing executable instructions of the processor;

[0034] Wherein, the processor is configured to execute the steps of the above data shunting method by executing the executable instructions.

[0035] An embodiment of the present invention further provides a computer-readable storage medium for storing a program, which implements the steps of the above-mentioned data diversion method when executed by a processor.

[0036] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0037] The data diversion method, system, program product, device and storage medium of the present invention have the following beneficial effects: the data diversion method provided by the present invention can achieve effective evaluation of product optimization effects in product testing of online air ticket booking platforms by grouping flight information units determined based on routes and route departure dates, and implementing different product optimization operations on flight information units of different groups. Especially in scenarios where it is necessary to ensure the consistency of product types and prices seen by users when browsing flights on the same route, it avoids user confusion that may be caused by traditional AB experiments, and there is no need to rotate strategy delivery on a daily basis as in alternate-day experiments, thereby shortening the experimental cycle, improving test efficiency, and helping to quickly verify the effectiveness of product optimization strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Other features, objectives and advantages of the present invention will become more apparent from a reading of the detailed description of non-limiting embodiments made with reference to the following accompanying drawings.

[0039] Figure 1 is a flow chart of a data splitting method according to an embodiment of the present invention;

[0040] Figure 2 It is a schematic diagram of the structure of a data distribution system according to an embodiment of the present invention;

[0041] Figure 3 It is a structural schematic diagram of a data splitting device according to an embodiment of the present application;

[0042] Figure 4 It is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of the present application. DETAILED DESCRIPTION

[0043] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0044] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all steps. For example, some steps can be further decomposed, while some steps can be combined or partially combined. Therefore, the actual execution order may change according to the actual situation. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0045] The present invention relates to the field of computer data processing and Internet technologies, especially the data diversion technology for online reservation platforms. In the process of strategy iteration and product optimization of Internet platforms, the AB test is a commonly used method to evaluate the effectiveness of solutions. Its purpose is to divide users or traffic into an experimental group and a control group, and evaluate the optimization effect by comparing the key indicators of the two groups. However, in scenarios with high requirements for user fairness such as air ticket reservation, directly randomly diverting users may result in inconsistent prices or products seen by different users on the same route, affecting the user experience. Therefore, it is necessary to find a diversion solution that can both ensure the consistency of the user experience and effectively isolate the experimental group and the control group. The present invention takes the route and the origin-destination date (ODT) as the minimum diversion unit, ensuring the consistency of the user experience under the same ODT. At the same time, by optimizing the diversion algorithm, it tries to minimize the differences in various indicators of each group before the experiment, thereby ensuring the reliability of the experimental results. This diversion solution can effectively shorten the experimental cycle, improve the experimental efficiency, and avoid user experience problems caused by diversion, providing an efficient and reliable solution for the product optimization and strategy iteration of the online air ticket reservation platform.

[0046] As Figure 1 shown, the invention provides a data diversion method, which is applied to the product test of an online air ticket reservation platform based on flight information units, and includes the following steps;

[0047] S100. Obtain multiple flight information units, which are determined based on the flight route and the departure date of the flight route. The flight information unit (ODT) is the smallest unit for data diversion in the present invention. It takes the flight route plus the departure date as the smallest diversion unit. To implement product testing, it is necessary to obtain a certain number of flight information units, that is, no less than two flight information units. Being determined based on the flight route and the departure date means that the flights taking off on a specific day on a specific flight route form an independent unit. In the air ticket reservation scenario, flights on the same flight route and departure date usually have similar user groups and market demands. Diversifying them as a whole can avoid confusion caused by users seeing different products or prices on the same flight route and date, and ensure the consistency of the user experience. In the embodiments of the present invention, the flight route refers to the flight route determined by the origin airport and the destination airport. For example, Beijing - Shanghai, Chengdu - Guangzhou, etc. More specifically, the flight route can be further refined to the specific flight number. The departure date refers to the actual departure date of the flight. The flight information unit can be obtained by querying from the database of the online air ticket reservation platform, purchasing from a third - party data provider, or scraping from other channels through web crawler technology, etc. The obtained flight information unit needs to contain sufficient information for subsequent grouping and testing. The number of flight information units depends on specific testing requirements and data volume. Usually, it is necessary to obtain a sufficient number of flight information units to ensure that the data volume of each group after grouping is large enough to obtain reliable test results. The flight information unit can also be determined based on factors such as the flight route, departure time, and cabin class. For example, the economy - class flights on the same flight route and taking off on the same day are regarded as one unit, while the business - class flights are regarded as another unit. This can be adjusted according to specific business scenarios and testing requirements.

[0048] By defining the flight information unit as the smallest diversion unit, it ensures that the product information seen by users on the same flight route and date is consistent, avoids unnecessary confusion, and lays a foundation for subsequent product testing and strategy evaluation. At the same time, it provides a basic data set for subsequent diversion strategies to facilitate strategy optimization comparison on an equal basis.

[0049] S200. Based on a preset grouping rule, divide multiple flight information units into at least two groups for comparative testing. In the embodiments of the present invention, the preset grouping rule refers to the criteria and methods for dividing flight information units into different groups. The purpose of grouping is to conduct comparative testing, so it is necessary to ensure comparability between groups. By dividing flight information units into different groups and then applying different product optimization operations to different groups, the effects of different strategies can be compared. The grouping rule affects the accuracy and reliability of the experimental results. The preset grouping rule can be designed based on various factors, including random assignment, assignment based on clustering algorithms, assignment based on business rules, etc. For example, the random assignment method can be used to randomly assign flight information units to different groups; the clustering algorithm can also be used to divide similar flight information units into the same group; or according to business rules, certain specific flight information units can be divided into specific groups. The number of groups can be adjusted according to specific test requirements. Usually, at least two groups are required: one as the experimental group to apply the new product optimization strategy; the other as the control group without applying any new strategy. Multiple experimental groups can also be set up to apply different product optimization strategies respectively. The ratio between groups can also be adjusted according to specific test requirements. For example, the ratio of the experimental group to the control group can be set to 1:1, or adjusted according to specific business scenarios and test requirements. To ensure comparability between groups, it is usually necessary to perform statistical analysis on the data of each group to ensure that the key indicators of each group have a similar distribution. For example, it is necessary to ensure that indicators such as user portraits, historical order volumes, and seat occupancy rates of each group have a similar distribution. In addition, the rule can also adopt the A / A testing method, that is, randomly divide the flight information units into two groups, and both groups adopt the same strategy to verify the randomness and consistency of the traffic splitting.

[0050] Traffic splitting through the preset grouping rule provides a basis for subsequent comparative testing. Different groups adopt different product optimization operations, which can effectively compare the effects of different strategies, thus supporting product optimization decisions. Moreover, through the flexible design of the grouping rule, different test requirements can be met, ensuring the accuracy and reliability of the experimental results.

[0051] S300. According to the preset product testing strategy, different product optimization operations are performed on different groups of flight information units. In the embodiments of the present invention, the product optimization operation refers to various improvement measures taken for the online air ticket reservation platform, aiming to enhance the user experience, increase the conversion rate, or increase revenue. The preset product testing strategy determines how to apply these optimization operations to different groups for comparative testing. By applying different product optimization operations to different groups of flight information units, the impacts of different strategies on user behavior and business metrics can be observed, thereby evaluating the effectiveness of the strategies and selecting the optimal solution. The product optimization operation may include adjusting the air ticket price, changing the page display layout, recommending different auxiliary products, providing personalized search results, etc. The product testing strategy may include A / B testing, multivariate testing, gray release, etc. For example, a new price strategy can be applied to the flight information units in the experimental group, while the original price strategy is maintained for the flight information units in the control group; or a new page layout can be displayed for the flight information units in the experimental group, while the original page layout is displayed for the flight information units in the control group. When performing product optimization operations, it is necessary to pay attention to controlling variables to avoid the influence of other factors on the experimental results. For example, it is necessary to ensure that the user groups in each group have similar characteristics to avoid deviation of the experimental results caused by user differences. The product optimization operation can also be a personalized recommendation strategy for specific user groups. For example, according to the historical behavior and preferences of users, different air ticket products or auxiliary products are recommended for different groups.

[0052] By means of the preset product testing strategy, performing different product optimization operations on different groups of flight information units can effectively evaluate the effects of different strategies and provide data support for product optimization. By comparing the user behavior data and business metrics under different strategies, the optimal solution can be found, thereby enhancing the user experience, increasing the conversion rate, or increasing revenue.

[0053] In some embodiments, the flight information unit includes at least one of the basic ticket attributes and business scenario requirements. When performing data shunting in the embodiments of the present invention, not only the inherent attributes of the ticket itself are considered, but also the requirements of users in specific business scenarios can be taken into account. By comprehensively considering the basic ticket attributes and business scenario requirements, the embodiments of the present invention can achieve more refined data shunting, thereby providing a more accurate and reliable data basis for the product comparison test of the online ticket booking platform. This helps the platform better understand user needs, optimize product strategies, and improve user experience and operation efficiency. For example, by analyzing the user behavior data of different groups, it can be found that different user groups have preferences for different product display strategies, thereby providing a basis for personalized recommendations; or the impact of different marketing activities on user purchase behavior can be evaluated, thereby optimizing marketing strategies. The basic ticket attributes refer to the attributes directly related to the ticket product, such as route, departure date, cabin class, and flight type, etc. These attributes are the core elements constituting a ticket and directly affect user choices. The business scenario requirements refer to the specific situations when users purchase tickets, such as user historical behavior, user preference settings, and marketing activity information, etc. These factors reflect the personalized needs and purchase intentions of users. The route refers to the specific departure and destination, such as from Beijing to Shanghai. The departure date refers to the actual departure date of the flight, specific to year, month, and day. The cabin class refers to the seat class of the ticket, such as economy class, business class, first class, etc. The flight type can be divided into direct flights, connecting flights, etc. The business scenario requirements can include user historical behavior, user preference settings, and marketing activity information. User historical behavior refers to the browsing, searching, purchasing, etc. historical records of users on the platform, such as the routes and cabin classes that users often purchase. User preference settings refer to the personal preferences set by users on the platform, such as preferred airlines, seat positions, etc. Marketing activity information refers to various marketing activities being carried out on the platform, such as promotional activities, coupons, etc. The comprehensive use of this information can more accurately group the flight information units, thereby improving the effect of product testing.

[0054] By comprehensively considering the basic ticket attributes and business scenario requirements, the characteristics of the flight information unit can be more comprehensively described, thereby providing richer information for subsequent grouping operations. For example, users who often purchase high-price tickets in their historical behavior can be grouped into the same group for the test of high-end products; users who are sensitive to price can also be grouped into the same group for the test of promotional activities. This refined grouping method can more effectively evaluate the effects of different product strategies, thereby providing a more reliable basis for the product optimization of the online ticket booking platform.

[0055] In some embodiments, after obtaining multiple flight information units determined based on the route and the departure date of the route, the following steps are further included:

[0056] S110. For each flight information unit, extract at least one feature, including: historical average occupancy rate, historical average ticket price, and advance booking days. In an embodiment of the present invention, the historical average occupancy rate refers to the average seat utilization rate of flights on a specific route and departure date over a past period. The historical average occupancy rate is one of the important indicators to measure the popularity of a flight, and is used to assist subsequent clustering and grouping of flight information units, ensuring that each group has similar market demand characteristics and reducing experimental bias caused by occupancy rate differences. The higher the occupancy rate, the more popular the flight on this route and departure date, and vice versa. In a specific implementation, the historical average occupancy rate can be obtained by calculating the ratio of the actual number of sold seats to the total number of seats of this flight over a past period. This feature can be calculated using historical data such as the past week, month, quarter, or year, and the specific time range can be adjusted according to actual business needs. The historical average ticket price refers to the average selling price of tickets on a specific route and departure date over a past period. The historical average ticket price is one of the important indicators to measure the value of a flight, and is also used to assist subsequent clustering and grouping of flight information units, ensuring that each group has similar profitability characteristics and reducing experimental bias caused by ticket price differences. The higher the ticket price, the higher the value of the flight on this route and departure date, and vice versa. In a specific implementation, the historical average ticket price can be obtained by calculating the ratio of the actual total sales amount to the number of sold tickets of this flight over a past period. The statistical time range of the historical average ticket price can also be adjusted according to actual business needs, and the ticket price differences of different cabin classes can be considered. The advance booking days refer to the number of days when a user books a ticket before the flight takes off. The advance booking days reflect the ticket purchasing behavior habits of users and the changing trend of market demand. Using it as a feature for clustering and grouping can ensure that each group has similar user booking behavior characteristics and reduce experimental bias caused by booking time differences. For example, flights with longer advance booking days may be more popular among business travelers, while flights with shorter advance booking days may be more popular among travelers with temporary trips. In a specific implementation, the advance booking days can be calculated by counting all orders of this flight over a past period, calculating the difference between the booking date and the departure date for each order, and then taking the average. Extract features such as historical average occupancy rate, historical average ticket price, and advance booking days, which can reflect the popularity, profitability, and user booking behavior habits of flights. For feature extraction, in addition to historical average occupancy rate, historical average ticket price, and advance booking days, other flight-related features can also be extracted, such as: holiday coefficient reflecting the impact of holidays on flight occupancy rate and ticket price. The number of competing routes reflecting the degree of competition on this route. Aircraft type, different aircraft types may affect user choices. On-time rate, flights with higher on-time rates are more favored by users.

[0057] S120. Based on the features, use a clustering algorithm to pre-cluster the flight information units, and divide the similar flight information units into the same cluster. In the embodiments of the present invention, a clustering algorithm (such as K-means, DBSCAN, etc.) is used to analyze the extracted features. The clustering algorithm is an unsupervised learning method that divides similar objects into the same group (cluster). In the embodiments of the present invention, the clustering algorithm is used to divide the flight information units with similar historical occupancy rates, historical average ticket prices, or advance booking days into the same cluster. The purpose of pre-clustering is to reduce the computational complexity of the subsequent greedy algorithm, improve the diversion efficiency, and avoid falling into a local optimum prematurely. Among them, the choice of the clustering algorithm can be adjusted according to the actual situation. For example, the K-means algorithm is suitable for the case where the number of clusters is known. By iteratively optimizing the cluster centers, the data points are divided into the nearest clusters. The DBSCAN algorithm is suitable for the case where the number of clusters is unknown. By density clustering, the data points in the high-density area are divided into the same cluster. The hierarchical clustering algorithm can flexibly select different levels of clustering results by constructing a hierarchical clustering structure.

[0058] S130. Assign each cluster as a whole to different groups. In the embodiments of the present invention, assigning each cluster as a whole to different groups ensures the difference between the groups and avoids similar flight information units being concentrated in the same group, thus affecting the accuracy of the experimental results. The overall assignment method of the clusters can adopt random assignment or assignment based on specific rules. For example, it can be assigned according to the size of the cluster, and the larger clusters are assigned to different groups to ensure the balance of the sample sizes between the groups. In specific implementation, each cluster can be randomly assigned to different groups, or assigned according to certain rules. For example, according to the number of flight information units in each cluster, the clusters with a larger number are assigned to different groups to balance the number of flight information units between the groups.

[0059] It should be noted that the above historical average occupancy rate, historical average ticket price, and advance booking days are only exemplary features. In actual applications, other features can be selected according to specific business requirements, such as flight type, cabin class, user historical behavior, user preference settings, marketing activity information, etc. In addition, the choice of the clustering algorithm can also be adjusted according to the actual data situation, and is not limited to the K-means clustering, hierarchical clustering, and DBSCAN clustering mentioned above. Through the pre-clustering method, the flight information units with similar features can be divided into the same cluster, thereby reducing the computational complexity in the subsequent grouping process and improving the diversion efficiency. At the same time, assigning each cluster as a whole to different groups can ensure the difference between the groups and avoid similar flight information units being concentrated in the same group, thereby improving the accuracy of the experimental results.

[0060] In some embodiments, the preset grouping rule uses a greedy algorithm to group multiple flight information units, including the following steps:

[0061] S210, initially grouping multiple flight information units, and the initial grouping is achieved by random allocation or allocation based on clustering results. Random allocation is to randomly allocate flight information units to different groups. Based on the clustering result allocation, the flight information units are first clustered using a clustering algorithm (such as the K-Means algorithm), and then each cluster is allocated to different groups as a whole. This method can ensure that the similarity of flight information units in the same group is high, but it may not ensure that the index differences between the groups are small. Greedy algorithm, firstly, the flight information units are initially grouped (which can be achieved by random allocation or allocation based on clustering results), and then iteratively exchange flight information units between groups. If the difference in the largest index between the two becomes smaller after the exchange, and the difference in other indexes does not become larger, it is considered that the exchange can be performed to minimize the difference between groups. This method can ensure that the index differences between groups are as small as possible. When using the greedy algorithm, an objective function is defined to measure the difference between groups, such as the sum of the variances of each group on key indicators (such as the historical average passenger load factor, the historical average ticket price, etc.). In each iteration, two flight information units that can reduce the objective function value the most after the exchange are selected for exchange. In order to avoid falling into the local optimal solution, some randomness can be introduced, such as accepting the exchange with a certain probability that the objective function value increases after the exchange.

[0062] S220: Based on the initial grouping, iteratively exchange flight information units between groups to minimize differences between groups.

[0063] In order to further improve efficiency, a flexible diversion mode can be adopted. In this mode, the diversion difference value of at least one indicator is allowed to increase within a preset range, but the diversion difference values ​​of other indicators are restricted to decrease. For example, when evaluating the optimization strategy of the bundling rate of auxiliary products, the difference in air ticket prices can be allowed to increase slightly in exchange for a significant reduction in the difference in bundling rate.

[0064] In addition, in order to cope with the huge number of diverted objects in complex scenarios, when the number of diverted objects reaches hundreds of thousands, the diversion cost is calculated on a daily basis, and the cost increases exponentially. When the number of diverted objects reaches hundreds of millions, the memory load is very high. The following further optimization solutions can be used for data diversion:

[0065] Bucket storage, that is, for each object to be diverted, after a diversion indicator is given, each feature to be diverted is bucketed and stored in order, and the extreme value and number of each bucket are recorded, and a mapping table is constructed to record the bucket information and the location information of the diversion object under each indicator dimension.

[0066] Quickly locate exchangeable objects based on bucket information and location information to avoid global brute force traversal.

[0067] Pre-clustering is used to pre-cluster and group the diversion object groups with sparse features, which reduces the time consumed by calling the greedy algorithm for formal diversion. Test data shows that it can reduce the time consumed by calling the greedy algorithm for formal diversion by 76%, while effectively avoiding falling into the local optimum too early.

[0068] By adopting the above-mentioned grouping rules, the time consumption of diversion can be reduced exponentially, the flight information units can be effectively divided into different groups, and the difference in indicators between the groups can be ensured to be small. Compared with the traditional every-other-day experiment, the embodiment of the present invention can greatly shorten the experimental cycle and improve the experimental efficiency. In the successful diversion scenario, the time consumption of the 100,000-level is reduced by about 10%, the time consumption of the 1 million-level is reduced by about 22%, and the time consumption of the 10 million-level is reduced by about 48%. At the same time, the embodiment of the present invention can be applied to a variety of product optimization scenarios, such as dynamic pricing of flights, dynamic pricing of auxiliary products, flight Upsell, etc., and has broad application prospects.

[0069] The embodiment of the present invention further provides a data diversion method for comparing products on an online air ticket booking platform, comprising the following steps:

[0070] S400, using the above-mentioned data diversion method, multiple flight information units are divided into at least two groups. In an embodiment of the present invention, the above-mentioned data diversion method defines the core technical solution of data diversion, including the acquisition of flight information units, grouping algorithm (such as greedy algorithm), performance optimization (such as pre-clustering), and other optional technical features. Flight information unit refers to the basic unit used to describe the information of air ticket products, which is determined at least based on the route and the departure date of the route, and may further include cabin class information, flight type information, etc. This step uses an effective data diversion method to divide user access requests into different groups, thereby providing a basis for subsequent comparative tests. Different data diversion methods will affect the effect of grouping, such as the balance between groups, the similarity of users in the group, etc., and thus affect the accuracy and efficiency of the comparative test. Specifically, call the data diversion module, call the pre-developed data diversion module on the server side of the online air ticket booking platform, and pass the flight information unit to be diverted as an input parameter to the module. Select the number of groups, and select the appropriate number of groups according to actual needs. For example, it can be divided into group A and group B (the most common comparative test), or it can be divided into multiple groups (for more complex comparative tests). The implementation of the data diversion module, the specific implementation method of the data diversion module can be selected and combined according to the various technical features described in the above data diversion method. For example, you can choose to use a greedy algorithm for grouping, and use pre-clustering to optimize the grouping effect.

[0071] Although at least two groups are described in the above embodiments, in some cases, the flight information unit can also be divided into three or more groups. For example, a control group, experimental group A, and experimental group B can be set up. Various strategies such as random allocation, polling allocation, and hash allocation can be adopted. This embodiment does not limit the specific group allocation strategy, which can be selected according to actual needs.

[0072] S500. For different groups, different product display strategies are adopted. In the embodiment of the present invention, for different groups obtained by the data shunting method, different product display strategies are adopted to observe the impact of different strategies on user behavior. The product display strategy refers to the way of displaying ticket products to users on the online ticket reservation platform, including product sorting, price display, additional product recommendation, etc. Through the method of controlling variables, comparative tests are carried out on different product display strategies to find the optimal strategy and improve user experience and platform revenue. Specifically, configure the product display strategy. In the configuration system of the online ticket reservation platform, different product display strategies are configured for different groups. For example, the default product display strategy is configured for group A, and the optimized product display strategy is configured for group B. Route user requests. When a user accesses the online ticket reservation platform, according to the group to which the user belongs, route the user request to different servers or code branches, so as to adopt different product display strategies. Display different products. Select different tickets for display through data analysis, such as displaying tickets that are more in line with the user portrait of this group of users, or displaying tickets with higher profits. It should be noted that multiple different product display strategies can be adopted. This embodiment does not limit the specific type of product display strategy, which can be selected according to actual needs. For example, different product sorting rules, different price display methods, different additional product recommendation strategies, etc. can be adopted. The product display strategy can also be dynamically adjusted according to the real-time behavior of users. For example, the recommended additional products can be adjusted according to the user's click behavior, purchase behavior, etc. Through different product display strategies, the impact of different strategies on user behavior can be observed, providing data support for subsequent product optimization.

[0073] S600. Collect and analyze the user behavior data of each group, and evaluate the effects of different product display strategies. In the embodiments of the present invention, collect the behavior data of users in each group, such as clicks, browsing, purchases, etc., and analyze these data to evaluate the effects of different product display strategies. By statistically analyzing the user behavior data, it is possible to understand the impact of different product display strategies on user behavior, thereby evaluating the advantages and disadvantages of the strategies. Commonly used evaluation indicators include click-through rate, conversion rate, order volume, user satisfaction, etc. Specifically, embed points in the code of the online flight reservation platform to record the user behavior data, such as clicks, browsing, purchases, etc. Store the collected user behavior data in a database or data warehouse. Use data analysis tools (such as SQL, Python, etc.) to analyze the stored user behavior data and calculate the evaluation indicators for each group. It should be noted that multiple different data analysis tools can be used. The present embodiment does not limit the specific data analysis tool, and a suitable data analysis tool can be selected for analysis. In addition, different evaluation indicators can also be selected according to actual needs. For example, click-through rate, conversion rate, order volume, user satisfaction, etc. can be selected as evaluation indicators.

[0074] S700. According to the evaluation results, adjust the product display strategy of the online flight reservation platform to improve user experience and platform revenue. By iteratively optimizing the product display strategy, the effect of the strategy can be continuously improved, thereby achieving a win-win situation for user experience and platform revenue. Specifically, formulate a specific adjustment strategy according to the evaluation results. For example, if the conversion rate of group A is higher than that of group B, then apply the product display strategy of group A to all users. In the configuration system of the online flight reservation platform, update the configuration information of the product display strategy. Publish the updated configuration information to the server of the online flight reservation platform to make the new product display strategy take effect. It should be noted that multiple different adjustment strategies can be adopted. For example, the product display strategy with the best effect can be directly applied to all users, or a personalized adjustment strategy can be adopted according to the characteristics of users. The adjustment frequency of the strategy can also be adjusted according to actual needs. For example, it can be adjusted once a day or once a week.

[0075] Through the above steps, the embodiments of the present invention provide a complete data shunting method for product comparison testing of an online flight reservation platform. This method can effectively divide users into different groups and conduct comparative tests on different product display strategies, thereby providing a scientific basis for product optimization. At the same time, the embodiments of the present invention also have high flexibility and scalability, and can be adjusted and optimized according to actual needs.

[0076] Such as Figure 2As shown, an embodiment of the present invention further provides a data shunting system, which is applied to the product test of an online air ticket reservation platform based on flight information units and is used to implement the above data shunting method. The system includes:

[0077] M100, an acquisition module, which is used to acquire a plurality of flight information units, and the flight information units are determined based on the flight route and the departure date of the flight route. In the embodiment of the present invention, the acquisition module acquires a plurality of flight information units from a data source (such as a database, a data interface, etc.). These flight information units will serve as the basic data for subsequent grouping and comparative testing. Among them, the flight information unit is the basic unit for describing air ticket product information, which is determined at least based on the flight route and the departure date of the flight route, and may further include information such as cabin class information and flight type information. Ensure that a data basis for data shunting can be provided for subsequent modules. The acquisition module M100 acquires scheduling information or other data from the database or data interface of the online reservation platform to form flight information units. Specifically, the acquisition module M100 can acquire flight information units by calling internal data interfaces of the online air ticket reservation platform, such as flight scheduling interfaces and flight information interfaces. Preferably, data interaction can be carried out in ways such as RESTful API or gRPC. This acquisition module can also acquire flight information units by directly querying the database of the online air ticket reservation platform, such as MySQL, Oracle, etc. In order to improve query efficiency, technologies such as indexing and caching can be adopted. The type, address, authentication information, etc. of the data source can be flexibly configured through a configuration file or a management interface. It should be noted that the embodiment of the present invention does not specifically limit the data source for acquiring flight information units. For example, data can be acquired from an internal database, a third-party data provider, or a crawler program, etc. Through this acquisition module, flight information units can be acquired efficiently and stably, providing a reliable data source for subsequent data shunting and product comparative testing.

[0078] M200, a grouping module, is used to divide multiple flight information units into at least two groups based on a preset grouping rule for comparative testing. In an embodiment of the present invention, the grouping module M200 divides the multiple flight information units provided by the acquisition module into at least two groups according to the preset grouping rule for subsequent comparative testing. The grouping rule refers to the strategy for determining the group to which a flight information unit belongs, such as random assignment, assignment based on specific attributes, prediction based on machine learning algorithms, etc. According to the preset grouping rule, the flight information units are classified to achieve data shunting. The selection of the grouping rule directly affects the grouping effect, such as the balance between groups and the similarity of users within the group, thereby affecting the accuracy and efficiency of the comparative testing. Specifically, a rule engine can be used to implement the configuration and execution of the grouping rule. For example, the Drools rule engine can be used. Custom grouping algorithms can also be developed according to specific business requirements. For example, greedy algorithms, clustering algorithms, etc. can be adopted. The initial grouping method can be based on the initial grouping, and the following operations are iteratively executed to minimize the difference between groups: select flight information units from different groups for exchange. The grouping strategy can configure the grouping rule through a visual interface, including algorithm selection, parameter setting, and rule priority, etc., to support flexible grouping strategy configuration, such as A / B testing, multivariate testing, etc. It should be noted that multiple different grouping algorithms can be adopted for the grouping algorithm. For example, random assignment, hash assignment, clustering algorithms, machine learning algorithms, etc. can be adopted. The number of groups can also be divided according to actual needs, and the flight information units are divided into different numbers of groups. For example, it can be divided into group A and group B (the most common comparative testing), or it can be divided into multiple groups (for more complex comparative testing). Through this grouping module M200, flexible and efficient data shunting can be achieved, providing different experimental groups for subsequent product comparative testing.

[0079] M300, an optimization module, is used to perform different product optimization operations on flight information units in different groups according to a preset product test strategy. In an embodiment of the present invention, the optimization module is an execution module of the data shunting system, and its main function is to perform different product optimization operations on flight information units in different groups according to a preset product test strategy, such as price adjustment, product recommendation, page layout optimization, etc. The product test strategy refers to different marketing means or product strategies adopted for different groups to verify the effectiveness of the strategies. The optimization module M300 selects a better strategy based on the test results of controlling variables. Specifically, the strategy configuration is to configure different product test strategies for different groups through a configuration system, including various different strategy types, such as price adjustment, product recommendation, page layout optimization, etc. The strategy execution engine uses a strategy execution engine to achieve the dynamic adjustment and execution of the strategy. For example, the MVEL expression engine can be used. By calling the internal interface of the online air ticket reservation platform, product optimization operations are realized, such as adjusting the product price, adjusting the recommendation list, adjusting the page layout, etc. A unified interface specification is adopted to facilitate expansion and maintenance. In the specific implementation process, different product test strategies can be selected according to actual needs. For example, strategies such as price discounts, point rewards, and coupons can be adopted. Multiple strategy effectiveness modes such as taking effect immediately, taking effect at a scheduled time, and being triggered by events can also be adopted. Through this optimization module, different product test strategies can be implemented for flight information units in different groups, and data support is provided for subsequent effect evaluation.

[0080] The above-mentioned modules work together to implement a data shunting system based on flight information units. This system can effectively divide users into different groups and conduct comparative tests on different product test strategies, thus providing a scientific basis for product optimization. At the same time, this system also has high flexibility and scalability and can be adjusted and optimized according to actual needs.

[0081] An embodiment of the present invention also provides a data shunting program product. The program product includes computer instructions, and when the computer instructions are executed by a processor, the steps of the data shunting method in the above-mentioned embodiment are implemented. The data shunting program product refers to a software product containing computer-executable instructions and can be installed and run on a computer or other devices with processing capabilities. By converting the steps of the data shunting method into computer-executable instructions, this program product enables the computer to automatically execute these steps, thereby realizing the automation of data shunting and product comparative testing. This program product can be implemented using different programming languages, such as Java, Scala, Python, etc. It can be compiled and optimized for different operating systems and hardware platforms. This program product can be deployed in different deployment methods, such as directly deployed on a physical machine, deployed on a virtual machine, deployed in a containerized environment, etc.

[0082] An embodiment of the present invention further provides a data shunting device, including:

[0083] a processor;

[0084] a memory, which stores executable instructions of the processor;

[0085] The processor is configured to execute the steps of the above data shunting method by executing the executable instructions.

[0086] Next, the electronic device 600 according to this embodiment of the present application will be described with reference to Figure 3 The electronic device 600 shown is only an example, and should not impose any limitations on the functions and usage scope of the embodiments of the present application. Figure 3

[0087] Figure 3 As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The 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 system components (including the storage unit 620 and the processing unit 610), a display unit 640, etc.

[0088] Figure 1 Among them, the storage unit stores program codes, and the program codes 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 application described in the above data shunting method part of this specification. For example, the processing unit 610 can execute the steps as shown in

[0089]

[0090] 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.

[0091] The storage unit 620 may further include a program / utilities 6204 having a set (at least one) of program modules 6205. Such program modules 6205 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. The implementation of a network environment may be included in each or some combination of these examples.

[0092] ​​​The electronic device 600 can also communicate with one or more external devices 700 (such as a keyboard, a pointing device, a Bluetooth device, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (such as a router, a modem, etc.). Such communication can be carried out through the input / output (I / O) interface 650. Moreover, the electronic device 600 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through the network adapter 660. The network adapter 660 can communicate with other modules of the electronic device 600 through the bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in combination with the 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 systems, etc.

[0093] In the data shunting device, when the program in the memory is executed by the processor, the steps of the data shunting method are implemented. Therefore, the device can also obtain the technical effects of the above data shunting method.

[0094] An embodiment of the present invention also provides a computer-readable storage medium for storing a program, characterized in that when the program is executed by the processor, the steps of the above data shunting method are implemented. In some possible implementation manners, various aspects of the present application can also be implemented in the form of a program product, which includes program code. When the program product is executed on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments described in the above data shunting method part of this specification.

[0095] Reference Figure 4 As shown, a program product 800 for implementing the above method according to an embodiment of the present application is described. It can be a portable compact disc read-only memory (CD-ROM) and includes program code, and can be executed on a terminal device, such as a personal computer. However, the program product of the present application is not limited thereto. In this document, the readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, device, or device.

[0096] The program product may employ any combination of one or more readable media. The readable media 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 any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0097] A computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, in which case the data signal carries the readable program code. Such a propagated data signal may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The readable storage medium may also be any readable medium other than a readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0098] The program code for performing the operations of this application may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and also including conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, executed as a stand-alone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., by using an Internet service provider to connect through the Internet).

[0099] The steps of the data diversion method are implemented when the program in the computer storage medium is executed by a processor. Therefore, the computer storage medium can also achieve the technical effects of the above data diversion method.

[0100] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A data diversion method, applied to product testing of an online air ticket booking platform based on flight information units, characterized in that: The following steps are involved: Acquire a plurality of flight information units, wherein the flight information units are determined based on routes and route departure dates; Based on a preset grouping rule, the plurality of flight information units are divided into at least two groups for comparative testing; According to the preset product testing strategy, different product optimization operations are performed on different groups of flight information units.

2. The data offloading method according to claim 1, characterized in that: The flight information unit includes at least one of basic attributes of a ticket and business scenario requirements.

3. The data splitting method according to claim 2, characterized in that: The basic attributes of the air ticket include route, departure date, cabin class, and flight type, and the business scenario requirements include user historical behavior, user preference settings, and marketing campaign information.

4. The data offloading method according to claim 1, characterized in that: After obtaining a plurality of flight information units, the flight information units are determined based on routes and route departure dates, the following steps are also included: For each flight information unit, extract at least one feature, the feature comprising: a historical average passenger load factor, a historical average ticket price, and advance booking days; Based on the features, a clustering algorithm is used to pre-cluster the flight information units, and similar flight information units are divided into the same cluster; Assign each cluster as a whole to different groups.

5. The data offloading method according to claim 1, characterized in that: The preset grouping rule uses a greedy algorithm to group the multiple flight information units, including the following steps: Initially grouping the multiple flight information units, wherein the initial grouping is achieved by random allocation or allocation based on clustering results; Based on the initial grouping, the flight information units between groups are iteratively exchanged to minimize the differences between groups.

6. The data offloading method according to claim 5, characterized in that: In the step of iteratively exchanging flight information units between groups, the diversion difference value of at least one indicator is allowed to increase within a preset range, but at the same time, the diversion difference values ​​of the remaining indicators are restricted to decrease.

7. A data diversion method for comparing products on an online air ticket booking platform, characterized in that: The following steps are involved: Using the data splitting method according to any one of claims 1 to 6, the plurality of flight information units are divided into at least two groups; Use different product display strategies for different groups; Collect and analyze user behavior data for each group to evaluate the effectiveness of different product display strategies; Adjust the product display strategy of the online air ticket booking platform based on the evaluation results.

8. A data diversion system, applied to product testing of an online air ticket booking platform based on a flight information unit, for implementing the data diversion method according to any one of claims 1 to 6, characterized in that: The system comprises: An acquisition module, used to acquire a plurality of flight information units, wherein the flight information units are determined based on routes and route departure dates; A grouping module, used for dividing the plurality of flight information units into at least two groups based on a preset grouping rule for comparative testing; The optimization module is used to perform different product optimization operations on different groups of flight information units according to preset product testing strategies.

9. A data diversion program product, characterized in that: The program product comprises computer instructions, which implement the steps of the method according to any one of claims 1 to 6 when executed by a processor.

10. A data distribution device, characterized in that: include: processor; a memory storing executable instructions of the processor; The processor is configured to execute the steps of the data diversion method according to any one of claims 1 to 6 by executing the executable instructions.

11. A computer-readable storage medium for storing a program, characterized in that: When the program is executed by a processor, the steps of the data diversion method described in any one of claims 1 to 6 are implemented.

Images