Deep learning-based cabin rising service recommendation method and apparatus, and electronic device

Abstract

The invention discloses a deep learning-based cabin rising service recommendation method and device and electronic equipment, and relates to the field of artificial intelligence, and the method comprises the steps: obtaining the user portrait information of a to-be-tripped user, the flight information of a target flight, and the cabin rising service information; pre-processing the data to obtain a user portrait feature vector, a flight feature vector and a rising service feature vector; inputting the user portrait feature vector, the flight feature vector and the cabin rising service feature vector into a cabin rising probability prediction model for any user waiting to go out, and outputting the cabin rising probability of the user waiting to go out; sorting the cabin rising probabilities of the N users to go out to obtain a sorting result; and selecting M to-be-tripped users from the sorting result based on the cabin rising shipping space of the target flight to recommend the cabin rising service of the target flight. According to the method and the device, the technical problem that accurate cabin rising recommendation is difficult to carry out on passengers in a cabin rising recommendation mode of carrying out batch recommendation through full-quantity recommendation or manual passenger circle selection in related technologies is solved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method and device for recommending a cabin upgrade service based on deep learning, and an electronic device. Background Art

[0002] With the growth of travel demand and the intensification of competition in the aviation market, airlines are actively seeking innovative sales methods to improve their product competitiveness and customer satisfaction. In this context, upgrade services have become one of the important strategies for airlines to promote sales and customer relationship development. By issuing upgrade services, airlines can motivate customers to upgrade seats and provide a higher level of service and comfort experience. This can not only increase the revenue of airlines, but also improve customer satisfaction. By issuing upgrade coupons through official channels or third-party channels, airlines can attract more customers to choose to upgrade, thereby increasing ticket sales and increasing upgrade rates. It brings new opportunities for airlines and provides passengers with more choices and upgrade opportunities.

[0003] Multi-scenario upgrade services can influence flight occupancy and airline profits through different pricing strategies. Lower upgrade voucher prices can attract more customers to choose to upgrade, thereby increasing flight occupancy and improving airline revenue. Higher upgrade voucher prices can increase airline profits, especially for high-demand routes or busy travel seasons. At the same time, this multi-scenario pricing strategy can also better serve customers. Lower-priced upgrade vouchers can allow more customers to enjoy the opportunity to upgrade, providing better comfort and services. Higher-priced upgrade vouchers provide customers who are willing to pay a higher price with higher-level seats and more luxurious service experience. Airlines can flexibly adjust the price of upgrade vouchers according to the actual situation of the flight and customer needs to increase occupancy, increase profits and optimize flight resource utilization, while providing customers with a smooth transition between economy class and business class, allowing customers to enjoy more comfortable seats and services, and improve customer satisfaction.

[0004] In the related technology, upgrade services can be recommended to passengers through full recommendation or batch recommendation of manually selected customer groups. This full recommendation or manual selection method is difficult to achieve accurate positioning. Inaccurate recommendation methods may cause adverse effects on customers such as information interference, difficulty in purchasing decisions, and unsatisfactory upgrade experience. In severe cases, it may lead to customer loss of airlines.

[0005] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiments of the present invention provide a method and device for recommending a class upgrade service based on deep learning, and an electronic device, so as to at least solve the technical problem in the related art that it is difficult to make accurate class upgrade recommendations to passengers by recommending all or manually selecting passengers for batch recommendation.

[0007] According to one aspect of an embodiment of the present invention, a method for recommending an upgrade service based on deep learning is provided, comprising: obtaining user portrait information of N users to be transported corresponding to a target flight, flight information of the target flight, and upgrade service information, wherein N is a positive integer; preprocessing the user portrait information of the users to be transported, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the users to be transported, a flight feature vector of the target flight, and an upgrade service feature vector; for any of the users to be transported, the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector are preprocessed; The vector is input into an upgrade probability prediction model, and the upgrade probability of the user to be traveling for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be traveling, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; the upgrade probabilities of N users to be traveling on the target flight are sorted to obtain a sorting result; based on the M upgrade seats of the target flight, M users to be traveling are selected from the sorting result as target upgrade users, and the upgrade service of the target flight is recommended to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0008] Optionally, the step of inputting the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into the upgrade probability prediction model, and outputting the upgrade probability of the user to be transported for the target flight comprises: inputting the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into a linear model in the upgrade probability prediction model, and outputting a linear feature vector through the linear model, wherein the linear feature vector is used to characterize the linear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector; and , the flight feature vector and the upgrade service feature vector of the target flight are input into a nonlinear model in the upgrade probability prediction model, and a nonlinear feature vector is output through the nonlinear model, wherein the nonlinear feature vector is used to characterize the nonlinear relationship between the features corresponding to the user portrait feature vector, the flight feature vector and the upgrade service feature vector, and the complexity of the linear relationship is greater than the complexity of the nonlinear relationship; weighted calculation is performed on the linear feature vector and the nonlinear feature vector to obtain a calculation result; the upgrade probability is calculated based on the calculation result and an activation function, and the upgrade probability of the user to be traveling for the target flight is output through the upgrade probability prediction model.

[0009] Optionally, the steps of pre-processing the user portrait information of the user to travel, the flight information of the target flight, and the upgrade service information to obtain the user portrait feature vector of the user to travel, the flight feature vector of the target flight, and the upgrade service feature vector include: performing data cleaning processing on the user portrait information of the user to travel, the flight information of the target flight, and the upgrade service information to obtain the processed user portrait information, the flight information, and the upgrade service information; performing feature extraction on the processed user portrait information, the flight information of the target flight, and the upgrade service information based on preset feature extraction rules to obtain user portrait feature data of the user to travel, flight feature data of the target flight, and upgrade service feature data; encoding the user portrait feature data of the user to travel, the flight feature data of the target flight, and the upgrade service feature data based on preset feature encoding rules to obtain the user portrait feature vector of the user to travel, the flight feature vector of the target flight, and the upgrade service feature vector.

[0010] Optionally, the user portrait feature data includes user attribute feature data and user behavior feature data, wherein the user attribute feature data includes at least one of the following: user gender feature data, user age feature data, user occupation feature data, user residence feature information, user level feature information; the user behavior feature data includes at least one of the following: user historical air travel feature data, user historical ticket purchase feature data, user historical upgrade feature data.

[0011] Optionally, the flight characteristic data of the target flight includes at least one of the following: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, season characteristic data, date characteristic data, characteristic data of the take-off time of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, and characteristic data of the mileage of the flight segments of the target flight.

[0012] Optionally, the upgrade service characteristic data includes at least one of the following: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, and upgrade location characteristic data.

[0013] Optionally, the step of constructing the upgrade probability prediction model includes: obtaining user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users; configuring upgrade probability labels for the historical travel users based on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of the historical travel users; performing feature processing and feature encoding on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of the historical travel users to obtain a historical feature vector of the historical travel users, wherein the historical feature vector includes: a historical user portrait feature vector, a historical flight feature vector, a historical upgrade service feature vector, and historical lines between the historical features. The method comprises the steps of: constructing a sample set based on the historical feature vectors of the historical travel users and the upgrade probability labels, and dividing the sample set to obtain a training set and a test set; jointly training an initial linear model and an initial nonlinear model based on the training set to obtain an initial upgrade probability prediction model, wherein the joint training is used to indicate that the initial linear model and the initial nonlinear model use the same training set and the same loss function for iteration during the training process; testing the initial upgrade probability prediction model based on the test set to obtain a test result, and obtaining the trained upgrade probability prediction model when the test result indicates that the initial upgrade probability prediction model passes the test.

[0014] According to another aspect of an embodiment of the present invention, there is further provided an upgrade service recommendation device based on deep learning, comprising: an acquisition unit, configured to acquire user portrait information of N users to be transported corresponding to a target flight, flight information of the target flight, and upgrade service information, wherein N is a positive integer; a preprocessing unit, configured to preprocess the user portrait information of the users to be transported, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the users to be transported, a flight feature vector of the target flight, and an upgrade service feature vector; an output unit, configured to output the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector to any of the users to be transported. The upgrade probability prediction model is used to input the upgrade service feature vector into the upgrade probability prediction model, and output the upgrade probability of the user to be transported for the target flight, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be transported, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; a sorting unit is used to sort the upgrade probabilities of the N users to be transported on the target flight to obtain a sorting result; a recommendation unit is used to select M users to be transported from the sorting result as target upgrade users based on the M upgrade cabins of the target flight, and recommend the upgrade service of the target flight to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0015] Optionally, the output unit includes: a first output module, used to input the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into a linear model in the upgrade probability prediction model, and output a linear feature vector through the linear model, wherein the linear feature vector is used to characterize the linear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector; a second output module, used to input the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into the upgrade probability prediction model. A nonlinear model in the rate prediction model is used to output a nonlinear feature vector through the nonlinear model, wherein the nonlinear feature vector is used to characterize the nonlinear relationship between the features corresponding to the user portrait feature vector, the flight feature vector and the upgrade service feature vector, and the complexity of the linear relationship is greater than the complexity of the nonlinear relationship; a first calculation module is used to perform weighted calculation on the linear feature vector and the nonlinear feature vector to obtain a calculation result; a third output module is used to calculate the upgrade probability based on the calculation result and an activation function, and output the upgrade probability of the user to be traveling for the target flight through the upgrade probability prediction model.

[0016] Optionally, the preprocessing unit includes: a first cleaning module, used to perform data cleaning processing on the user portrait information of the user to be traveled, the flight information of the target flight, and the upgrade service information to obtain the processed user portrait information, the flight information, and the upgrade service information; a first extraction module, used to perform feature extraction on the processed user portrait information, the flight information of the target flight, and the upgrade service information based on preset feature extraction rules to obtain user portrait feature data of the user to be traveled, flight feature data of the target flight, and upgrade service feature data; a first encoding module, used to encode the user portrait feature data of the user to be traveled, the flight feature data of the target flight, and the upgrade service feature data based on preset feature encoding rules to obtain the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector.

[0017] Optionally, the user portrait feature data includes user attribute feature data and user behavior feature data, wherein the user attribute feature data includes at least one of the following: user gender feature data, user age feature data, user occupation feature data, user residence feature information, user level feature information; the user behavior feature data includes at least one of the following: user historical air travel feature data, user historical ticket purchase feature data, user historical upgrade feature data.

[0018] Optionally, the flight characteristic data of the target flight includes at least one of the following: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, season characteristic data, date characteristic data, characteristic data of the take-off time of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, and characteristic data of the mileage of the flight segments of the target flight.

[0019] Optionally, the upgrade service characteristic data includes at least one of the following: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, and upgrade location characteristic data.

[0020] Optionally, the deep learning-based upgrade service recommendation device also includes: a first acquisition module, used to acquire user portrait information, historical flight information and historical upgrade service information corresponding to the historical flights of historical travel users; a first configuration module, used to configure an upgrade probability label for the historical travel user based on the user portrait information, historical flight information and historical upgrade service information corresponding to the historical flights of the historical travel user; a second acquisition module, used to perform feature processing and feature encoding on the user portrait information, historical flight information and historical upgrade service information corresponding to the historical flights of the historical travel user to obtain a historical feature vector of the historical travel user, the historical feature vector including: a historical user portrait feature vector, a historical flight feature vector, a historical upgrade service feature vector and historical relationships between the historical features. a linear feature vector and a historical nonlinear feature vector; a first construction module, used to construct a sample set based on the historical feature vectors of the historical travel users and the upgrade probability labels, and divide the sample set to obtain a training set and a test set; a first training module, used to jointly train the initial linear model and the initial nonlinear model based on the training set to obtain an initial upgrade probability prediction model, wherein the joint training is used to indicate that the initial linear model and the initial nonlinear model use the same training set and the same loss function for iteration during the training process; a first testing module, used to test the initial upgrade probability prediction model based on the test set to obtain a test result, and when the test result indicates that the initial upgrade probability prediction model passes the test, the trained upgrade probability prediction model is obtained.

[0021] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned deep learning-based upgrade service recommendation methods.

[0022] According to another aspect of an embodiment of the present invention, an electronic device is further provided, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned deep learning-based upgrade service recommendation methods.

[0023] In the present application, the following steps are performed: first, user portrait information of N users to be transported corresponding to the target flight, flight information of the target flight, and upgrade service information are obtained, wherein N is a positive integer; then, the user portrait information of the users to be transported, the flight information of the target flight, and the upgrade service information are preprocessed to obtain the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector; and for any user to be transported, the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector are input into an upgrade probability prediction model, and the upgrade probability of the user to be transported for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the users to be transported, and the upgrade probability prediction model includes a linear model and a nonlinear model; the upgrade probabilities of the N users to be transported for the target flight are sorted to obtain a sorting result; finally, based on the M upgrade cabins of the target flight, M users to be transported are selected from the sorting result as target upgrade users, and the upgrade service of the target flight is recommended to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0024] In the present application, the flight information, upgrade service information and user portrait information of the current flight corresponding to the traveling user are obtained, and the flight information, upgrade service information and user portrait information are analyzed and calculated through a pre-built upgrade probability prediction model to predict the user's upgrade probability. Users with a higher upgrade probability are selected as target upgrade users based on the upgrade cabin space of the target flight, and upgrade services are accurately recommended to the target upgrade users to improve user satisfaction, while avoiding information interference to users who do not need upgrades, and achieving accurate recommendation of upgrade services, which is convenient for airlines to better maintain customer sources, thereby solving the technical problem in the related art that it is difficult to make accurate upgrade recommendations for passengers through full recommendation or manual selection of passengers for batch recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0026] Figure 1 is a flowchart of an optional deep learning-based upgrade service recommendation method according to an embodiment of the present invention;

[0027] Figure 2 is a schematic diagram of an optional deep learning-based upgrade service recommendation process according to an embodiment of the present invention;

[0028] Figure 3is a schematic diagram of a construction process of an optional upgrade probability prediction model according to an embodiment of the present invention;

[0029] Figure 4 is a schematic diagram of an optional deep learning-based upgrade service recommendation device according to an embodiment of the present invention;

[0030] Figure 5 This is a hardware structure block diagram of an electronic device (or mobile device) according to a method for recommending a class upgrade service based on deep learning according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] To facilitate those skilled in the art to understand the present invention, some terms or nouns involved in the embodiments of the present invention are explained below:

[0034] DataX, an offline data synchronization tool, is used to efficiently, stably and reliably transfer large amounts of data between different data storage systems.

[0035] Flink, a stream processing framework, is used for processing and analyzing real-time data streams.

[0036] Kafka, a stream processing platform, is mainly used to build real-time data pipelines and stream processing applications.

[0037] CDH, the full name of Cloudera Distr ibut ion Hadoop, includes the core components of Apache Hadoop and many other related data storage and analysis tools.

[0038] Hudi, or Apache Hudi, is the full name of Hadoop Upserts and Incremental Processing, a storage and processing framework for managing large analytical data sets in the Apache Hadoop ecosystem. It aims to improve the write performance and query efficiency of the data lake, while supporting a variety of data updates and incremental processing operations.

[0039] Spark, a distributed big data processing framework, is used to handle complex data processing tasks on large data sets.

[0040] HDFS, the full name of Hadoop Distributed File System, is a distributed file system used to store and process large amounts of data.

[0041] It should be noted that the deep learning-based upgrade service recommendation method and device in the present application can be used in the field of artificial intelligence. When recommending upgrade services to users based on deep learning technology, it can also be used in any field other than the field of artificial intelligence. When recommending upgrade services to users based on deep learning technology, the application field of the deep learning-based upgrade service recommendation method and device is not limited in the present application.

[0042] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good customs, and provide corresponding operation entrances for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0043] It should be noted that in this application, when collecting and analyzing customer information, corresponding operation entrances are provided for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered.

[0044] The following embodiments of the present invention can be applied to various upgrade service recommendation systems / applications / devices based on deep learning. The present invention uses deep learning technology to process and analyze a large amount of flight information, user profile information and upgrade service information, and calculates the user's upgrade probability based on the upgrade service information as well as the flight information and user profile information, and then selects users with a higher probability of upgrade for accurate recommendation. The upgrade probability prediction model built based on deep learning technology includes two parts: a linear model and a nonlinear model. When processing relevant information, the cross-relationship and nonlinear relationship between features can be considered simultaneously, thereby improving the accuracy and precision of the recommendation results.

[0045] The present invention is described in detail below in conjunction with various embodiments.

[0046] Embodiment 1

[0047] According to an embodiment of the present invention, an embodiment of a method for recommending a class upgrade service based on deep learning is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0048] Figure 1 is a flowchart of an optional deep learning-based upgrade service recommendation method according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0049] Step S101, obtaining user portrait information of N users to travel corresponding to a target flight, flight information of the target flight, and upgrade service information, where N is a positive integer;

[0050] Step S102, pre-processing the user portrait information of the user to be traveled, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the user to be traveled, a flight feature vector of the target flight, and an upgrade service feature vector;

[0051] Step S103: for any user to travel, input the user portrait feature vector of the user to travel, the flight feature vector of the target flight, and the upgrade service feature vector into the upgrade probability prediction model, and output the upgrade probability of the user to travel for the target flight, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to travel, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model;

[0052] Step S104, sorting the upgrade probabilities of N users waiting to travel on the target flight to obtain a sorting result;

[0053] Step S105 , based on the M upgrade seats of the target flight, M users to be transported are selected from the sorting results as target upgrade users, and upgrade services of the target flight are recommended to the target upgrade users, where M is a positive integer and N is greater than M.

[0054] Through the above steps, firstly, user portrait information of N users to be transported corresponding to the target flight, flight information of the target flight and upgrade service information are obtained, wherein N is a positive integer, and then the user portrait information of the users to be transported, the flight information of the target flight and the upgrade service information are preprocessed to obtain the user portrait feature vector of the users to be transported, the flight feature vector of the target flight and the upgrade service feature vector, and for any user to be transported, the user portrait feature vector of the users to be transported, the flight feature vector of the target flight and the upgrade service feature vector are input into the upgrade probability prediction model, and the upgrade probability of the users to be transported for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the users to be transported, and the upgrade probability prediction model includes a linear model and a nonlinear model, and the upgrade probabilities of the N users to be transported for the target flight are sorted to obtain a sorting result, and finally, based on the M upgrade cabins of the target flight, M users to be transported are selected from the sorting result as target upgrade users, and the upgrade service of the target flight is recommended to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0055] In this embodiment, flight information, upgrade service information of the current flight, and user portrait information of the user to be traveled corresponding to the current flight are obtained, and the flight information, upgrade service information and user portrait information are analyzed and calculated through a pre-built upgrade probability prediction model to predict the user's upgrade probability. Users with a higher upgrade probability are selected as target upgrade users based on the upgrade cabin space of the target flight, and upgrade services are accurately recommended to the target upgrade users to improve user satisfaction, while avoiding information interference to users who do not need upgrades, and achieving accurate recommendation of upgrade services, which is convenient for airlines to better maintain customer sources, thereby solving the technical problem in the related art that it is difficult to make accurate upgrade recommendations for passengers through full recommendation or manually selecting passengers for batch recommendation.

[0056] The embodiment of the present invention is described in detail below in combination with the above steps.

[0057] At present, air travel has become the most convenient mode of travel, which can save users a lot of time and become the first choice for users. In order to provide users with a more comfortable riding experience, flights generally have more comfortable cabins such as first class. When the number of passengers in high-class cabins is not full, upgrade services can be recommended to users to improve user satisfaction. At the same time, it can also bring higher profits to airlines and achieve a win-win situation between airlines and users.

[0058] It should be noted that the present invention predicts the upgrade probability of users waiting to travel on the target flight based on a pre-built upgrade probability prediction model. The upgrade probability prediction model includes two parts: a linear model and a nonlinear model. It can be used to explore the cross-relationships and nonlinear relationships between features, thereby achieving accurate predictions, improving the accuracy of prediction results, and further accurately recommending upgrade services to passengers and improving passenger satisfaction.

[0059] Step S101, obtaining user portrait information of N users to travel corresponding to the target flight, flight information of the target flight, and upgrade service information.

[0060] It should be noted that the target flight is a flight that can currently be recommended for upgrade services. Before the flight takes off, the upgrade service will be processed for passengers taking the flight based on the remaining upgrade seats of the flight. In order to achieve accurate recommendations for passengers, the embodiment of the present invention first needs to obtain user portrait information of users to be traveling, flight information of the target flight, and upgrade service information. The users to be traveling can be all passengers on the target flight this time, or they can be users traveling in economy class, business class, etc. on the target flight.

[0061] Specifically, user portrait information is the user file or user information built by airlines for passengers based on big data technology. It is a digital description of user characteristics, which may cover basic attribute information such as the user's name, gender, age, occupation, education level, geographical information such as the user's place of residence, workplace, and common activity areas, and multi-dimensional information such as the user's travel behavior information and travel habits information based on historical travel records. By analyzing user portrait information, multiple features can be obtained to calculate the user's upgrade probability.

[0062] It should be noted that the flight information of the target flight may include: information about the airline to which the flight belongs, information about the flight model, current season information, holiday information, flight departure time, flight departure airport code, flight destination airport code, flight class information, flight price information, flight delay rate data, flight occupancy rate data, flight segment mileage, price per kilometer data and other information; the upgrade service information may include: upgrade class information, upgrade price information, upgrade discount information and other information.

[0063] Step S102, pre-processing the user portrait information of the user to be traveled, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the user to be traveled, a flight feature vector of the target flight, and an upgrade service feature vector.

[0064] Optionally, the steps of preprocessing the user portrait information of the user to travel, the flight information of the target flight, and the upgrade service information to obtain the user portrait feature vector of the user to travel, the flight feature vector of the target flight, and the upgrade service feature vector include: performing data cleaning processing on the user portrait information of the user to travel, the flight information of the target flight, and the upgrade service information to obtain processed user portrait information, flight information, and upgrade service information; performing feature extraction on the processed user portrait information, the flight information of the target flight, and the upgrade service information based on preset feature extraction rules to obtain user portrait feature data of the user to travel, flight feature data of the target flight, and upgrade service feature data; encoding the user portrait feature data of the user to travel, the flight feature data of the target flight, and the upgrade service feature data based on preset feature encoding rules to obtain the user portrait feature vector of the user to travel, the flight feature vector of the target flight, and the upgrade service feature vector.

[0065] It should be noted that after obtaining the relevant information of the user to be traveled, it is necessary to pre-process the user portrait information, flight information of the target flight, and upgrade service information in the relevant information. The pre-processing operation first includes cleaning the data, filling in missing data values, or correcting erroneous data, etc., so as to obtain cleaned user portrait information, flight information, and upgrade service information. Then, according to the preset feature extraction rules, feature extraction is performed on the cleaned relevant information to extract important features in the information, and then the features are encoded according to the preset encoding rules to obtain the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector, so as to facilitate the upgrade probability model to analyze and calculate the feature vector. The feature encoding rules are pre-constructed, and the extracted feature data can be directly converted into feature codes, such as converting the residence characteristics of the user to be travel into the area code of the region to which the residence belongs, converting the characteristics of the airline to which the target flight belongs into the digital number of the airline, etc., so as to organize the original text information into a format suitable for model input.

[0066] Optionally, the user portrait feature data includes user attribute feature data and user behavior feature data, wherein the user attribute feature data includes at least one of the following: user gender feature data, user age feature data, user occupation feature data, user residence feature information, user level feature information; the user behavior feature data includes at least one of the following: user historical air travel feature data, user historical ticket purchase feature data, user historical upgrade feature data.

[0067] It should be noted that user portrait feature data is feature data extracted based on user portrait information, which is used to characterize the important characteristics of users. It is specifically divided into user attribute feature data and user behavior feature data. User attribute feature data is obtained by extracting features based on the user's basic attribute information, and may include user gender feature data, user age feature data, user occupation feature data, user residence feature information, user level feature information, etc. User behavior feature data is feature data extracted based on the user's air travel behavior, and the user behavior feature data may include: user's historical air travel feature data, such as the flights taken by the user in historical air travel, flight time, the airline to which the flight belongs, etc., user's historical ticket purchase feature data, such as the flight price and flight class purchased by the user in historical air travel, etc., and user's historical upgrade feature data, such as the number of upgrades and upgraded class during the user's historical air travel, etc.

[0068] Optionally, the flight characteristic data of the target flight includes at least one of the following: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, season characteristic data, date characteristic data, characteristic data of the take-off time of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, and characteristic data of the mileage of the flight segments of the target flight.

[0069] It should be noted that the flight characteristic data of the target flight is obtained by extracting features based on the flight information of the target flight, and may include: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, seasonal characteristic data, date characteristic data, characteristic data of the take-off period of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, characteristic data of the mileage of the flight segments of the target flight, etc.

[0070] Optionally, the upgrade service characteristic data includes at least one of the following: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, and upgrade location characteristic data.

[0071] It should be noted that the upgrade service characteristic data is characteristic data extracted based on the upgrade service information of the target flight, and may specifically include: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, upgrade location characteristic data, etc.

[0072] Step S103: for any user to travel, the user portrait feature vector of the user to travel, the flight feature vector of the target flight and the upgrade service feature vector are input into the upgrade probability prediction model, and the upgrade probability of the user to travel for the target flight is output.

[0073] It should be noted that after encoding the relevant information, the user portrait feature vector of each user to travel, the flight feature vector of the target flight, and the upgrade service feature vector are input one by one into the upgrade probability prediction model. The linear and nonlinear relationships between the various features are mined through the upgrade probability prediction model, and then the upgrade probability of each user to travel is calculated based on the linear and nonlinear relationships, and the upgrade service is accurately recommended to passengers based on the upgrade probability.

[0074] It should be noted that the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of traveling users. The upgrade probability prediction model includes two parts: a linear model and a nonlinear model.

[0075] Optionally, the step of constructing the upgrade probability prediction model includes: obtaining user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users; configuring upgrade probability labels for historical travel users based on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users; performing feature processing and feature encoding on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users to obtain historical feature vectors of historical travel users, the historical feature vectors including: historical user portrait feature vectors, historical flight feature vectors, historical upgrade service feature vectors, and historical relationships between the historical features. historical linear feature vectors and historical nonlinear feature vectors; construct a sample set based on the historical feature vectors and upgrade probability labels of historical travel users, and divide the sample set into a training set and a test set; jointly train the initial linear model and the initial nonlinear model based on the training set to obtain an initial upgrade probability prediction model, wherein the joint training is used to indicate that the initial linear model and the initial nonlinear model use the same training set and the same loss function for iteration during the training process; test the initial upgrade probability prediction model based on the test set to obtain a test result, and when the test result indicates that the initial upgrade probability prediction model passes the test, a trained upgrade probability prediction model is obtained.

[0076] Specifically, when constructing an upgrade probability model, it is first necessary to construct a sample set, obtain user portrait information of historical travel users, historical flight information involved in historical travel users and upgrade service information corresponding to the historical flights from multiple information acquisition channels, and configure an upgrade probability label for the user based on the user's historical travel behavior and other data, and perform feature processing and feature encoding on the user portrait information, historical flight information and historical upgrade service information corresponding to the historical flights of the historical travel user to obtain a historical feature vector of the historical travel user, concatenate the historical feature vector and the configured upgrade probability label, construct a sample set for model training, and divide the sample set according to a pre-set division ratio to obtain a training set for user model training and a test set for model testing, and jointly train the initial linear model and the initial nonlinear model through the test set. During the joint training process, the initial linear model and the initial nonlinear model are iterated using the same loss function. After the training is completed to obtain the initial upgrade probability prediction model, the upgrade probability prediction model is tested through the test set. If the test passes, the trained upgrade probability prediction model is obtained.

[0077] Optionally, the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector are input into the upgrade probability prediction model, and the step of outputting the upgrade probability of the user to be traveled for the target flight includes: inputting the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector into a linear model in the upgrade probability prediction model, and outputting a linear feature vector through the linear model, wherein the linear feature vector is used to characterize the linear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector; inputting the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector into a nonlinear model in the upgrade probability prediction model, and outputting a nonlinear feature vector through the nonlinear model, wherein the nonlinear feature vector is used to characterize the nonlinear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector, and the complexity of the linear relationship is greater than the complexity of the nonlinear relationship; performing weighted calculation on the linear feature vector and the nonlinear feature vector to obtain a calculation result; calculating the upgrade probability based on the calculation result and the activation function, and outputting the upgrade probability of the user to be travel for the target flight through the upgrade probability prediction model.

[0078] It should be noted that when the upgrade probability of the user to be traveling is calculated and predicted by the upgrade probability prediction model, the user portrait feature vector of the user to be traveling, the flight feature vector of the target flight and the upgrade service feature vector are simultaneously input into the linear model and the nonlinear model in the upgrade probability prediction model, and the linear model outputs a linear feature vector, which is used to characterize the linear relationship between various features. For example, the flight duration feature is obtained based on the flight departure time and the flight arrival time, and the nonlinear model outputs a nonlinear feature vector, which is used to characterize nonlinear relationships that are more complex than feature linear relationships, such as the impact of seasons on user upgrades, etc.

[0079] Step S104, sorting the upgrade probabilities of the N users waiting to travel on the target flight to obtain a sorting result.

[0080] It should be noted that after calculating the upgrade probability of each user to be traveling, the upgrade probability can be sorted from large to small according to the probability value to obtain the sorting result. Passengers with higher upgrade probability are more likely to choose the upgrade service and can be the focus of recommendation.

[0081] Step S105 , based on the M upgrade seats of the target flight, M users to be transported are selected from the sorting results as target upgrade users, and upgrade services of the target flight are recommended to the target upgrade users.

[0082] It should be noted that after calculating the upgrade probability of each user to be upgraded, the upgrade seats and the number M of upgrade seats of the target flight are obtained, and the users to be upgraded who are ranked first M in the sorting results are selected from large to small, and they are used as target upgrade users. Upgrade services are recommended to the target upgrade users, thereby achieving accurate recommendations. Customers who are not selected do not need to be pushed messages, avoiding unnecessary information interference to passengers, improving user satisfaction and avoiding the loss of customers.

[0083] Another optional specific implementation is described in detail below.

[0084] Figure 2 is a schematic diagram of an optional deep learning-based upgrade service recommendation process according to an embodiment of the present invention, such as Figure 2 As shown in the figure, the upgrade service recommendation process based on deep learning includes:

[0085] Step 1, start;

[0086] Step 2: Obtain user and scenario flight information;

[0087] It should be noted that, for data processing, the embodiments of the present invention can use DataX, Flink, Kafka, and other batch and stream data collection components to build a configurable, low-code data collection platform to achieve access and collection of full market data, airline sales data, upgrade service data, and user portrait data. In addition, a data lake platform for upgrade services can be built based on CDH and Hud i to achieve storage of massive data and quasi-real-time data services.

[0088] Based on the consideration of data entry efficiency and data quality, and to ensure data reliability and consistency, this embodiment proposes a big data lake warehouse integrated solution, using a combination of flink and spark. For historical data with high data quality and no duplication, spark is used to directly enter the lake; for duplicate data, flink's streaming processing is used to deduplicate the data in the partition, and then spark is used to perform global deduplication in the backend. At the same time, the parallelism and computing resources are adjusted to improve the efficiency of data entry into the lake. Offline data is entered into the lake in a batch running manner. First, the data is extracted from the database or file, and the data is written into the Apache Hud i table using spark, and stored using the data storage method provided by Apache Hud i itself. For real-time data, the data is first extracted into kafka. In this process, data skew needs to be prevented, and it is distributed to the topic's partition according to the table name and PK hash (hash value); then flink is used to consume the data from kafka, and different tasks are established to consume data from different partitions, and written to hdfs in parallel.

[0089] Step 3: Calculate the probability based on the probability prediction model of the probability upgrade;

[0090] The upgrade probability model is pre-built. Figure 3 FIG. 1 is a schematic diagram of a construction process of an optional upgrade probability prediction model according to an embodiment of the present invention. Figure 3 As shown, the steps of constructing the upgrade probability model include:

[0091] First, start;

[0092] Second, data set division;

[0093] 3. Feature processing;

[0094] Before building the model, feature engineering is required. This includes feature selection, feature scaling, feature crossover, and other operations on flight information, user information, and upgrade service information to extract useful features and enhance the model's expressiveness. For example, the flight's departure and arrival times are converted into time difference features to capture the flight's duration information.

[0095] After completing feature engineering, you can start training the model. Input the prepared data into the models of each part for training. During the training process, the model will optimize the parameters according to the input features and labels to minimize the loss function. Use common optimization algorithms such as stochastic gradient descent (SGD) or Adam optimizer, and set the appropriate learning rate and batch size.

[0096] Fourth, nonlinear model training and linear model training;

[0097] The upgrade probability prediction model is composed of a nonlinear model and a linear model. The linear model is mainly used to capture the cross-relationship between features, and the nonlinear model is mainly used to learn the nonlinear relationship between features. A multi-layer perceptron (MLP) can be used as a nonlinear model, and multiple hidden layers can be used to learn the complex relationship between features.

[0098] V. Linear model test and linear model test;

[0099] 6. Obtaining the upgrade probability prediction model;

[0100] During the entire training process, attention should be paid to data quality and model tuning. The performance of the model can be improved through data cleaning, feature selection, model parameter adjustment, etc. In addition, some techniques such as regularization and batch standardization can be used to prevent overfitting and improve the generalization ability of the model.

[0101] Seven, the end.

[0102] After the model training is completed, new flight information, user information, and upgrade service information are input into the model according to specific task requirements, and decisions or personalized recommendations are made through the model output. For example, based on the user's historical purchase records, current upgrade service information, and designated flight information, it is predicted whether the user will purchase an upgrade voucher for a designated flight, and upgrade vouchers of different denominations are recommended to the user.

[0103] Step 4: predict the probability of passenger upgrade;

[0104] Step 5: Select target passengers from the sorting results and recommend upgrade services to the target passengers;

[0105] The upgrade probability prediction model can accurately predict the upgrade probability of a passenger given a flight scenario and upgrade price information. In the rescheduling stage, in order to maximize the benefits for the airline, the upgrade price and the passenger upgrade probability will be re-selected. All passengers with an upgrade probability higher than 80% will be included in the selection range, and then the passengers will be arranged in descending order of upgrade price-descending probability, and combined with the number of available high-class seats on the current flight, they will be selected and recommended from top to bottom.

[0106] Step six, end.

[0107] In the embodiment of the present invention, deep learning technology is used to process and analyze a large amount of flight information, user profile information and upgrade service information, and the user's upgrade probability is calculated based on the upgrade service information, flight information and user profile information, and then users with a higher upgrade probability are selected for accurate recommendation. The upgrade probability prediction model built based on deep learning technology includes two parts: a linear model and a nonlinear model. When processing relevant information, the cross-relationship and nonlinear relationship between features can be considered at the same time, thereby improving the accuracy and precision of the recommendation results.

[0108] The following is a detailed description in conjunction with another embodiment.

[0109] Embodiment 2

[0110] A deep learning-based upgrade service recommendation device provided in this embodiment includes multiple implementation units, each of which corresponds to each implementation step in the above-mentioned embodiment 1. Its specific implementation methods and beneficial effects can refer to the above-mentioned method embodiments and will not be repeated here.

[0111] Figure 4 is a schematic diagram of an optional deep learning-based upgrade service recommendation device according to an embodiment of the present invention, such as Figure 4 As shown, the upgrade service recommendation device based on deep learning may include: an acquisition unit 41, a preprocessing unit 42, an output unit 43, an output unit 43, and a recommendation unit 45, wherein:

[0112] An acquisition unit 41 is used to acquire user portrait information of N users to travel corresponding to a target flight, flight information of the target flight, and upgrade service information, where N is a positive integer;

[0113] A preprocessing unit 42 is used to preprocess the user portrait information of the user to be traveled, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the user to be traveled, a flight feature vector of the target flight, and an upgrade service feature vector;

[0114] The output unit 43 is used to input the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into the upgrade probability prediction model for any user to be transported, and output the upgrade probability of the user to be transported for the target flight, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be transported, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model;

[0115] A sorting unit 44 is used to sort the upgrade probabilities of N users waiting to travel on the target flight to obtain a sorting result;

[0116] The recommendation unit 45 is configured to select M users to be transported as target upgrade users from the sorting results based on the M upgrade seats of the target flight, and recommend upgrade services of the target flight to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0117] The above-mentioned upgrade service recommendation device based on deep learning obtains user portrait information of N users to be transported corresponding to the target flight, flight information of the target flight, and upgrade service information through the acquisition unit 41, wherein N is a positive integer; the user portrait information of the users to be transported, the flight information of the target flight, and the upgrade service information are preprocessed through the preprocessing unit 42 to obtain the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector; and the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector are output to the output unit 43 for any user to be transported. The upgrade probability prediction model is input into the upgrade probability prediction model, and the upgrade probability of the user to be transported for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be transported, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; the upgrade probabilities of the N users to be transported of the target flight are sorted by the sorting unit 44 to obtain a sorting result; the recommendation unit 45 selects M users to be transported from the sorting result as target upgrade users based on the M upgrade seats of the target flight, and recommends the upgrade service of the target flight to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0118] In this embodiment, flight information, upgrade service information of the current flight, and user portrait information of the user to be traveled corresponding to the current flight are obtained, and the flight information, upgrade service information and user portrait information are analyzed and calculated through a pre-built upgrade probability prediction model to predict the user's upgrade probability. Users with a higher upgrade probability are selected as target upgrade users based on the upgrade cabin space of the target flight, and upgrade services are accurately recommended to the target upgrade users to improve user satisfaction, while avoiding information interference to users who do not need upgrades, and achieving accurate recommendation of upgrade services, which is convenient for airlines to better maintain customer sources, thereby solving the technical problem in the related art that it is difficult to make accurate upgrade recommendations for passengers through full recommendation or manually selecting passengers for batch recommendation.

[0119] Optionally, the output unit 43 includes: a first output module, which is used to input the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector into a linear model in the upgrade probability prediction model, and output a linear feature vector through the linear model, wherein the linear feature vector is used to characterize the linear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector; a second output module, which is used to input the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector into a nonlinear model in the upgrade probability prediction model, and output a nonlinear feature vector through the nonlinear model, wherein the nonlinear feature vector is used to characterize the nonlinear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector, and the complexity of the linear relationship is greater than the complexity of the nonlinear relationship; a first calculation module, which is used to perform weighted calculation on the linear feature vector and the nonlinear feature vector to obtain a calculation result; a third output module, which is used to calculate the upgrade probability based on the calculation result and the activation function, and output the upgrade probability of the user to be travel for the target flight through the upgrade probability prediction model.

[0120] Optionally, the preprocessing unit 42 includes: a first cleaning module, used to perform data cleaning processing on the user portrait information of the user to be traveled, the flight information of the target flight, and the upgrade service information, so as to obtain the processed user portrait information, flight information, and upgrade service information; a first extraction module, used to perform feature extraction on the processed user portrait information, the flight information of the target flight, and the upgrade service information based on preset feature extraction rules, so as to obtain user portrait feature data of the user to be traveled, flight feature data of the target flight, and upgrade service feature data; a first encoding module, used to encode the user portrait feature data of the user to be traveled, the flight feature data of the target flight, and the upgrade service feature data based on preset feature encoding rules, so as to obtain the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector.

[0121] Optionally, the user portrait feature data includes user attribute feature data and user behavior feature data, wherein the user attribute feature data includes at least one of the following: user gender feature data, user age feature data, user occupation feature data, user residence feature information, user level feature information; the user behavior feature data includes at least one of the following: user historical air travel feature data, user historical ticket purchase feature data, user historical upgrade feature data.

[0122] Optionally, the flight characteristic data of the target flight includes at least one of the following: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, season characteristic data, date characteristic data, characteristic data of the take-off time of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, and characteristic data of the mileage of the flight segments of the target flight.

[0123] Optionally, the upgrade service characteristic data includes at least one of the following: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, and upgrade location characteristic data.

[0124] Optionally, the deep learning-based upgrade service recommendation device further includes: a first acquisition module, used to acquire user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users; a first configuration module, used to configure upgrade probability labels for historical travel users based on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users; a second acquisition module, used to perform feature processing and feature encoding on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of historical travel users to obtain a historical feature vector of the historical travel user, the historical feature vector including: a historical user portrait feature vector, a historical flight feature vector, a historical upgrade service feature vector, and a probability distribution of the historical features. A historical linear feature vector and a historical nonlinear feature vector; a first construction module, used to construct a sample set based on the historical feature vectors and upgrade probability labels of historical travel users, and divide the sample set to obtain a training set and a test set; a first training module, used to jointly train the initial linear model and the initial nonlinear model based on the training set to obtain an initial upgrade probability prediction model, wherein the joint training is used to indicate that the initial linear model and the initial nonlinear model use the same training set and the same loss function for iteration during the training process; a first testing module, used to test the initial upgrade probability prediction model based on the test set to obtain a test result, and when the test result indicates that the initial upgrade probability prediction model passes the test, a trained upgrade probability prediction model is obtained.

[0125] The above-mentioned deep learning-based upgrade service recommendation device may also include a processor and a memory. The above-mentioned acquisition unit 41, preprocessing unit 42, output unit 43, output unit 43, recommendation unit 45, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0126] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the upgrade service can be accurately recommended to passengers by adjusting the kernel parameters.

[0127] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory (fl ash RAM), and the memory includes at least one storage chip.

[0128] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned deep learning-based upgrade service recommendation methods.

[0129] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement any one of the above-mentioned deep learning-based upgrade service recommendation methods.

[0130] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program having the following method steps: obtaining user portrait information of N users to be transported corresponding to a target flight, flight information of the target flight, and upgrade service information, wherein N is a positive integer; preprocessing the user portrait information of the users to be transported, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the users to be transported, a flight feature vector of the target flight, and an upgrade service feature vector; for any user to be transported, the user portrait feature vector of the users to be transported, the flight feature vector of the target flight, and the upgrade service feature vector are preprocessed; and the upgrade service feature vector are input into the upgrade probability prediction model, and the upgrade probability of the user to be traveling for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be traveling, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; the upgrade probabilities of N users to be traveling on the target flight are sorted to obtain a sorting result; based on the M upgrade seats of the target flight, M users to be traveling are selected from the sorting result as target upgrade users, and the upgrade service of the target flight is recommended to the target upgrade users, wherein M is a positive integer and N is greater than M.

[0131] Figure 5 is a hardware structure block diagram of an electronic device (or mobile device) according to a method for recommending a class upgrade service based on deep learning according to an embodiment of the present invention. Figure 5 As shown, the electronic device may include one or more ( Figure 5 502a, 502b, ..., 502n are used to illustrate) processor 502 (processor 502 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), memory 504 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 5 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 5 More or fewer components as shown, or with Figure 5 Different configurations shown.

[0132] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0133] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0136] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0137] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, 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, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0138] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for recommending cabin upgrade services based on deep learning, characterized in that: include: Obtain user profile information of N users to travel corresponding to a target flight, flight information of the target flight, and upgrade service information, where N is a positive integer; Preprocessing the user portrait information of the user to be transported, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the user to be transported, a flight feature vector of the target flight, and an upgrade service feature vector; For any of the users to be transported, the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector are input into the upgrade probability prediction model, and the upgrade probability of the user to be transported for the target flight is output, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be transported, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; Sorting the upgrade probabilities of N users waiting to travel on the target flight to obtain a sorting result; Based on the M upgrade seats of the target flight, M users to be transported are selected from the sorting results as target upgrade users, and upgrade services of the target flight are recommended to the target upgrade users, wherein: M is a positive integer, and N is greater than M.

2. The recommendation method according to claim 1, characterized in that: The step of inputting the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into the upgrade probability prediction model, and outputting the upgrade probability of the user to be transported for the target flight comprises: Inputting the user portrait feature vector of the traveling user, the flight feature vector of the target flight, and the upgrade service feature vector into a linear model in the upgrade probability prediction model, and outputting a linear feature vector through the linear model, wherein the linear feature vector is used to characterize the linear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector; Inputting the user portrait feature vector of the user to be traveled, the flight feature vector of the target flight, and the upgrade service feature vector into a nonlinear model in the upgrade probability prediction model, and outputting a nonlinear feature vector through the nonlinear model, wherein the nonlinear feature vector is used to characterize the nonlinear relationship between the features corresponding to the user portrait feature vector, the flight feature vector, and the upgrade service feature vector, and the complexity of the linear relationship is greater than the complexity of the nonlinear relationship; Performing weighted calculation on the linear feature vector and the nonlinear feature vector to obtain a calculation result; The upgrade probability is calculated based on the calculation result and the activation function, and the upgrade probability of the user to be traveling for the target flight is output through the upgrade probability prediction model.

3. The recommendation method according to claim 1, characterized in that: The steps of preprocessing the user portrait information of the user to be transported, the flight information of the target flight, and the upgrade service information to obtain the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector include: Performing data cleaning processing on the user portrait information of the user to be transported, the flight information of the target flight, and the upgrade service information to obtain the processed user portrait information, the flight information, and the upgrade service information; Performing feature extraction on the processed user portrait information, the flight information of the target flight, and the upgrade service information based on a preset feature extraction rule to obtain user portrait feature data of the user to be traveled, flight feature data of the target flight, and upgrade service feature data; The user portrait feature data of the user to travel, the flight feature data of the target flight and the upgrade service feature data are encoded based on preset feature encoding rules to obtain a user portrait feature vector of the user to travel, a flight feature vector of the target flight and an upgrade service feature vector.

4. The recommendation method according to claim 3, characterized in that: The user portrait feature data includes user attribute feature data and user behavior feature data, wherein the user attribute feature data includes at least one of the following: user gender feature data, user age feature data, user occupation feature data, user residence feature information, and user level feature information; the user behavior feature data includes at least one of the following: user historical air travel feature data, user historical ticket purchase feature data, and user historical upgrade feature data.

5. The recommendation method according to claim 3, characterized in that: The flight characteristic data of the target flight includes at least one of the following: characteristic data of the airline to which the target flight belongs, characteristic data of the aircraft model of the target flight, season characteristic data, date characteristic data, characteristic data of the take-off time period of the target flight, price characteristic data of the target flight, characteristic data of the delay rate of the target flight, characteristic data of the occupancy rate of the target flight, and characteristic data of the mileage of the flight segments of the target flight.

6. The recommendation method according to claim 3, characterized in that: The upgrade service characteristic data includes at least one of the following: upgrade class characteristic data, upgrade price characteristic data, upgrade discount characteristic data, and upgrade location characteristic data.

7. The recommendation method according to claim 1, characterized in that: The steps of constructing the upgrade probability prediction model include: Obtain user profile information of historical travel users, historical flight information, and historical upgrade service information corresponding to historical flights; configuring an upgrade probability label for the historical travel user based on the user portrait information of the historical travel user, the historical flight information, and the historical upgrade service information corresponding to the historical flight; Performing feature processing and feature encoding on the user portrait information, historical flight information, and historical upgrade service information corresponding to the historical flights of the historical travel user to obtain a historical feature vector of the historical travel user, wherein the historical feature vector includes: a historical user portrait feature vector, a historical flight feature vector, a historical upgrade service feature vector, and a historical linear feature vector and a historical nonlinear feature vector between the historical features; Constructing a sample set based on the historical feature vectors of the historical travel users and the upgrade probability labels, and dividing the sample set into a training set and a test set; Based on the training set, the initial linear model and the initial nonlinear model are jointly trained to obtain an initial upgrade probability prediction model, wherein the joint training is used to indicate that the initial linear model and the initial nonlinear model use the same training set and the same loss function for iteration during the training process; The initial upgrade probability prediction model is tested based on the test set to obtain a test result, and when the test result indicates that the initial upgrade probability prediction model passes the test, the trained upgrade probability prediction model is obtained.

8. A deep learning-based upgrade service recommendation device, characterized in that: include: An acquisition unit, configured to acquire user portrait information of N users to be transported corresponding to a target flight, flight information of the target flight, and upgrade service information, wherein N is a positive integer; A preprocessing unit, configured to preprocess the user portrait information of the user to be transported, the flight information of the target flight, and the upgrade service information to obtain a user portrait feature vector of the user to be transported, a flight feature vector of the target flight, and an upgrade service feature vector; an output unit, for any of the users to be transported, inputting the user portrait feature vector of the user to be transported, the flight feature vector of the target flight, and the upgrade service feature vector into an upgrade probability prediction model, and outputting the upgrade probability of the user to be transported for the target flight, wherein the upgrade probability prediction model is a model pre-built based on deep learning technology for predicting the upgrade probability of the user to be transported, and the upgrade probability prediction model includes two parts: a linear model and a nonlinear model; A sorting unit, used to sort the upgrade probabilities of the N users waiting to travel on the target flight to obtain a sorting result; A recommendation unit is used to select M of the users to be transported from the sorting results as target upgrade users based on the M upgrade seats of the target flight, and recommend the upgrade service of the target flight to the target upgrade users, wherein M is a positive integer and N is greater than M.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the deep learning-based upgrade service recommendation method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The invention comprises one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the deep learning-based upgrade service recommendation method according to any one of claims 1 to 7.

Images