Method and system for predicting user willingness to pay, electronic device, and storage medium

Abstract

The present application discloses a method and system for predicting user willingness to pay, an electronic device, and a storage medium, wherein the method comprises: obtaining the user's current pre-purchase information of a target air ticket; generating a network graph corresponding to the pre-purchase information according to the relationship between each information in the pre-purchase information; performing graph convolution on the network graph of the pre-purchase information to obtain the graph features corresponding to the pre-purchase information; inputting the graph features corresponding to the pre-purchase information into a pre-trained prediction model, analyzing and calculating the graph features corresponding to the pre-purchase information through the prediction model, and obtaining the discount information of the target air ticket corresponding to the current user; wherein the prediction model is trained using a data set consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information; the adversarial network includes an information generator and an information discriminator. Thus, accurate prediction of user willingness to pay is achieved.

Description

Technical Field

[0001] The present application relates to the field of information prediction technology, and in particular to a method and system for predicting user willingness to pay, an electronic device, and a storage medium. Background Art

[0002] Airplanes are luxury goods among transportation tools due to their capacity and operating costs, so the cost of air tickets is significantly different from that of land transportation. The purchase volume of air tickets will fluctuate with the time period, season, weather, and even the air ticket pricing of competitor companies. Air ticket pricing will largely affect the user's willingness to pay, so how to set the price of air tickets to maximize revenue while satisfying the user's willingness to pay for air tickets is a problem that all airlines are extremely concerned about.

[0003] Currently, ticket pricing is mainly based on the inventory model, which is essentially to apply statistical data and set pricing strategies to price, so as to sell tickets to customers in need at a suitable price and in a suitable way. Alternatively, recommendation algorithms are used to analyze users' purchasing preferences based on a large amount of user information, so as to provide a decision-making basis for ticket pricing.

[0004] However, the pricing method based on the stock model needs to be based on statistical laws, so the data is cumulatively shared and has a lag. The pricing method based on the recommendation method cannot be directly priced. And because air tickets are relatively expensive, the number of times an individual appears on a plane is limited, and the tickets they may buy may be from different airlines. Therefore, it is impossible to get the preferences of a user based on a small amount of data of a certain user, and analysis is often based on the data of a large number of users. Therefore, there is a problem of low accuracy. Therefore, these existing methods cannot accurately predict the user's willingness to pay, so as to make the most reasonable pricing, and then, under the maximum profit, it can meet the psychological expectations of each user for air tickets, so that they can buy tickets from this airline. Summary of the invention

[0005] Based on the above-mentioned deficiencies of the prior art, the present application provides a method and system for predicting user willingness to pay, an electronic device, and a storage medium to solve the problem that the prior art cannot predict user willingness to pay.

[0006] In order to achieve the above objectives, this application provides the following technical solutions:

[0007] The first aspect of the present application provides a method for predicting a user's willingness to pay, comprising:

[0008] Get the user's current pre-purchase information for the target flight;

[0009] Generate a network diagram corresponding to the pre-purchased ticket information according to the relationship between each piece of information in the pre-purchased ticket information;

[0010] Performing graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information;

[0011] The graph features corresponding to the pre-purchased ticket information are input into a pre-trained prediction model, and the graph features corresponding to the pre-purchased ticket information are analyzed and calculated by the prediction model to obtain the discount information of the target air ticket corresponding to the current user; wherein the prediction model is trained using a data set consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information; the adversarial network includes an information generator and an information discriminator.

[0012] Optionally, in the above method, the training method of the adversarial network includes:

[0013] Acquire the real ticket purchase information, and pre-process the real ticket purchase information;

[0014] Based on the preprocessed real ticket purchase information, construct the information generator and the information discriminator;

[0015] Calling the information generator to randomly generate a set of initial ticket purchase information;

[0016] Shuffling the initial ticket purchase information with the official ticket purchase information to obtain a first data set;

[0017] Dividing the first data set into a first training set and a first test set;

[0018] Performing a round of iterative training on the current information discriminator using the first training set to obtain the latest current information discriminator;

[0019] Using the first test set to test the current information discriminator, to obtain a discrimination probability value of the current information discriminator;

[0020] Determine whether the current discrimination probability value of the information discriminator is greater than a first discrimination threshold, or whether the number of iterative training of the information discriminator reaches a first specified number of iterations;

[0021] If it is determined that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold, and the number of iterative training of the information discriminator has not reached the first specified number of iterations, then returning to execute the iterative training of the current information discriminator using the first training set to obtain the latest current information discriminator;

[0022] If it is determined that the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, or it is determined that the number of iterative training of the information discriminator reaches the first specified number of iterations, then the current information discriminator is stored;

[0023] Based on the current predefined learning rate, the current information generator is iteratively trained using the current information discriminator to obtain the latest current information generator; wherein, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information;

[0024] Using the current information generator to randomly generate a set of ticket purchase information to be identified, and shuffle the ticket purchase information to be identified with the real ticket purchase information to obtain a current set of data to be identified;

[0025] Using the current information discriminator to discriminate the current data set to be discriminated, and obtaining a discriminant value corresponding to the current data set to be discriminated;

[0026] Determine whether the discrimination value corresponding to the current data set to be discriminated is greater than a second discrimination threshold;

[0027] If it is determined that the discrimination value corresponding to the current data set to be discriminated is greater than the second discrimination threshold, the current information generator and the current information discriminator are stored;

[0028] If it is determined that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold, then determining whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate;

[0029] If it is determined that the number of iterative training of the information generator under the current predefined learning rate has not reached the second specified number of iterations, then return to execute the iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator;

[0030] If it is determined that the number of iterative training of the information generator has reached the second specified number of iterations under the current predefined learning rate, the current predefined learning rate is updated to obtain the latest current predefined learning rate, and for the latest current predefined learning rate, the process returns to execute a round of iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator.

[0031] Optionally, in the above method, after determining that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold, the method further includes:

[0032] Determine whether the total number of iterative training of the adversarial network is greater than the total number of predictions;

[0033] If it is determined that the total number of iterative trainings for the adversarial network is greater than the total number of predictions, then storing the current information generator and the current information discriminator is executed;

[0034] If it is determined that the total number of iterative training times for the adversarial network is not greater than the total number of predictions, then the determination is performed as to whether the number of iterative training times for the information generator reaches a second specified number of iterations under the current predefined learning rate.

[0035] Optionally, in the above method, the training method of the prediction model includes:

[0036] Based on the real ticket purchase information after preprocessing, the prediction model is constructed; wherein the neural network structure of the prediction model is a deep neural network with temporal causality obtained by superimposing multiple stack structures; the stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on a double residual superposition method;

[0037] Using the stored information generator to randomly generate a set of ticket purchase information, and confuse it with the real ticket purchase information to obtain a training data set;

[0038] Generate a network diagram corresponding to the training data set according to the relationship between each piece of information in the training data set;

[0039] Performing graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set;

[0040] The graph features corresponding to the training data set form a time series graph which is input into the prediction model, and the prediction model is iteratively trained until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, and the trained prediction model is obtained and stored.

[0041] Optionally, in the above method, after obtaining the user's current pre-purchase information of the target air ticket, the method further includes:

[0042] Preprocessing the pre-purchased ticket information to obtain the pre-purchased ticket information in a preset format;

[0043] The step of generating a network diagram corresponding to the pre-purchased ticket information according to the relationship between each piece of information in the pre-purchased ticket information includes:

[0044] According to the relationship between each piece of information in the pre-purchased ticket information in a preset format, a network diagram corresponding to the pre-purchased ticket information is generated.

[0045] The second aspect of the present application provides a user willingness to pay prediction system, comprising:

[0046] A first acquisition unit is used to acquire the user's current pre-purchase information of the target air ticket;

[0047] A first generating unit, configured to generate a network graph corresponding to the pre-purchased ticket information according to the relationship between each information in the pre-purchased ticket information;

[0048] A first convolution unit is used to perform graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information;

[0049] A prediction unit is used to input the graph features corresponding to the pre-purchased ticket information into a pre-trained prediction model, and analyze and calculate the graph features corresponding to the pre-purchased ticket information through the prediction model to obtain the discount information of the target air ticket corresponding to the current user; wherein the prediction model is trained using a data set consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information; the adversarial network includes an information generator and an information discriminator.

[0050] Optionally, the above system further includes an adversarial network training unit, wherein the adversarial network training unit includes:

[0051] A second acquisition unit, used to acquire the real ticket purchase information and pre-process the real ticket purchase information;

[0052] A first construction unit, configured to construct the information generator and the information discriminator based on the preprocessed real ticket purchase information;

[0053] A calling unit, used for calling an information generator to randomly generate a set of initial ticket purchase information;

[0054] A first processing unit, configured to shuffle the initial ticket purchase information with the formal ticket purchase information to obtain a first data set;

[0055] A data division unit, used for dividing the first data set into a first training set and a first test set;

[0056] A first training unit, configured to perform a round of iterative training on the current information discriminator using the first training set to obtain the latest current information discriminator;

[0057] A testing unit, used to test the current information discriminator using the first test set to obtain a discrimination probability value of the current information discriminator;

[0058] A first judging unit, used to judge whether the current discrimination probability value of the information discriminator is greater than a first discrimination threshold, or whether the number of iterative training of the information discriminator reaches a first specified number of iterations;

[0059] A first returning unit is configured to return to the first training unit to perform a round of iterative training of the current information discriminator using the first training set to obtain the latest current information discriminator when the first judging unit judges that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold and the number of iterative training of the information discriminator has not reached the first specified number of iterations;

[0060] A first storage unit, configured to store the current information discriminator when the first judgment unit judges that the discrimination probability value of the current information discriminator is greater than a first discrimination threshold, or judges that the number of iterative training of the information discriminator reaches a first specified number of iterations;

[0061] The second training unit is used to perform a round of iterative training on the current information generator using the current information discriminator based on the current predefined learning rate to obtain the latest current information generator; wherein, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information;

[0062] A second processing unit is used to randomly generate a set of ticket purchase information to be determined by using the current information generator, and shuffle the ticket purchase information to be determined with the real ticket purchase information to obtain a current set of data to be determined;

[0063] A discriminant unit, used for discriminating the current data set to be discriminated by using the current information discriminator, and obtaining a discriminant value corresponding to the current data set to be discriminated;

[0064] A second judgment unit, used to judge whether the discrimination value corresponding to the current data set to be judged is greater than a second discrimination threshold;

[0065] A second storage unit, configured to store the current information generator and the current information discriminator when the second judging unit judges that the discrimination value corresponding to the current data set to be discriminated is greater than a second discrimination threshold;

[0066] A third judgment unit is used to judge whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold;

[0067] A second returning unit is used to return to the second training unit to perform a round of iterative training on the current information generator based on the current predefined learning rate using the current information discriminator, to obtain the latest current information generator, when the third judging unit judges that the number of iterative training of the information generator under the current predefined learning rate has not reached the second specified number of iterations;

[0068] An updating unit is used for updating the current predefined learning rate when the third judgment unit determines that the number of iterative training of the information generator has reached a second specified number of iterations under the current predefined learning rate to obtain the latest current predefined learning rate, and for the latest current predefined learning rate, returning to the second training unit to execute a round of iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator.

[0069] Optionally, in the above system, the adversarial network training unit further includes:

[0070] The fourth judgment unit is used to judge whether the total number of iterative training of the adversarial network is greater than the predicted total number when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold; wherein, if it is judged that the total number of iterative training of the adversarial network is greater than the predicted total number, the second storage unit executes the storage of the current information generator and the current information discriminator; if it is judged that the total number of iterative training of the adversarial network is not greater than the predicted total number, the third judgment unit executes the judgment whether the number of iterative training of the information generator reaches the second specified number of iterations under the current predefined learning rate.

[0071] Optionally, the above system further includes a prediction model training unit, wherein the prediction model training unit includes:

[0072] A second construction unit is used to construct the prediction model based on the real ticket purchase information after preprocessing; wherein the neural network structure of the prediction model is a deep neural network with temporal causality obtained by superimposing multiple stack structures; the stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on a double residual superposition method;

[0073] A third processing unit, configured to randomly generate a set of ticket purchase information using the stored information generator, and confuse the generated set of ticket purchase information with the real ticket purchase information to obtain a training data set;

[0074] A second generating unit, configured to generate a network graph corresponding to the training data set according to the relationship between each information in the training data set;

[0075] A second convolution unit is used to perform graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set;

[0076] The third training unit is used to input the timing diagram composed of the graph features corresponding to the training data set into the prediction model, and iteratively train the prediction model until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, to obtain and store the trained prediction model.

[0077] Optionally, in the above system, it also includes:

[0078] A preprocessing unit, used for preprocessing the pre-purchased ticket information to obtain the pre-purchased ticket information in a preset format;

[0079] Wherein, when the first generating unit generates the network graph corresponding to the pre-purchased ticket information according to the relationship between each information in the pre-purchased ticket information, it is used to:

[0080] According to the relationship between each piece of information in the pre-purchased ticket information in a preset format, a network diagram corresponding to the pre-purchased ticket information is generated.

[0081] A third aspect of the present application provides an electronic device, including:

[0082] one or more processing devices;

[0083] a memory having one or more programs stored therein;

[0084] When the one or more programs are executed by the one or more processing devices, the one or more processing devices implement the method for predicting user willingness to pay as described in any one of the above.

[0085] A fourth aspect of the present application provides a computer storage medium for storing a program, which, when executed, is used to implement a method for predicting user willingness to pay as described in any one of the above.

[0086] The method for predicting user willingness to pay provided by the present application is to train an adversarial network including an information generator and an information discriminator, and then train the prediction model with a data set consisting of ticket purchase information generated by the trained adversarial network and real ticket purchase information, so as to ensure that there is enough data to train the model, thereby ensuring the accuracy of the model. When a prediction is required, the user's current pre-purchase information for the target air ticket is obtained. Then, according to the relationship between each information in the pre-purchase information, a network graph corresponding to the pre-purchase information is generated, and the network graph of the pre-purchase information is convolved to obtain the graph features corresponding to the pre-purchase information, so that by means of graph convolution, various influencing factors can be fully considered to ensure the accuracy of the prediction results. Finally, the graph features corresponding to the pre-purchase information are input into the pre-trained prediction model, and the graph features corresponding to the pre-purchase information are analyzed and calculated by the prediction model to obtain the discount information that the current user is willing to pay to purchase the target air ticket, so that the user's willingness to pay can be accurately predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0088] Figure 1 A schematic diagram of the architecture of a prediction system provided in an embodiment of the present application;

[0089] Figure 2 A flowchart of a method for predicting user willingness to pay provided in another embodiment of the present application;

[0090] Figure 3 A flowchart of a training method for an adversarial network model provided in another embodiment of the present application;

[0091] Figure 4 A flowchart of a prediction model training method provided in another embodiment of the present application;

[0092] Figure 5 A schematic diagram of a neural network structure of a prediction model provided in another embodiment of the present application;

[0093] Figure 6 A schematic diagram of the structure of a system for predicting user willingness to pay provided in another embodiment of the present application;

[0094] Figure 7 A schematic diagram of the structure of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0095] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0096] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0097] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0098] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0099] The present application provides a method for predicting user willingness to pay, so as to solve the problem that the prior art cannot predict user willingness to pay.

[0100] Optionally, another embodiment of the present application provides a prediction system for implementing the method for predicting user willingness to pay provided by the present application. Figure 1 As shown, the prediction system provided by the embodiment of the present application includes: an interaction module 101, a preprocessing module 102, a scheduling module 103, a construction module 104, a training module 105, and a prediction module 106.

[0101] The interactive module 101 is mainly used to provide the user with an interface for transmitting information and to perform operations such as importing and outputting information.

[0102] The preprocessing module 102 is mainly used to preprocess the data, which may specifically include preprocessing the data and formatting the data in the subsequent model training process.

[0103] The scheduling module 103 is mainly used to realize parameter transfer between modules and complete the management of the algorithm iteration process.

[0104] The construction module 104 is mainly used to construct an information discriminator, an information generator and a prediction model according to the formatted information during initialization.

[0105] The training module 105 is mainly used to train the model by inputting data. This module contains deep neural network models such as feature extraction network, information generation network, information discrimination network, graph convolution network and prediction network, which are used to process various training processes of the algorithm.

[0106] The prediction module 106 is mainly used to realize the use of the prediction model, specifically to process the data, convert the data into a data structure used by the prediction model, and translate it into a usable data structure after completing the prediction calculation.

[0107] The present application embodiment provides a method for predicting a user's willingness to pay, such as Figure 2 As shown, the following steps are included:

[0108] S201. Obtain the user's current pre-purchase information for the target air ticket.

[0109] Among them, the pre-purchase ticket information mainly includes the attribute information of the target ticket, such as the departure place, destination, take-off and landing time, original ticket price, route type, warehouse type, etc. Of course, it can also include user information, such as purchasing power, historical ticket purchase records, etc.

[0110] Optionally, in another embodiment of the present application, after executing step S201, the method may further include: pre-processing the pre-purchased ticket information to obtain the pre-purchased ticket information in a preset format, thereby converting the data into a unified format for subsequent processing.

[0111] In the embodiment of the present application, step S202 corresponds to: generating a network diagram corresponding to the pre-purchased ticket information according to the relationship between each information in the pre-purchased ticket information in a preset format.

[0112] S202: Generate a network diagram corresponding to the pre-purchase ticket information according to the relationship between each piece of information in the pre-purchase ticket information.

[0113] Specifically, there is a relationship between each piece of information in the pre-purchase ticket information. Each corresponding node can be created, and then the corresponding nodes can be connected according to the relationship between the information, so as to reflect the relationship between each piece of information through a network diagram.

[0114] S203: Perform graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information.

[0115] In the embodiment of the present application, graph features are extracted from the network graph of the pre-purchase information by means of graph convolution, so that various factors affecting the ticket price can be more fully considered.

[0116] S204: Input the graph features corresponding to the pre-purchased ticket information into a pre-trained prediction model, analyze and calculate the graph features corresponding to the pre-purchased ticket information through the prediction model, and obtain discount information of the target air ticket corresponding to the current user.

[0117] Among them, the discount information of the target air ticket corresponding to the current user refers to the discount that the user is willing to pay to purchase the target air ticket at the moment. The discount can be the highest discount. Therefore, air ticket pricing based on the discount information can not only increase the user's purchase rate, but also effectively guarantee the airline's profits.

[0118] The prediction model is trained using a dataset consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information. The real ticket purchase information refers to the ticket purchase information of the air tickets that the user has purchased before.

[0119] It should be noted that considering that the number of times a user uses an airplane is limited, and the number of times a user uses a certain airline is even rarer, there is a problem of limited existing information. Therefore, a sufficient amount of ticket purchase information is generated through a trained adversarial network, and then combined with the real ticket purchase information to form a data set with a large enough amount of data to train the model, thereby avoiding the overfitting problem caused by insufficient data and effectively ensuring the accuracy of the prediction results.

[0120] The adversarial network includes an information generator and an information discriminator. The information generator is used to generate ticket purchase information, while the information discriminator is used to judge the accuracy of the ticket purchase information generated by the information generator.

[0121] Optionally, another embodiment of the present application provides a training method for an adversarial network model, such as Figure 3 As shown, the following steps are included:

[0122] S301, obtaining real ticket purchase information, and preprocessing the real ticket purchase information.

[0123] After obtaining the real ticket purchase information, the real ticket purchase information is pre-processed to process the information into information in a specified format.

[0124] S302: Based on the pre-processed real ticket purchase information, construct an information generator and an information discriminator.

[0125] S303: Calling an information generator to randomly generate a set of initial ticket purchase information.

[0126] S304: Shuffle the initial ticket purchase information and the official ticket purchase information to obtain a first data set.

[0127] S305: Divide the first data set into a first training set and a first test set.

[0128] S306: Perform a round of iterative training on the current information discriminator using the first training set to obtain the latest current information discriminator.

[0129] S307 , using the first test set to test the current information discriminator to obtain a discrimination probability value of the current information discriminator.

[0130] S308: Determine whether the discrimination probability value of the current information discriminator is greater than the first discrimination threshold.

[0131] If it is determined that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold, step S309 is executed. If it is determined that the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, step S310 is executed.

[0132] S309: Determine whether the number of iterative training of the information discriminator reaches a first specified number of iterations.

[0133] If it is determined that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold, and it is determined that the number of iterative training of the information discriminator has not reached the first specified number of iterations, then the information discriminator is trained, so when the judgment in step S309 is no, it returns to step S306. If it is determined that the number of iterative training of the information discriminator has reached the first specified number of iterations, it means that the training of the information discriminator can be ended, so step S310 is executed at this time.

[0134] S310, storing the current information discriminator.

[0135] After the information discriminator is trained, the subsequent steps are performed using the stored current information discriminator to train the information generator.

[0136] S311. Based on the current predefined learning rate, use the current information discriminator to perform a round of iterative training on the current information generator to obtain the latest current information generator.

[0137] Among them, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information.

[0138] The predefined characteristic information refers to information within a preset characteristic range. For example, if the passenger age is between 0 and 100, any value between 0 and 100 belongs to the predefined characteristic information. In order to allow the information generator to generate complete ticket purchase information in a targeted manner, it is equivalent to inputting part of the ticket purchase information into the information generator so that the information generator can generate complete ticket purchase information based on the input information.

[0139] S312. Use the current information generator to randomly generate a set of ticket purchase information to be judged, and shuffle the ticket purchase information to be judged with the real ticket purchase information to obtain the current set of data to be judged.

[0140] S313: Use the current information discriminator to discriminate the current data set to be discriminated, and obtain a discriminant value corresponding to the current data set to be discriminated.

[0141] S314: Determine whether the discrimination value corresponding to the current data set to be discriminated is greater than a second discrimination threshold.

[0142] If it is determined that the discrimination value corresponding to the current dataset to be discriminated is greater than the second discrimination threshold, step S315 is executed. If it is determined that the discrimination value corresponding to the current dataset to be discriminated is not greater than the second discrimination threshold, step S316 is executed.

[0143] Optionally, in another embodiment of the present application, after determining that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold, it is possible to further perform: determining whether the total number of iterative training times for the adversarial network is greater than the total number of predictions.

[0144] It should be noted that when the total number of iterative training times for the adversarial network is greater than the total number of predictions, it means that further optimization of the adversarial network is already very limited. Therefore, in order to improve efficiency, in an embodiment of the present application, if it is determined that the total number of iterative training times for the adversarial network is greater than the total number of predictions, step S315 is executed.

[0145] If it is determined that the total number of iterative training times for the adversarial network is not greater than the total number of predictions, step S316 is executed.

[0146] S315. Store the current information generator and the current information discriminator.

[0147] It should be noted that the information generator and the information discriminator are adversarial networks, so during the training of the information generator, the information discriminator can also be adjusted according to the feedback. Therefore, the current information generator and the current information discriminator need to be stored in the end.

[0148] S316. Determine whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate.

[0149] If it is determined that the number of iterative training of the information generator has not reached the second specified number of iterations under the current predefined learning rate, the process returns to step S311. If it is determined that the number of iterative training of the information generator has reached the second specified number of iterations under the current predefined learning rate, the process executes step S317.

[0150] S317: Update the current predefined learning rate to obtain the latest current predefined learning rate.

[0151] It should be noted that after obtaining the latest current predefined learning rate, the process returns to execute step S311 for the latest current predefined learning rate.

[0152] Optionally, based on Figure 3 The training method of the adversarial network model shown in the figure, another embodiment of the present application provides a training method of a prediction model, such as Figure 4 As shown, the following steps are included:

[0153] S401: Build a prediction model based on the pre-processed real ticket purchase information.

[0154] Among them, the neural network structure of the prediction model can be specifically as follows Figure 5 As shown, a deep neural network with temporal causality is obtained by superimposing multiple stack structures. The stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on a double residual superposition method.

[0155] S402: Use the stored information generator to randomly generate a set of ticket purchase information, and confuse it with the real ticket purchase information to obtain a training data set.

[0156] S403: Generate a network diagram corresponding to the training data set according to the relationship between each information in the training data set.

[0157] S404: Perform graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set.

[0158] S405. Input the graph features corresponding to the training data set into a time series graph into the prediction model, perform iterative training on the prediction model until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, and obtain and store the trained prediction model.

[0159] The method for predicting user willingness to pay provided in the embodiment of the present application is to train an adversarial network including an information generator and an information discriminator, and then train the prediction model with a data set consisting of ticket purchase information generated by the trained adversarial network and real ticket purchase information, so as to ensure that there is enough data to train the model, thereby ensuring the accuracy of the model. When a prediction is required, the user's current pre-purchase information for the target air ticket is obtained. Then, according to the relationship between each information in the pre-purchase information, a network graph corresponding to the pre-purchase information is generated, and the network graph of the pre-purchase information is convolved to obtain the graph features corresponding to the pre-purchase information, so that by means of graph convolution, various influencing factors can be fully considered to ensure the accuracy of the prediction results. Finally, the graph features corresponding to the pre-purchase information are input into the pre-trained prediction model, and the graph features corresponding to the pre-purchase information are analyzed and calculated by the prediction model to obtain the discount information that the current user is willing to pay to purchase the target air ticket, so that the user's willingness to pay can be accurately predicted.

[0160] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0161] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0162] Although operations are depicted in a particular order, this should not be understood as requiring that the operations be performed in the particular order shown or in a sequential order.Multitasking and parallel processing may be advantageous under certain circumstances.

[0163] Another embodiment of the present application provides a system for predicting user willingness to pay, such as Figure 6 As shown, it includes the following units:

[0164] The first acquisition unit 601 is used to acquire the user's current pre-purchase information of the target air ticket.

[0165] The first generating unit 602 is used to generate a network graph corresponding to the pre-purchase ticket information according to the relationship between each information in the pre-purchase ticket information.

[0166] The first convolution unit 603 is used to perform graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information.

[0167] The prediction unit 604 is used to input the graph features corresponding to the pre-purchased ticket information into a pre-trained prediction model, analyze and calculate the graph features corresponding to the pre-purchased ticket information through the prediction model, and obtain the discount information of the target air ticket corresponding to the current user.

[0168] The prediction model is trained using a dataset consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information. The adversarial network includes an information generator and an information discriminator.

[0169] Optionally, in the user willingness to pay prediction system provided by another embodiment of the present application, an adversarial network training unit is also included, wherein the adversarial network training unit includes:

[0170] The second acquisition unit is used to acquire the real ticket purchase information and pre-process the real ticket purchase information.

[0171] The first construction unit is used to construct an information generator and an information discriminator based on the preprocessed real ticket purchase information.

[0172] A calling unit, used for calling an information generator to randomly generate a set of initial ticket purchase information;

[0173] The first processing unit is used to shuffle the initial ticket purchase information and the formal ticket purchase information to obtain a first data set.

[0174] The data division unit is used to divide the first data set into a first training set and a first test set.

[0175] The first training unit is used to perform a round of iterative training on the current information discriminator using the first training set to obtain the latest current discriminator.

[0176] The testing unit is used to test the current information discriminator using the first test set to obtain the discrimination probability value of the current information discriminator.

[0177] The first judgment unit is used to judge whether the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, or whether the number of iterative training of the information discriminator reaches a first specified number of iterations.

[0178] The first return unit is used to return to the first training unit to perform a round of iterative training on the current information discriminator using the first training set to obtain the latest current discriminator when the first judgment unit determines that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold and the number of iterative training of the information discriminator has not reached the first specified number of iterations.

[0179] The first storage unit is used to store the current information discriminator when the first judgment unit determines that the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, or determines that the number of iterative training of the information discriminator reaches a first specified number of iterations.

[0180] The second training unit is used to perform a round of iterative training on the current information generator using the current information discriminator based on the current predefined learning rate to obtain the latest current information generator.

[0181] Among them, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information.

[0182] The second processing unit is used to randomly generate a set of ticket purchase information to be judged by using the current information generator, and shuffle the ticket purchase information to be judged with the real ticket purchase information to obtain the current set of data to be judged.

[0183] The discriminant unit is used to discriminate the current data set to be discriminated by using the current information discriminator to obtain a discriminant value corresponding to the current data set to be discriminated.

[0184] The second judgment unit is used to judge whether the discrimination value corresponding to the current data set to be judged is greater than a second discrimination threshold.

[0185] The second storage unit is used to store the current information generator and the current information discriminator when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is greater than the second discrimination threshold.

[0186] The third judgment unit is used to judge whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold.

[0187] The second return unit is used to return to the second training unit to perform a round of iterative training on the current information generator based on the current predefined learning rate using the current information discriminator when the third judgment unit determines that the number of iterative training of the information generator has not reached a second specified number of iterations under the current predefined learning rate, so as to obtain the latest current information generator.

[0188] The updating unit is used to update the current predefined learning rate when the third judgment unit determines that the number of iterative training of the information generator has reached a second specified number of iterations under the current predefined learning rate, so as to obtain the latest current predefined learning rate, and return to the second training unit to perform a round of iterative training on the current information generator based on the current predefined learning rate and using the current information discriminator to obtain the latest current information generator.

[0189] Optionally, in the user willingness to pay prediction system provided by another embodiment of the present application, the adversarial network training unit further includes:

[0190] The fourth judgment unit is used to judge whether the total number of iterative training of the adversarial network is greater than the predicted total number when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold; wherein, if it is judged that the total number of iterative training of the adversarial network is greater than the predicted total number, the second storage unit executes to store the current information generator and the current information discriminator; if it is judged that the total number of iterative training of the adversarial network is not greater than the predicted total number, the third judgment unit executes to judge whether the number of iterative training of the information generator reaches the second specified number of iterations under the current predefined learning rate.

[0191] Optionally, the user willingness to pay prediction system provided in another embodiment of the present application further includes a prediction model training unit, wherein the prediction model training unit includes:

[0192] The second construction unit is used to construct a prediction model based on the preprocessed real ticket purchase information.

[0193] The neural network structure of the prediction model is a deep neural network with temporal causality obtained by superimposing multiple stack structures. The stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on the double residual superposition method.

[0194] The third processing unit is used to randomly generate a set of ticket purchase information using the stored information generator, and confuse it with the real ticket purchase information to obtain a training data set.

[0195] The second generating unit is used to generate a network graph corresponding to the training data set according to the relationship between each information in the training data set.

[0196] The second convolution unit is used to perform graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set.

[0197] The third training unit is used to input the graph feature composition time series graph corresponding to the training data set into the prediction model, and iteratively train the prediction model until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, to obtain and store the trained prediction model.

[0198] Optionally, in the user willingness to pay prediction system provided in another embodiment of the present application, the system further includes:

[0199] The pre-processing unit is used to pre-process the pre-purchased ticket information to obtain the pre-purchased ticket information in a preset format.

[0200] Wherein, when the first generating unit generates a network diagram corresponding to the pre-purchased ticket information according to the relationship between each information in the pre-purchased ticket information, it is used to:

[0201] According to the relationship between each piece of information in the pre-purchased ticket information in a preset format, a network diagram corresponding to the pre-purchased ticket information is generated.

[0202] It should be noted that the various units provided in the above embodiments of the present application may belong to Figure 1 The specific working process of each unit can refer to the implementation of the corresponding steps in the above method embodiment, which will not be repeated here. The units involved in the above embodiments can be implemented by software or hardware. The name of the unit does not constitute a limitation on the unit itself under certain circumstances.

[0203] Furthermore, the functions of the various units described in the above embodiments may be at least partially performed by one or more hardware logic components. For example, exemplary types of hardware logic components that may be used include, without limitation, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), complex programmable logic devices (CPLDs), etc.

[0204] Another embodiment of the present application provides an electronic device, such as Figure 7 , which shows a schematic diagram of the structure of an electronic device 700 suitable for implementing the embodiment of the present disclosure. The electronic device in the embodiment of the present disclosure may include but is not limited to electronic devices such as desktop computers, laptop computers, tablet computers, and vehicle-mounted terminals.

[0205] like Figure 7As shown, the electronic device 700 includes one or more processing devices 701, such as a central processing unit, a graphics processing unit, etc., and a memory 702, on which one or more programs are stored. When the one or more programs are executed by the one or more processing devices 701, the one or more processing devices 701 implement the method for generating boarding information provided in any of the above embodiments.

[0206] Optionally, the electronic device may also include other components, see also Figure 7 , the processing device 701, the read-only memory ROM 702 and the random access memory RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704. Typically, the following devices can be connected to the I / O interface 705: an input device 706 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 707 including, for example, a liquid crystal display, a speaker, a vibrator, etc., a storage device 706 including, for example, a magnetic tape, a hard disk, etc., and a communication device 709. The communication device 709 can allow the electronic device 700 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The electronic device 700 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0207] Another embodiment of the present application provides a computer storage medium for storing a program, which, when executed, is used to implement a method for predicting user willingness to pay as provided in any of the above embodiments.

[0208] Computer storage media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0209] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0210] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting user willingness to pay, characterized in that: include: Acquire the user's current pre-purchase information for the target air ticket; wherein the pre-purchase information includes attribute information of the target air ticket; Generate a network diagram corresponding to the pre-purchased ticket information according to the relationship between each piece of information in the pre-purchased ticket information; Performing graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information; Input the graph features corresponding to the pre-purchased ticket information into a pre-trained prediction model, analyze and calculate the graph features corresponding to the pre-purchased ticket information through the prediction model, and obtain the discount information of the target air ticket corresponding to the current user; wherein the prediction model is trained using a data set consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information; the adversarial network includes an information generator and an information discriminator; the information generator is used to generate ticket purchase information, and the information discriminator is used to discriminate the accuracy of the ticket purchase information generated by the information generator; The training method of the adversarial network includes: Acquire the real ticket purchase information, and pre-process the real ticket purchase information; Based on the preprocessed real ticket purchase information, construct the information generator and the information discriminator; Calling the information generator to randomly generate a set of initial ticket purchase information; Shuffling the initial ticket purchase information with the real ticket purchase information to obtain a first data set; Dividing the first data set into a first training set and a first test set; Performing a round of iterative training on the current information discriminator using the first training set to obtain the latest current information discriminator; Using the first test set to test the current information discriminator, to obtain a discrimination probability value of the current information discriminator; Determine whether the current discrimination probability value of the information discriminator is greater than a first discrimination threshold, or whether the number of iterative training of the information discriminator reaches a first specified number of iterations; If it is determined that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold, and the number of iterative training of the information discriminator has not reached the first specified number of iterations, then returning to execute the iterative training of the current information discriminator using the first training set to obtain the latest current information discriminator; If it is determined that the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, or it is determined that the number of iterative training of the information discriminator reaches the first specified number of iterations, the current information discriminator is stored.

2. The method according to claim 1, characterized in that After storing the current information discriminator, the method further includes: Based on the current predefined learning rate, the current information generator is iteratively trained using the current information discriminator to obtain the latest current information generator; wherein, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information; Using the current information generator to randomly generate a set of ticket purchase information to be identified, and shuffle the ticket purchase information to be identified with the real ticket purchase information to obtain a current set of data to be identified; Using the current information discriminator to discriminate the current data set to be discriminated, and obtaining a discriminant value corresponding to the current data set to be discriminated; Determine whether the discrimination value corresponding to the current data set to be discriminated is greater than a second discrimination threshold; If it is determined that the discrimination value corresponding to the current data set to be discriminated is greater than the second discrimination threshold, the current information generator and the current information discriminator are stored; If it is determined that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold, then determining whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate; If it is determined that the number of iterative training of the information generator under the current predefined learning rate has not reached the second specified number of iterations, then return to execute the iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator; If it is determined that the number of iterative training of the information generator has reached the second specified number of iterations under the current predefined learning rate, the current predefined learning rate is updated to obtain the latest current predefined learning rate, and for the latest current predefined learning rate, the process returns to execute a round of iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator.

3. The method according to claim 2, characterized in that After determining that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold, the method further includes: Determine whether the total number of iterative training of the adversarial network is greater than the total number of predictions; If it is determined that the total number of iterative trainings for the adversarial network is greater than the total number of predictions, then storing the current information generator and the current information discriminator is executed; If it is determined that the total number of iterative training times for the adversarial network is not greater than the total number of predictions, then the determination is performed as to whether the number of iterative training times for the information generator reaches a second specified number of iterations under the current predefined learning rate.

4. The method according to claim 2, characterized in that: The training method of the prediction model comprises: Based on the real ticket purchase information after preprocessing, the prediction model is constructed; wherein the neural network structure of the prediction model is a deep neural network with temporal causality obtained by superimposing multiple stack structures; the stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on a double residual superposition method; Using the stored information generator to randomly generate a set of ticket purchase information, and confuse it with the real ticket purchase information to obtain a training data set; Generate a network diagram corresponding to the training data set according to the relationship between each piece of information in the training data set; Performing graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set; The graph features corresponding to the training data set form a time series graph which is input into the prediction model, and the prediction model is iteratively trained until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, and the trained prediction model is obtained and stored.

5. The method according to claim 1, characterized in that After obtaining the user's current pre-purchase information of the target air ticket, the method further includes: Preprocessing the pre-purchased ticket information to obtain the pre-purchased ticket information in a preset format; The step of generating a network diagram corresponding to the pre-purchased ticket information according to the relationship between each piece of information in the pre-purchased ticket information includes: According to the relationship between each piece of information in the pre-purchased ticket information in a preset format, a network diagram corresponding to the pre-purchased ticket information is generated.

6. A user willingness to pay prediction system, characterized in that: include: A first acquisition unit is used to acquire the user's current pre-purchase information of the target air ticket; wherein the pre-purchase information includes the attribute information of the target air ticket; A first generating unit, configured to generate a network graph corresponding to the pre-purchased ticket information according to the relationship between each information in the pre-purchased ticket information; A first convolution unit is used to perform graph convolution on the network graph of the pre-purchased ticket information to obtain graph features corresponding to the pre-purchased ticket information; A prediction unit, used for inputting the graph features corresponding to the pre-purchased ticket information into a pre-trained prediction model, analyzing and calculating the graph features corresponding to the pre-purchased ticket information through the prediction model, and obtaining the discount information of the target air ticket corresponding to the current user; wherein the prediction model is trained using a data set consisting of ticket purchase information generated by a trained adversarial network and real ticket purchase information; the adversarial network includes an information generator and an information discriminator; the information generator is used to generate ticket purchase information, and the information discriminator is used to discriminate the accuracy of the ticket purchase information generated by the information generator; A second acquisition unit, used to acquire the real ticket purchase information and pre-process the real ticket purchase information; A first construction unit, configured to construct the information generator and the information discriminator based on the real ticket purchase information after preprocessing; A calling unit, used for calling an information generator to randomly generate a set of initial ticket purchase information; A first processing unit, configured to shuffle the initial ticket purchase information with the real ticket purchase information to obtain a first data set; A data division unit, used for dividing the first data set into a first training set and a first test set; A first training unit, configured to perform a round of iterative training on the current information discriminator using the first training set to obtain the latest current information discriminator; A testing unit, configured to test the current information discriminator using the first test set to obtain a discrimination probability value of the current information discriminator; A first judging unit, used to judge whether the current discrimination probability value of the information discriminator is greater than a first discrimination threshold, or whether the number of iterative training of the information discriminator reaches a first specified number of iterations; A first returning unit is configured to return to the first training unit to perform a round of iterative training of the current information discriminator using the first training set to obtain the latest current information discriminator when the first judging unit judges that the discrimination probability value of the current information discriminator is not greater than the first discrimination threshold and the number of iterative training of the information discriminator has not reached the first specified number of iterations; The first storage unit is used to store the current information discriminator when the first judgment unit judges that the discrimination probability value of the current information discriminator is greater than the first discrimination threshold, or judges that the number of iterative training of the information discriminator reaches a first specified number of iterations.

7. The system according to claim 6, characterized in that Also includes: The second training unit is used to perform a round of iterative training on the current information generator using the current information discriminator based on the current predefined learning rate to obtain the latest current information generator; wherein, during the iterative training process, the information discriminator is used to discriminate the complete ticket purchase information generated by the information generator based on the input predefined feature information; A second processing unit is used to randomly generate a set of ticket purchase information to be determined by using the current information generator, and shuffle the ticket purchase information to be determined with the real ticket purchase information to obtain a current set of data to be determined; A discriminant unit, used to discriminate the current data set to be discriminated by using the current information discriminator, and obtain a discriminant value corresponding to the current data set to be discriminated; A second judgment unit, used to judge whether the discrimination value corresponding to the current data set to be judged is greater than a second discrimination threshold; A second storage unit, configured to store the current information generator and the current information discriminator when the second judging unit judges that the discrimination value corresponding to the current data set to be discriminated is greater than a second discrimination threshold; A third judgment unit is used to judge whether the number of iterative training of the information generator reaches a second specified number of iterations under the current predefined learning rate when the second judgment unit judges that the discrimination value corresponding to the current data set to be discriminated is not greater than the second discrimination threshold; A second returning unit is used to return to the second training unit to perform a round of iterative training on the current information generator based on the current predefined learning rate using the current information discriminator, to obtain the latest current information generator, when the third judging unit judges that the number of iterative training of the information generator under the current predefined learning rate has not reached the second specified number of iterations; An updating unit is used for updating the current predefined learning rate when the third judgment unit determines that the number of iterative training of the information generator has reached a second specified number of iterations under the current predefined learning rate to obtain the latest current predefined learning rate, and for the latest current predefined learning rate, returning to the second training unit to execute a round of iterative training of the current information generator based on the current predefined learning rate using the current information discriminator to obtain the latest current information generator.

8. The system according to claim 7, characterized in that The system further comprises a prediction model training unit, wherein the prediction model training unit comprises: A second construction unit is used to construct the prediction model based on the real ticket purchase information after preprocessing; wherein the neural network structure of the prediction model is a deep neural network with temporal causality obtained by superimposing multiple stack structures; the stack structure is obtained by stacking multiple layers of fully connected blocks of activation functions based on a double residual superposition method; A third processing unit, configured to randomly generate a set of ticket purchase information using the stored information generator, and confuse the generated set of ticket purchase information with the real ticket purchase information to obtain a training data set; A second generating unit, configured to generate a network graph corresponding to the training data set according to the relationship between each information in the training data set; A second convolution unit is used to perform graph convolution on the network graph of the training data set to obtain graph features corresponding to the training data set; The third training unit is used to input the timing diagram composed of the graph features corresponding to the training data set into the prediction model, and iteratively train the prediction model until the accuracy of the output result of the prediction model is greater than a preset threshold, or the iterative training reaches a third specified number of iterations, to obtain and store the trained prediction model.

9. An electronic device, characterized in that: include: one or more processing devices; a memory having one or more programs stored therein; When the one or more programs are executed by the one or more processing devices, the one or more processing devices implement the method for predicting user willingness to pay as described in any one of claims 1 to 5.

10. A computer storage medium, characterized in that: Used to store a program, which, when executed, is used to implement the method for predicting user willingness to pay as described in any one of claims 1 to 5.

Images