Dynamic packaging method, device and program product for airline tickets and ancillary products

Abstract

This application discloses a dynamic packaging method, device, and program for airline tickets and ancillary products, relating to the field of civil aviation service technology, aiming to solve the problem of inaccurate recommended ancillary products. The method includes: constructing a multi-dimensional feature vector based on interaction behavior characteristics, user profile characteristics, and flight scenario characteristics corresponding to the user's booked flight; inputting the multi-dimensional feature vector into an MMOE model to obtain the conversion probability of each ancillary product among multiple ancillary products; calculating the correlation coefficient between different categories of ancillary products based on the correlation matrix of multiple categories of ancillary products; determining the packaging score of multiple ancillary product combinations based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product; and determining the target ancillary product combination with the highest packaging score from the multiple ancillary product combinations.

Description

Technical Field

[0001] This application relates to the field of civil aviation service technology, and in particular to a method, equipment and program product for dynamically packaging air tickets and ancillary products. Background Technology

[0002] In the context of civil aviation ticketing services, users typically perform operations such as flight selection, information filling, and purchasing ancillary products during the ticket booking process.

[0003] During the flight booking process, airline ticket booking systems can typically recommend and display ancillary products such as seat selection, baggage, and insurance to users according to fixed rules (such as descending order of purchase popularity).

[0004] However, the ancillary products currently recommended and displayed are not accurate enough. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, and program for dynamically packaging airline tickets and ancillary products, aiming to solve the problem that the recommended ancillary products are not accurate enough.

[0006] Firstly, this application provides a method for dynamically packaging airline tickets and ancillary products. The method includes: acquiring user interaction behavior characteristics during the ticket booking process; constructing a multi-dimensional feature vector based on the interaction behavior characteristics, user profile characteristics, and flight scenario characteristics corresponding to the flight booked by the user; inputting the multi-dimensional feature vector into a multi-gate hybrid expert MMOE model to obtain the conversion probability of the user for each of the multiple ancillary products predicted by the MMOE model; the MMOE model includes multiple expert networks, each expert network corresponding to a category of ancillary products, and each expert network is used to predict the corresponding category of ancillary products based on the multi-dimensional feature vector. The conversion probability is calculated; based on historical order data of various ancillary products, a correlation matrix of various ancillary products is constructed; based on the correlation matrix, the correlation coefficient between different categories of ancillary products is calculated; based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product, the package score of multiple ancillary product combinations is determined; the target ancillary product combination with the highest package score is determined from multiple ancillary product combinations; a display instruction for the target ancillary product combination is sent to the user's terminal device so that the terminal device displays the target ancillary product combination on the display interface of the flight booking process.

[0007] Optionally, the interactive behavior characteristics include at least one of the following: price comparison behavior characteristics, time attention characteristics, model attention characteristics, comfort preference characteristics, paid seat selection preference characteristics, seat selection preference characteristics, and decision-making time characteristics.

[0008] Optionally, user profile features may include at least one of the following: membership level features, accumulated mileage features, membership duration features, time sensitivity features, price sensitivity features, travel structure features, refund and change preference features, baggage preference features, seat selection preference features, and insurance preference features.

[0009] Optionally, flight scenario features include at least one of the following: red-eye flight features, remaining seat features, on-time performance features, and weather features.

[0010] Optionally, based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product, the package score of multiple ancillary product combinations is determined, including: calculating the package score according to the following formula: ; in, Indicates ancillary products and ancillary products Package score of the ancillary product portfolio; Indicates ancillary products Product profitability coefficient; Indicates ancillary products Product profitability coefficient; Indicates ancillary products and ancillary products The probability of combinational transformation; ; in, Indicates ancillary products The conversion probability; Indicates ancillary products The conversion probability; Indicates the first Ancillary products and the first Correlation coefficients between ancillary products; ancillary products Belongs to the Ancillary products; Ancillary products Belongs to the Ancillary products.

[0011] Optionally, the MMOE model is trained in the following way: obtaining a training sample set; the training sample set includes multiple training samples, each training sample including the interaction behavior characteristics of historical users in the flight booking process, the user profile characteristics of historical users, the flight scenario characteristics of the flight corresponding to the flight booked by historical users, and the purchase tags of historical users for various categories of ancillary products; and performing multi-task training on the initial MMOE model based on the training sample set to obtain the trained MMOE model.

[0012] Secondly, this application provides a dynamic packaging device for airline tickets and ancillary products, which includes various functional modules for the method described in the first aspect above.

[0013] Thirdly, this application provides an electronic device comprising: a processor and a memory; the memory storing instructions executable by the processor; the processor being configured to, when executing the instructions, cause the electronic device to perform the method provided in the first aspect above.

[0014] Fourthly, this application provides a computer-readable storage medium comprising: computer software instructions; when the computer software instructions are executed in an electronic device, the electronic device causes the electronic device to implement the method provided in the first aspect above.

[0015] Fifthly, this application provides a computer program product including computer instructions that, when executed on an electronic device, cause the electronic device to perform the method provided in the first aspect above.

[0016] The technical solution provided in this application has at least the following beneficial effects: The dynamic packaging method for airline tickets and ancillary products provided in this application no longer relies on single purchase popularity data. Instead, it constructs a multi-dimensional feature vector by acquiring user interaction behavior characteristics, user profile characteristics, and flight scenario characteristics. This comprehensively portrays users' real-time preferences and scenario attributes, providing a data foundation for subsequent accurate predictions. It can avoid the problem that traditional fixed rules cannot adapt to users' personalized needs, thereby improving the accuracy of the recommended ancillary products.

[0017] Furthermore, this application employs an MMOE model comprising multiple independent expert networks, with each expert network corresponding to a category of ancillary products. It predicts the conversion probability of each category in parallel based on a unified multi-dimensional feature vector. This approach not only achieves knowledge sharing among different categories but also improves the accuracy of conversion prediction for each ancillary product by learning the unique user preference patterns of each category through independent expert networks, overcoming the deficiency of traditional recommendation methods in lacking user preference modeling capabilities.

[0018] Furthermore, this application also constructs a correlation matrix for ancillary products based on historical order data, which quantifies the relationships between different product categories. This effectively identifies high-frequency combinations that users purchase simultaneously, avoids recommending products with poor correlation or mutual exclusion, improves the rationality of product combinations, and solves the problem that traditional fixed rules cannot take into account the logic of product pairing.

[0019] Finally, this application incorporates the conversion probability, cross-category correlation coefficient, and product revenue coefficient of each ancillary product into a unified packaged scoring system. It comprehensively evaluates the ancillary product combination from multiple dimensions, including user acceptance, reasonableness of the combination, and business value, ultimately determining the target ancillary product combination with the highest score. This approach ensures both the personalization and rationality of recommendations while optimizing business revenue, overcoming the technical limitations of traditional recommendation methods that rely on a single rule and cannot simultaneously address multiple objectives. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A schematic diagram illustrating the composition of a dynamic packaging system for airline tickets and ancillary products provided in an embodiment of this disclosure; Figure 2 A flowchart illustrating the dynamic packaging method for airline tickets and ancillary products provided in this embodiment of the disclosure; Figure 3 This is a schematic diagram of the structure of the MMOE model provided in the embodiments of this application; Figure 4 This is a schematic diagram of the MMOE model training process provided in the embodiments of this application; Figure 5 A schematic diagram of the composition of the dynamic packaging device for airline tickets and ancillary products provided in the embodiments of this application; Figure 6 This is a schematic diagram illustrating the composition of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0024] In embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, article, or apparatus that includes that element.

[0025] In the embodiments of this application, the words "exemplarily" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of the words "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0026] In the context of civil aviation ticketing services, users typically perform operations such as flight selection, information filling, and purchasing ancillary products during the ticket booking process.

[0027] During the flight booking process, airline ticket booking systems can typically recommend and display ancillary products such as seat selection, baggage, and insurance to users according to fixed rules (such as descending order of purchase popularity).

[0028] However, the ancillary products currently recommended and displayed are not accurate enough.

[0029] Based on this, embodiments of this application provide a method, device, and program product for dynamically packaging airline tickets and ancillary products. This method can recommend and display target ancillary product combinations by combining the user's interactive behavior characteristics during the ticket booking process, the user's user profile characteristics, and the flight scenario characteristics of the flight corresponding to the user's booked ticket. This allows for accurate identification of the user's current immediate needs and improves the accuracy of the recommended ancillary products.

[0030] The following description is provided in conjunction with the accompanying drawings.

[0031] Figure 1 This is a schematic diagram illustrating the composition of a dynamic packaging system for airline tickets and ancillary products provided in an embodiment of this disclosure. Figure 1 As shown, the dynamic packaging system may include: a terminal device 100 and an auxiliary product packaging device 200. The terminal device 100 is communicatively connected to the auxiliary product packaging device 200.

[0032] Terminal device 100 can be an electronic device with network interaction and interface display functions, such as a smartphone, tablet computer, laptop computer, desktop computer, or counter ticketing terminal. Figure 1(Taking a smartphone as an example), the terminal 100 may be equipped with a flight booking client or a flight booking webpage interactive interface.

[0033] Terminal device 100 can be used to collect user interaction data during the flight booking process and send the interaction data to ancillary product packaging device 200.

[0034] As an example, before collecting user interaction data during the flight booking process, the terminal device 100 or the flight booking client or flight booking webpage interface installed on the terminal device 100 can obtain the user's authorization for collecting interaction data.

[0035] For example, terminal device 100 can display authorization options for collecting interactive behavior data during the flight booking process in the flight booking client or flight booking webpage interactive interface, and in response to the user's check-in operation or signature operation of the authorization option, determine that the user's authorization operation for collecting interactive behavior data has been obtained.

[0036] In some embodiments, the terminal device 100 can also be used to receive a display instruction for a target ancillary product combination sent by the ancillary product packaging device 200, and in response to the display instruction, render and display the target ancillary product combination in the interactive interface of the flight booking client or the flight booking webpage, so as to complete the product and operation response at the human-computer interaction level.

[0037] The auxiliary product packaging device 200 can be an electronic device with computing and processing functions, such as a computer or server.

[0038] The server can be a single server or a server cluster consisting of multiple servers. In some embodiments, the server cluster can also be a distributed cluster. Optionally, the server can also be implemented on a cloud platform, such as a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, inter-cloud, and multi-cloud, or any combination thereof. This disclosure does not impose any limitations on these embodiments.

[0039] The ancillary product packaging device 200 can be used to extract interactive behavior features from the interactive behavior data sent by the terminal device 100, determine the target ancillary product combination based on user profile features and flight scenario features corresponding to the user's booked flight, and send a display instruction for the target ancillary product combination to the terminal device 100. The specific process can be referred to the dynamic packaging method for air tickets and ancillary products provided in the following method embodiment, and will not be repeated here.

[0040] The executing entity of the dynamic packaging method for airline tickets and ancillary products provided in this application embodiment is a dynamic packaging device for airline tickets and ancillary products (hereinafter referred to as the ancillary product packaging device). As mentioned above, the ancillary product packaging device (ancillary product packaging device 200) can be an electronic device with computing processing capabilities, such as a computer or server. Optionally, the ancillary product packaging device can also be a processor (e.g., a central processing unit (CPU)) in the aforementioned electronic device; or, the ancillary product packaging device can also be a software system or platform deployed in the aforementioned electronic device; or, the ancillary product packaging device can also be a functional module in the aforementioned electronic device used to execute the dynamic packaging method for airline tickets and ancillary products, etc. This disclosure embodiment does not impose limitations here.

[0041] For simplicity, the following description will use the auxiliary product packaging device as the execution subject of the dynamic packaging method for airline tickets and ancillary products provided in the embodiments of this application.

[0042] Figure 2 This is a flowchart illustrating the dynamic packaging method for airline tickets and ancillary products provided in this embodiment of the disclosure. Figure 2 As shown, the method may include the following steps: S101. Obtain the user's interactive behavior characteristics during the flight booking process.

[0043] As an example, as described above, the ancillary product packaging device (ancillary product packaging device 200) can be communicatively connected to a terminal device (terminal device 200). The terminal device can collect user interaction behavior data during the flight booking process and send this interaction behavior data to the ancillary product packaging device. In this case, the ancillary product packaging device can receive the interaction behavior data sent by the terminal device and extract interaction behavior features from the interaction behavior data.

[0044] As an example, interactive behavior characteristics may include at least one of the following: price comparison behavior characteristics, time attention characteristics, model attention characteristics, comfort preference characteristics, paid seat selection preference characteristics, seat selection preference characteristics, and decision-making time characteristics.

[0045] Among them, price comparison behavior features can be used to represent the number of times a user switches dates on the flight list page and the number of times they click the price sorting control in a single session.

[0046] Time focus features can be used to represent the number of times a user switches dates on the flight list page and the number of times they click the time sorting control within a single session.

[0047] Aircraft type popularity can be used to indicate the number of times a user filters for aircraft types on the flight list page within a single session.

[0048] Comfort preference features can be used to represent the number of times a user filters for large aircraft types on the flight list page within a single session. Large aircraft types refer to aircraft with more than a preset seat capacity threshold or those classified as wide-body passenger aircraft.

[0049] Paid seat selection preference features can be used to indicate whether a user uses frequent flyer miles to select a seat within a historical period of a preset duration.

[0050] Seat selection preference features can be used to indicate whether a user has selected a seat within a historical period of a preset duration.

[0051] The decision time feature can be used to represent the difference between the time a user places an order and the time they first search for a flight.

[0052] S102. Construct a multi-dimensional feature vector based on interaction behavior features, user profile features, and flight scenario features corresponding to the flight booked by the user.

[0053] As an example, user profile features may include at least one of the following: membership level features, accumulated mileage features, membership duration features, time sensitivity features, price sensitivity features, travel structure features, refund and change preference features, baggage preference features, seat selection preference features, and insurance preference features.

[0054] Among them, the membership level feature can be used to represent the user's membership level identifier in the aviation service system. This membership level identifier is based on the user's historical consumption or travel data and is divided into a preset level.

[0055] The cumulative mileage feature can be used to represent the total amount of flight mileage points that a user has accumulated based on air travel within a preset statistical period, which can be used to redeem services.

[0056] Membership duration can be used to represent the duration of time a user has been a member since registration.

[0057] Time sensitivity features can be used to represent the time period during which a user purchased tickets in advance within their historical order history.

[0058] Price sensitivity can be used to represent the proportion of times a user selects low-priced flights out of their total flight history. A low-priced flight refers to a ticket price lower than a preset price threshold. This preset price threshold can be set to be 75%-80% lower than the average selling price of the same route and class of service over the past N days. For example, the preset price threshold for short-haul routes (flight distance ≤ 800 km) can be set to 200 yuan, 250 yuan, or 300 yuan; the preset price threshold for medium- and long-haul routes (flight distance > 800 km) can be set to 400 yuan, 500 yuan, or 600 yuan. This application does not limit the specific value of the preset price threshold.

[0059] Travel structure features can be used to indicate whether a user is traveling alone, on a family trip, or in a group; or in other words, to indicate the number of people traveling with the user.

[0060] The refund and change preference feature can be used to represent the frequency of a user's ticket refund or change behavior in historical orders within a preset statistical period.

[0061] Baggage preference features can be used to indicate whether a user has purchased additional baggage allowance in their historical orders.

[0062] Seat selection preference features can be used to indicate whether a user has chosen to pay for seat selection in their historical orders.

[0063] Insurance preference features can be used to indicate whether a user has purchased ancillary insurance products in their historical orders.

[0064] As an example, flight scenario features may include at least one of the following: red-eye flight features, remaining seat features, on-time performance features, and weather features.

[0065] Among these features, red-eye flight characteristics can be used to indicate whether the flight's departure and arrival times fall within a preset nighttime period (e.g., departure time later than a preset nighttime threshold or arrival time earlier than a preset early morning threshold). Red-eye flight passengers may prefer front-row seats.

[0066] Remaining seat characteristics can be used to represent the number of available seats for a flight at the time of a user's query, the proportion of available seats to the total number of seats, or the distribution of remaining seats. When the number of remaining available seats is lower than a seat quantity threshold, family travelers may be more inclined to choose seats together with their family members.

[0067] On-time performance is a metric used to represent the historical on-time performance of a flight booked by a user within a preset statistical period. When the on-time performance rate is below a threshold, users may be more inclined to cancel or change their tickets or purchase flight delay insurance.

[0068] Weather characteristics can be used to indicate the weather conditions in the airspace at the departure point, destination, or transit point of a booked flight during the scheduled flight period. In severe weather conditions, users may prefer to cancel or change their tickets or purchase insurance.

[0069] S103. Input the multi-dimensional feature vector into the Multi-gate Mixture-of-Experts (MMOE) model to obtain the conversion probability of the user for each of the multiple ancillary products predicted by the MMOE model.

[0070] The MMOE model includes multiple expert networks, each corresponding to a category of ancillary products. Each expert network is used to predict the conversion probability of the corresponding category of ancillary products based on multi-dimensional feature vectors.

[0071] For example, Figure 3 This is a schematic diagram of the structure of the MMOE model provided in an embodiment of this application. Figure 3 As shown, an MMOE model can include an input layer, an embedding layer, an expert network layer, and an output layer.

[0072] The input layer can be used to receive user interaction behavior features, user profile features, and flight scenario features of the flight corresponding to the user's booked ticket during the flight booking process. It maps the interaction behavior features, user profile features, and flight scenario features into an input format that can be read by the MMOE model, providing a data foundation for subsequent feature processing.

[0073] The embedding layer can be used to uniformly process the interactive behavior features, user profile features, and flight scenario features of the input layer, constructing a multi-dimensional feature vector. For example, discrete features can be embedded and encoded (e.g., one-hot encoding), while numerical features can be used directly. The embedding layer converts discrete features into matrices and concatenates them to obtain a multi-dimensional feature vector, which is then input into the expert network.

[0074] The expert network layer can include multiple expert networks, each corresponding to a category of ancillary products. Each expert network is used to predict the conversion probability of the corresponding category of ancillary products based on multi-dimensional feature vectors.

[0075] The output layer can be used to output the conversion probability of a user to each of the multiple ancillary products predicted by the MMOE model.

[0076] S104. Based on historical order data of various types of ancillary products, construct a correlation matrix for various types of ancillary products.

[0077] As an example, the ancillary product packaging device can extract historical order data within a preset statistical period, and statistically analyze the selection status of various ancillary products in each historical order (e.g., "1" indicates that the product was selected, and "0" indicates that it was not selected), forming a two-dimensional data matrix of orders and products. Then, based on the aforementioned data matrix, a correlation coefficient algorithm is used to calculate the co-occurrence correlation between any two categories of ancillary products in historical orders. Finally, all pairwise correlation coefficients are filled into a matrix of preset dimensions to form an N×N order correlation matrix (N is the number of categories of ancillary products). The element in the i-th row and j-th column of the matrix is ​​the correlation coefficient between the i-th category of products and the j-th category of products.

[0078] S105. Based on the correlation matrix, calculate the correlation coefficient between different categories of ancillary products.

[0079] As an example, the correlation coefficient between different categories of auxiliary tea products can be obtained using the Pearson correlation coefficient. The calculation process of the Pearson correlation coefficient can be found in the relevant technical documentation, and will not be repeated here.

[0080] S106. Based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product, determine the package score of multiple ancillary product combinations.

[0081] As an example, the packaging score for ancillary product packaging devices can be calculated using the following formula: ; in, Indicates ancillary products and ancillary products Package score of the ancillary product portfolio; Indicates ancillary products Product profitability coefficient; Indicates ancillary products Product profitability coefficient; Indicates ancillary products and ancillary products The probability of combinational transformation; ; in, Indicates ancillary products The conversion probability; Indicates ancillary products The conversion probability; Indicates the first Ancillary products and the first Correlation coefficients between ancillary products; ancillary products Belongs to the Ancillary products; Ancillary products Belongs to the Ancillary products.

[0082] S107. Identify the target ancillary product portfolio with the highest packaging score from multiple ancillary product portfolios.

[0083] As an example, the ancillary product packaging device can sort multiple ancillary product combinations according to their packaging scores from high to low, obtain a sorting result, and determine the top M ancillary product combinations in the sorting result as the target ancillary product combinations.

[0084] Where M is a positive integer. For example, M can be set to 1, 2, 3, 4, or 5, etc. The embodiments of this application do not limit the specific value of M.

[0085] S108. Send a display instruction for the target ancillary product combination to the terminal device operated by the user, so that the terminal device displays the target ancillary product combination on the display interface of the flight booking process.

[0086] As an example, the algorithm application process of the dynamic packaging method for airline tickets and ancillary products provided in this application embodiment is illustrated as follows: For the ticket purchase behavior of general users, after they select a flight and enter the secondary page, the system captures the user's "family leisure travel" characteristics (such as three people traveling together including children, flying to a tourist destination) based on passenger profiles, flight information, and other data. It determines the user's strong demand for "family seats" and "baggage allowance", as well as their psychological characteristics of price sensitivity. The system then no longer mechanically lists individual items, but dynamically generates a customized "worry-free family package" that includes "seat selection + extra baggage allowance", and actively calculates the optimal discount based on the user's willingness to pay, thereby naturally completing the conversion of ancillary products while improving the user experience.

[0087] The dynamic packaging method for airline tickets and ancillary products provided in this application no longer relies on single purchase popularity data. Instead, it constructs a multi-dimensional feature vector by acquiring user interaction behavior characteristics, user profile characteristics, and flight scenario characteristics. This comprehensively portrays the user's real-time preferences and scenario attributes, providing a data foundation for subsequent accurate prediction. It can avoid the problem that traditional fixed rules cannot adapt to users' personalized needs, thereby improving the accuracy of the recommended ancillary products.

[0088] Furthermore, this application employs an MMOE model comprising multiple independent expert networks, with each expert network corresponding to a category of ancillary products. It predicts the conversion probability of each category in parallel based on a unified multi-dimensional feature vector. This approach not only achieves knowledge sharing among different categories but also improves the accuracy of conversion prediction for each ancillary product by learning the unique user preference patterns of each category through independent expert networks, overcoming the deficiency of traditional recommendation methods in lacking user preference modeling capabilities.

[0089] Furthermore, this application also constructs a correlation matrix for ancillary products based on historical order data, which quantifies the relationships between different product categories. This effectively identifies high-frequency combinations that users purchase simultaneously, avoids recommending products with poor correlation or mutual exclusion, improves the rationality of product combinations, and solves the problem that traditional fixed rules cannot take into account the logic of product pairing.

[0090] Finally, this application incorporates the conversion probability, cross-category correlation coefficient, and product revenue coefficient of each ancillary product into a unified packaged scoring system. It comprehensively evaluates the ancillary product combination from multiple dimensions, including user acceptance, reasonableness of the combination, and business value, ultimately determining the target ancillary product combination with the highest score. This approach ensures both the personalization and rationality of recommendations while optimizing business revenue, overcoming the technical limitations of traditional recommendation methods that rely on a single rule and cannot simultaneously address multiple objectives.

[0091] In some possible embodiments, prior to S101 above, the ancillary product packaging device may also acquire the trained MMOE model.

[0092] In one possible implementation, the ancillary product packaging device can directly obtain the trained MMOE model from other devices. The process of training the MMOE model by other devices can be referred to the following... Figure 4 The steps described herein will not be repeated here.

[0093] For example, the auxiliary product packaging device can pull the trained MMOE model online through network transmission methods such as a dedicated intranet transmission link, an encrypted HyperText Transfer Protocol (HTTP) / HyperText Transfer Protocol Secure (HTTPS) interface, or a cloud model distribution service.

[0094] For example, the ancillary product packaging device can also copy and transfer the trained MMOE model via offline storage media such as USB flash drives, solid-state drives, or encrypted hard drives.

[0095] For example, the auxiliary product packaging device can also complete the cross-device MMOE model retrieval through cluster deployment methods such as server cluster image synchronization, containerized image loading, or microservice configuration distribution.

[0096] For example, an MMOE model can be trained by an independent training device, which can then push the trained MMOE model to an auxiliary product packaging device for local deployment via a preset data push interface.

[0097] It should be noted that the above-mentioned method of the ancillary product packaging device obtaining the trained MMOE model from other devices is only an example. The embodiments of this application do not limit the specific method of the ancillary product packaging device obtaining the trained MMOE model from other devices.

[0098] In another possible implementation, the ancillary product packaging device can also train an MMOE model based on a training sample set. In this case, Figure 4 This is a schematic diagram of the MMOE model training process provided in an embodiment of this application. Figure 4 As shown, the training process of an MMOE model can include the following steps: S201. Obtain the training sample set.

[0099] The training sample set includes multiple training samples. Each training sample includes the interactive behavior characteristics of historical users during the flight booking process, the user profile characteristics of historical users, the flight scenario characteristics of the flight corresponding to the flight booked by historical users, and the purchase tags of historical users for various categories of ancillary products.

[0100] S202. Perform multi-task training on the initial MMOE model based on the training sample set to obtain the trained MMOE model.

[0101] As an example, the ancillary product packaging device can input one or more training samples into the initial MMOE model each time to obtain the MMOE model's predicted purchase results for each category of ancillary products. Then, it compares the predicted purchase results with the purchase tags in the training samples, calculates the loss function, and adjusts the parameters in the initial MMOE model based on the loss function until the preset iteration stopping condition is reached, thus obtaining the trained MMOE model.

[0102] For example, the preset iteration stopping condition may include: the number of times the training samples are input into the initial MMOE model reaches a threshold; and / or, the error between the predicted purchase result output by the initial MMOE model and the purchase tag in the training samples is less than or equal to an error threshold.

[0103] The number of times threshold can be preset in the packaging device for auxiliary products. For example, the number of times threshold can be set to 500 times, 1000 times, 5000 times, or 10000 times, etc. This application embodiment does not limit the specific value of the number of times threshold. The error threshold can also be preset in the packaging device for auxiliary products. For example, the error threshold can be set to 5%, 10%, or 15%, etc. This application embodiment does not limit the specific value of the error threshold.

[0104] In some embodiments, for ancillary products in a newly added product category, a separate expert network can be trained for each ancillary product in that category. For example, an independent training sample set can be constructed for each ancillary product in the newly added product category. This independent training sample set can include multiple independent training samples. Each independent virtual sample includes historical user interaction behavior features during the flight booking process, historical user profile features, flight scenario features corresponding to the flight booked by the historical user, and historical user purchase tags for the ancillary product in the newly added product category. Then, the initial expert network can be trained based on this independent training sample set to obtain a trained expert network, which can then be connected to the expert network layer of the original MMOE model to complete the lightweight expansion and adaptation of the model.

[0105] The foregoing primarily describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the aforementioned functions, each device, such as an ancillary product packaging device, includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the algorithmic steps of the examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Experts may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0106] This application embodiment can divide the ancillary product packaging device into functional modules according to the above method embodiment. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one functional module. The integrated module can be implemented in hardware or software. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods. The following description uses the example of dividing each functional module according to each function.

[0107] In an exemplary embodiment, this application also provides a dynamic packaging device for airline tickets and ancillary products. Figure 5 This is a schematic diagram illustrating the composition of the dynamic packaging device for airline tickets and ancillary products provided in an embodiment of this application. Figure 5 As shown, the dynamic packaging device for airline tickets and ancillary products may include: an acquisition module 501, a processing module 502, and a sending module 503.

[0108] The acquisition module 501 is used to acquire the user's interactive behavior characteristics during the flight booking process.

[0109] Processing module 502 is used to construct a multi-dimensional feature vector based on interaction behavior characteristics, user profile characteristics, and flight scenario characteristics of the flight corresponding to the user's booked flight; input the multi-dimensional feature vector into a multi-gate hybrid expert MMOE model to obtain the conversion probability of the user for each of the multiple ancillary products predicted by the MMOE model; the MMOE model includes multiple expert networks, each expert network corresponding to a category of ancillary products, and each expert network is used to predict the conversion probability of the corresponding category of ancillary products based on the multi-dimensional feature vector; construct a correlation matrix of multiple categories of ancillary products based on historical order data of multiple categories of ancillary products; calculate the correlation coefficient between different categories of ancillary products based on the correlation matrix; determine the package score of multiple ancillary product combinations based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product; and determine the target ancillary product combination with the highest package score from multiple ancillary product combinations; The sending module 503 is used to send a display instruction for the target ancillary product combination to the terminal device operated by the user, so that the terminal device can display the target ancillary product combination on the display interface of the flight booking process.

[0110] In some possible embodiments, the interactive behavior characteristics include at least one of the following: price comparison behavior characteristics, time attention characteristics, model attention characteristics, comfort preference characteristics, paid seat selection preference characteristics, seat selection preference characteristics, and decision time characteristics.

[0111] In some possible embodiments, user profile features include at least one of the following: membership level features, accumulated mileage features, membership duration features, time sensitivity features, price sensitivity features, travel structure features, refund and change preference features, baggage preference features, seat selection preference features, and insurance preference features.

[0112] In some possible embodiments, flight scenario features include at least one of the following: red-eye flight features, remaining seat features, on-time performance features, and weather features.

[0113] In some possible embodiments, processing module 502 is specifically used to calculate the packaging score according to the following formula: ; in, Indicates ancillary products and ancillary products Package score of the ancillary product portfolio; Indicates ancillary products Product profitability coefficient; Indicates ancillary products Product profitability coefficient; Indicates ancillary products and ancillary products The probability of combinational transformation; ; in, Indicates ancillary products The conversion probability; Indicates ancillary products The conversion probability; Indicates the first Ancillary products and the first Correlation coefficients between ancillary products; ancillary products Belongs to the Ancillary products; Ancillary products Belongs to the Ancillary products.

[0114] In some possible embodiments, the acquisition module 501 is further configured to acquire a training sample set; the training sample set includes multiple training samples, each training sample including the interaction behavior characteristics of historical users in the flight booking process, the user profile characteristics of historical users, the flight scenario characteristics of the flight corresponding to the flight booked by historical users, and the purchase tags of historical users for various categories of ancillary products; the processing module 502 is further configured to perform multi-task training on the initial MMOE model based on the training sample set to obtain the trained MMOE model.

[0115] It should be noted that the above Figure 5 Modules in a module can also be called units; for example, a processing module can be called a processing unit. Additionally, in... Figure 5 In the embodiments shown, the names of the modules may not be the same as those shown in the figure. For example, the acquisition module may also be called the transceiver module or the communication module.

[0116] Figure 5 If each unit in the process is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a mobile phone, personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The storage medium for storing computer software products includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0117] In an exemplary embodiment, this application also provides an electronic device. Figure 6 This is a schematic diagram illustrating the composition of an electronic device provided in an embodiment of this application. For example... Figure 6 As shown, the electronic device includes a processor 602, a communication interface 603, and a bus 604. As an example, the electronic device may also include a memory 601.

[0118] Processor 602 may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 602 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 602 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0119] Communication interface 603 is used to connect to other devices via a communication network. This communication network can be Ethernet, wireless access network, wireless local area network (WLAN), etc.

[0120] The memory 601 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0121] As one possible implementation, the memory 601 can exist independently of the processor 602. The memory 601 can be connected to the processor 602 via a bus 604 and is used to store instructions or program code. When the processor 602 calls and executes the instructions or program code stored in the memory 601, it can implement the dynamic packaging method for airline tickets and ancillary products provided in this application embodiment.

[0122] In another possible implementation, the memory 601 can also be integrated with the processor 602.

[0123] Bus 604 can be an extended industry standard architecture (EISA) bus, etc. Bus 604 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0124] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the auxiliary product packaging device can be divided into different functional modules to complete all or part of the functions described above.

[0125] In exemplary embodiments, this application also provides a computer-readable storage medium including computer software instructions that, when executed in an electronic device, cause the electronic device to implement the methods described in the above embodiments. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device. Further, the computer-readable storage medium can include both internal storage units and external storage devices of the electronic device. The computer-readable storage medium is used to store the software instructions and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0126] In an exemplary embodiment, this application also provides a computer program product including computer instructions that, when executed on an electronic device, cause the electronic device to perform the methods described in the above method embodiments.

[0127] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer-executable instructions. When these computer-executable instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer-executable instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer-executable instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, Bluetooth, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape) or an optical medium (e.g., DVD), etc.

[0128] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0129] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

[0130] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamic packaging of airline tickets with ancillary products, characterized in that, The method includes: Obtain user interaction behavior characteristics during the flight booking process; Based on the interactive behavior features, the user profile features, and the flight scenario features of the flight corresponding to the user's booked ticket, a multi-dimensional feature vector is constructed; The multi-dimensional feature vector is input into a multi-gate hybrid expert MMOE model to obtain the conversion probability of the user for each of the multiple ancillary products predicted by the MMOE model; the MMOE model includes multiple expert networks, each expert network corresponds to a category of ancillary products, and each expert network is used to predict the conversion probability of the corresponding category of ancillary products based on the multi-dimensional feature vector. Based on historical order data of various ancillary products, a correlation matrix of various ancillary products is constructed. Based on the correlation matrix, the correlation coefficients between different categories of ancillary products are calculated; Based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product, the package score of multiple ancillary product combinations is determined. From the plurality of ancillary product portfolios, determine the target ancillary product portfolio with the highest packaging score; Send a display instruction for the target ancillary product combination to the terminal device operated by the user, so that the terminal device displays the target ancillary product combination on the display interface of the flight booking process.

2. The method according to claim 1, characterized in that, The interactive behavior characteristics include at least one of the following: price comparison behavior characteristics, time attention characteristics, model attention characteristics, comfort preference characteristics, paid seat selection preference characteristics, seat selection preference characteristics, and decision-making time characteristics.

3. The method according to claim 1, characterized in that, The user profile features include at least one of the following: membership level features, accumulated mileage features, membership duration features, time sensitivity features, price sensitivity features, travel structure features, refund and change preference features, baggage preference features, seat selection preference features, and insurance preference features.

4. The method according to claim 1, characterized in that, The flight scenario features include at least one of the following: red-eye flight features, remaining seat features, on-time performance features, and weather features.

5. The method according to claim 1, characterized in that, The determination of the packaged score for multiple ancillary product combinations based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product includes: The packaging score is calculated using the following formula: ； in, Indicates ancillary products and ancillary products Package score of the ancillary product portfolio; Indicates ancillary products Product profitability coefficient; Indicates ancillary products Product profitability coefficient; Indicates ancillary products and ancillary products The probability of combinational transformation; ； in, Indicates ancillary products The conversion probability; Indicates ancillary products The conversion probability; Indicates the first Ancillary products and the first Correlation coefficients between ancillary products; ancillary products Belongs to the Ancillary products; Ancillary products Belongs to the Ancillary products.

6. The method according to claim 1, characterized in that, The MMOE model was trained in the following way: Obtain a training sample set; the training sample set includes multiple training samples, each training sample including the interaction behavior characteristics of historical users in the flight booking process, the user profile characteristics of the historical users, the flight scenario characteristics of the flight corresponding to the flight booked by the historical users, and the purchase tags of the historical users for each category of ancillary products. The initial MMOE model is trained using the training sample set to obtain the trained MMOE model.

7. A dynamic packaging device for airline tickets and ancillary products, characterized in that, include: The module includes an acquisition module, a processing module, and a sending module. The acquisition module is used to acquire the user's interactive behavior characteristics during the flight booking process; The processing module is configured to: construct a multi-dimensional feature vector based on the interaction behavior features, the user profile features, and the flight scenario features corresponding to the flight booked by the user; input the multi-dimensional feature vector into a multi-gate hybrid expert MMOE model to obtain the conversion probability of the user for each of the multiple ancillary products predicted by the MMOE model; the MMOE model includes multiple expert networks, each expert network corresponding to a category of ancillary products, and each expert network is used to predict the conversion probability of the corresponding category of ancillary products based on the multi-dimensional feature vector; construct a correlation matrix for multiple categories of ancillary products based on historical order data of multiple categories of ancillary products; calculate the correlation coefficient between different categories of ancillary products based on the correlation matrix; determine the package score of multiple ancillary product combinations based on the conversion probability of each ancillary product, the correlation coefficient between different categories of ancillary products, and the product revenue coefficient of each ancillary product; and determine the target ancillary product combination with the highest package score from the multiple ancillary product combinations. The sending module is used to send a display instruction for the target ancillary product combination to the terminal device operated by the user, so that the terminal device displays the target ancillary product combination on the display interface of the flight booking process.

8. An electronic device, characterized in that, include: Memory and processor; The memory stores instructions that the processor can execute; When the processor is configured to execute the instructions, the electronic device performs the method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, include: Computer software instructions; When the computer software instructions are executed in an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-6.

10. A computer program product, characterized in that, include: Computer instructions; When the computer instructions are executed in an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-6.

Images