Product recommendation method and device for aviation journey and computer program product

Abstract

The invention discloses a product recommendation method and device for an aviation journey and a computer program product. The method comprises the steps that user information of a target user is acquired, and whether the target user is a new user without an aviation record or not is judged based on the user information; under the condition that the target user is a new user, a first recommended product is determined according to the similarity between the new user and other users in a target database, the first recommended product is recommended to the target user, and the other users are users except the target user in the target database; and when the target user is not the new user, obtaining a historical flight record of the target user, determining a second recommended product through the historical flight record, and recommending the second recommended product to the target user. Through the method and the device, the problem that products recommended to aviation passengers are inaccurate in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a product recommendation method, device, and computer program product for an air travel. Background Art

[0002] With the diversification of air travel demand and the intensification of competition in the aviation market, providing personalized services to passengers has become an important means for airlines to attract customers. By providing personalized products to passengers, airlines can not only increase their revenue, but also improve passengers' loyalty and satisfaction with the airline. By recommending products to customers through official channels or third-party channels, airlines can attract more one-way passengers to buy return tickets, thereby increasing ticket sales.

[0003] In the related art, the product recommendation method for air travel relies on sales staff to recommend products to all customers. However, this full product recommendation method has the problem that the recommended products do not meet customer needs, which is likely to cause adverse effects such as information interference and unsatisfactory experience to customers, and in serious cases may lead to customer loss of airlines.

[0004] Currently, no effective solution has been proposed to the problem of inaccurate product recommendations to airline passengers in related technologies. Summary of the invention

[0005] The main purpose of the present application is to provide a product recommendation method, device and computer program product for an air journey, so as to solve the problem of inaccurate product recommendations to air passengers in the related art.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a product recommendation method for an air travel is provided. The method comprises: obtaining user information of a target user, judging whether the target user is a new user without an air travel record based on the user information; if the target user is a new user, determining a first recommended product by the similarity between the new user and other users in a target database, and recommending the first recommended product to the target user, wherein the other users are users other than the target user in the target database; if the target user is not a new user, obtaining the historical flight record of the target user, determining a second recommended product by the historical flight record, and recommending the second recommended product to the target user.

[0007] Optionally, determining the first recommended product through the similarity between the new user and other users in the target database includes: extracting user features from user information to obtain first user features; matching associated users from the target database based on the first user features, wherein the associated users are users in the target database whose user features have the highest similarity with the first user features; and determining the associated user's historically recommended products as the first recommended products.

[0008] Optionally, associated users are determined in the following manner: multiple users are determined from a target database, and a first user feature of each user is obtained, wherein the first user feature includes at least one of the following: travel time, number of trips, passenger type, membership level, and transaction frequency; for each user, the cosine similarity between the user's first user feature and the first user feature of the target user is calculated to obtain a set of similarities; and the user corresponding to the maximum similarity in the set of similarities is determined as an associated user of the target user.

[0009] Optionally, after matching associated users from the target database based on the first user feature, the method also includes: if no associated user is matched in the target database, determining the sales volume of each product within a preset period; determining the product corresponding to the maximum sales volume as the first recommended product, and recommending the first recommended product to the target user.

[0010] Optionally, determining the second recommended product through historical flight records includes: extracting user features from the historical flight records to obtain second user features; inputting the second user features into a target model to obtain a second recommended product, wherein the target model is trained by multiple groups of training samples, and each group of training samples includes historical user features and historical recommended products.

[0011] Optionally, the target model is obtained by: obtaining the historical flight records of each user in the target database, and extracting the user's historical user features from the historical flight records; determining the historical recommended products of each user, and determining the historical user features and historical recommended products of each user as a set of training samples to obtain multiple sets of training samples; training a preset neural network model based on the multiple sets of training samples to obtain the target model.

[0012] Optionally, training a preset neural network model based on multiple groups of training samples to obtain a target model includes: iteratively training the preset neural network model through multiple groups of training samples, wherein each iterative training adjusts model parameters of the preset neural network model through a preset optimizer; in each iterative training, for each training sample, controlling the preset neural network model to process historical user characteristics of the training sample input to obtain an output result of the preset neural network model; determining a loss function of the preset neural network model, calculating a loss value based on the output results of multiple groups of training samples, historical recommended products and the loss function, and obtaining a loss value of the preset neural network model after each iterative training; when the loss value is less than or equal to a loss value threshold, determining the trained preset neural network model corresponding to the loss value as the target model.

[0013] Optionally, after obtaining the user information or historical flight records of the target user, the method further includes: eliminating missing values ​​from the user information or historical flight records to obtain data to be processed; determining abnormal values ​​from the data to be processed, and eliminating the abnormal values ​​from the data to be processed to obtain updated data to be processed; normalizing the updated data to be processed to obtain target data, and executing the step of extracting a first user feature from the user information based on the target data, or executing the step of extracting a second user feature from the historical flight records based on the target data.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a product recommendation device for an air journey is provided. The device comprises: an acquisition unit, used to acquire user information of a target user, and determine whether the target user is a new user without an air travel record based on the user information; a first recommendation unit, used to determine a first recommended product by the similarity between the new user and other users in a target database when the target user is a new user, and recommend the first recommended product to the target user, wherein the other users are users other than the target user in the target database; a second recommendation unit, used to acquire the historical flight record of the target user when the target user is not a new user, determine a second recommended product by the historical flight record, and recommend the second recommended product to the target user.

[0015] In order to achieve the above objective, according to another aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of the product recommendation method for an air trip described in each embodiment of the present application.

[0016] Through this application, the following steps are adopted: obtaining the user information of the target user, judging whether the target user is a new user without flight record based on the user information; in the case where the target user is a new user, determining the first recommended product through the similarity between the new user and other users in the target database, and recommending the first recommended product to the target user, wherein the other users are users other than the target user in the target database; in the case where the target user is not a new user, obtaining the historical flight record of the target user, determining the second recommended product through the historical flight record, and recommending the second recommended product to the target user, thereby solving the problem of inaccurate product recommendations to air passengers in the related art. For new users, the first recommended product is determined to be recommended to the target user through the similarity between the new user and other users, and for target users who are not new users, the second recommended product is determined to be recommended to the target user through the historical flight record of the target user, thereby achieving the effect of improving the accuracy of products recommended to air passengers. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 is a flow chart of a product recommendation method for an air travel provided in an embodiment of the present application;

[0019] Figure 2 is a schematic diagram of a method for calculating cosine similarity according to an embodiment of the present application;

[0020] Figure 3 is a schematic diagram of a training process of a target model provided according to an embodiment of the present application;

[0021] Figure 4 is a schematic diagram of an optional product recommendation method provided according to an embodiment of the present application;

[0022] Figure 5 is a schematic diagram of a product recommendation device for an air travel provided according to an embodiment of the present application;

[0023] Figure 6 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.

[0028] It should be noted that the information collected is 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 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.

[0029] For the convenience of description, some nouns or terms involved in the embodiments of the present application are explained below:

[0030] pandas: a data analysis package for Python (a programming language);

[0031] sklearn.preprocessing.StandardScaler(): A method to standardize data;

[0032] The present invention is described below in conjunction with preferred implementation steps. Figure 1 is a flow chart of a product recommendation method for an air travel provided in an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0033] Step S101, obtaining user information of a target user, and determining whether the target user is a new user without flight record based on the user information.

[0034] In step S101, the user information of the target user is retrieved from the data warehouse or database of the airline's business system. The user information may include the user's membership information, historical flight records, ticket purchase preferences, frequent flight routes, cabin selection habits, flight time preferences, etc. After obtaining the user information, the historical flight records of the target user are checked. If the target user does not have any travel records in the historical flight records, that is, the target user has no historical flights, it means that the target user is a new user with no aviation records.

[0035] For example, an API (Application Programming Interface) is developed in the airline's business system to query the relevant information of the target user from the user information table and historical flight data table in the database. A function or logical judgment statement is written to check the historical flight records of the target user. If the record is empty or there is no record within a specific time range (such as the past year), the user is determined to be a new user.

[0036] Step S102, when the target user is a new user, determine the first recommended product based on the similarity between the new user and other users in the target database, and recommend the first recommended product to the target user, wherein the other users are users in the target database other than the target user.

[0037] In step S102, after determining that the target user is a new user, the business system may activate a cold start recommendation strategy for the target user. The cold start recommendation strategy may be to recommend products based on the first user characteristics of the new user and the trend of the current aviation market (products may be various coupons, such as discount coupons, upgrade coupons, luggage coupons, meal coupons, and mileage coupons, etc.). For example, the business system may recommend the product with the highest recent sales volume as the first recommended product to the new user, or calculate the similarity between the new user and other users based on the new user's registration information (such as age, gender, registration source, etc.), and recommend the historically recommended products of other users to the new user as the first recommended product.

[0038] Step S103, when the target user is not a new user, obtain the historical flight record of the target user, determine the second recommended product through the historical flight record, and recommend the second recommended product to the target user.

[0039] In step S103, for old users who are not new users, the historical flight records of the target user are retrieved from the database. The historical flight records may include information such as flight date, departure place, destination, flight and cabin class, whether to upgrade, baggage check-in, dining options, etc. Based on the historical flight records, the travel patterns and preferences of the target user are analyzed. For example, if the user often takes flights of a certain airline, or prefers economy class, or often travels at specific times (such as holidays), this information can be used as a basis for product recommendations. Second user features are extracted from the historical flight records, such as destination preference, flight type preference, preference for a specific airline, cabin preference, whether to frequently apply for upgrades or baggage check-in, etc.

[0040] The extracted second user features are input into the trained target model. The target model can be a model based on machine learning, or a deep learning model such as a neural network model. The target model will output recommendation results based on the preferences and features of the target user. The target model can output multiple recommended products and their recommendation probabilities, and determine the second recommended product based on the output results of the target model. For example, the recommended products are sorted from large to small based on the recommendation probability, and the product ranked first is selected as the second recommended product. In some cases, if the product ranked first has been recommended or the user has used it, the next product in the sorted list can be selected as the second recommended product.

[0041] The product recommendation method for air travel provided in the embodiment of the present application obtains the user information of the target user, and determines whether the target user is a new user with no air travel record based on the user information; if the target user is a new user, the first recommended product is determined by the similarity between the new user and other users in the target database, and the first recommended product is recommended to the target user, wherein the other users are users other than the target user in the target database; if the target user is not a new user, the historical flight record of the target user is obtained, the second recommended product is determined by the historical flight record, and the second recommended product is recommended to the target user, thereby solving the problem of inaccurate product recommendations to air passengers in the related art. For new users, the first recommended product is determined by the similarity between the new user and other users and recommended to the target user. For target users who are not new users, the second recommended product is determined by the historical flight record of the target user and recommended to the target user, thereby achieving the effect of improving the accuracy of the products recommended to air passengers.

[0042] For a new user, a first recommended product is determined based on the similarity of user features. Optionally, in the product recommendation method for an air trip provided in an embodiment of the present application, determining the first recommended product by the similarity between the new user and other users in a target database includes: extracting user features from user information to obtain first user features; matching associated users from the target database based on the first user features, wherein the associated user is a user in the target database whose user features have the highest similarity to the first user features; and determining the associated user's historically recommended product as the first recommended product.

[0043] In some examples, user features are extracted from the user information of the new user. The user features may include age, gender, occupation, hobbies, registration source, travel time preference, number of premium class trips, number of economy class trips, passenger type, membership level, and transaction frequency. For example, Table 1 shows the first user features of the new user. These features can reflect the user's potential needs and preferences and provide a basis for subsequent similarity calculations. The extracted first user features are converted into numerical vector representations, such as converting category features into one-hot encoding (one-bit effective encoding), and the continuous features are standardized or normalized to ensure that the first user features are compared on the same scale.

[0044] Table 1

[0045]

[0046]

[0047] For each user in the target database (i.e., the airline's database), it is also necessary to extract its user features and generate a user feature vector. Then, by calculating the similarity between the first user feature vector of the new user and the user feature vector of each user in the database, the most matching associated user is found. The similarity calculation method may include cosine similarity, Pearson correlation coefficient, etc. After calculating the similarity of all users in the target database, the user with the highest similarity is selected as the associated user.

[0048] Obtain the historical recommendation records of the associated users, and select the product or service with the highest recommendation frequency and the best evaluation as the first recommended product from the historical recommendation records of the associated users. If there are multiple products that meet the conditions, you can select one or more for recommendation, or further filter according to the type, price, applicability, etc. of the product. Push the determined first recommended product to the new user, which can be done through email, application notification, SMS or other communication channels. The business system records this recommendation for subsequent analysis and optimization of the recommendation strategy.

[0049] If the new user has feedback on the recommended product, such as clicks, purchases, and reviews, this feedback information is collected and used to update the first user features and optimize the target model. The feedback mechanism can help the business system understand user needs and preferences more accurately, thereby making more personalized recommendations.

[0050] This embodiment recommends products by utilizing the similarity between new users and existing users and analyzing the preferences of users similar to the target user, taking into account the diversity and personalized needs of the user group, thereby improving the accuracy of recommended products and user satisfaction.

[0051] Associated users may be determined based on the cosine similarity between user features. Optionally, in the product recommendation method for an air trip provided in an embodiment of the present application, associated users are determined in the following manner: multiple users are determined from a target database, and a first user feature of each user is obtained, wherein the first user feature includes at least one of the following: travel time, number of trips, passenger type, membership level, and transaction frequency; for each user, the cosine similarity between the first user feature of the user and the first user feature of the target user is calculated to obtain a set of similarities; and a user corresponding to the maximum similarity in a set of similarities is determined as an associated user of the target user.

[0052] In some examples, Figure 2 is a schematic diagram of a method for calculating cosine similarity according to an embodiment of the present application, such as Figure 2 As shown, the similarity between users is calculated using cosine similarity matching. V1 represents the first user feature of the target user, V2 represents the first user features of other users in the database, and c represents the distance between the two vectors. The smaller the distance, the higher the user similarity.

[0053] After calculating the cosine similarity between all users and the new user, a set of similarity data will be obtained. This set of similarities is sorted and the user with the highest similarity is found as the associated user.

[0054] This embodiment determines the associated users by the cosine similarity between user features. Even if the new user has no historical data, it can find the most similar user as a reference by matching the similarity with the existing user features, thereby recommending products or services that meet the potential needs of the new user. It can provide fast and personalized product recommendations for new users.

[0055] For target users who are not matched with associated users, the product corresponding to the maximum sales volume can be used as the first recommended product. Optionally, in the product recommendation method for air journeys provided in an embodiment of the present application, after matching associated users from a target database based on the first user feature, the method further includes: when no associated users are matched in the target database, determining the sales volume of each product within a preset period; determining the product corresponding to the maximum sales volume as the first recommended product, and recommending the first recommended product to the target user.

[0056] In some examples, a similarity threshold is set based on expert experience. If the similarity between the user features of all users in the target database and the user features of the new user is less than the similarity threshold, it is determined that no associated user is matched in the target database. At this time, the sales volume data of each product in a preset period can be obtained from the target database. The preset period can be a week, a month, or a quarter. Perform statistical analysis on the collected sales data and calculate the total sales volume of each product. Sort the sales volume of all products from high to low. Select the product with the highest sales volume from the sorted list and determine it as the first recommended product.

[0057] This embodiment adopts a recommendation strategy based on product sales volume, which can ensure that new users at least receive market-proven and popular product recommendations. It is suitable for situations where new users do not have sufficient behavioral data, and it also helps to improve user acceptance and satisfaction with recommendations.

[0058] For non-new users, the second recommended product is determined based on historical flight records. Optionally, in the product recommendation method for an air journey provided in an embodiment of the present application, determining the second recommended product through historical flight records includes: extracting user features from the historical flight records to obtain second user features; inputting the second user features into a target model to obtain a second recommended product, wherein the target model is trained by multiple groups of training samples, and each group of training samples includes historical user features and historical recommended products.

[0059] In some examples, the second user features may include: travel time preferences (weekdays, holidays, specific time points, etc.), cabin preferences (economy class, business class, first class), flight type preferences (direct flights, transfers), frequent destinations, budget ranges (inferred through fare records), membership levels, whether cards, coupons, discount coupons, etc. are frequently used, baggage check-in preferences, and dining selection preferences. The extracted user features are preprocessed as necessary, such as encoding discrete features (for example, using one-hot encoding or label encoding) and scaling numerical features (for example, standardization or normalization) to ensure the format consistency of feature data and facilitate model processing. For example, Table 2 shows the second user features.

[0060] Table 2

[0061]

[0062]

[0063] The second user features extracted and preprocessed from historical flight records are input into the trained target model. The target model predicts based on the input second user features and outputs one or more recommended products. These products are the products that the user is most likely to be interested in based on historical behavior patterns and preferences. A product is selected from the prediction results of the target model as the second recommended product. The selection criteria can be based on the product's recommendation probability, historical sales records, inventory status, promotional activities, etc.

[0064] Through the above steps, even if the user is not a new user, this embodiment can use a machine learning model to provide personalized and refined second recommended products based on his or her historical flight records and behavioral characteristics, thereby improving user satisfaction and loyalty, while helping airlines or service providers optimize resource allocation and promotion strategies.

[0065] In order to predict the second recommended product for the target user, it is necessary to train the target model. Optionally, in the product recommendation method for the air journey provided in the embodiment of the present application, the target model is obtained in the following manner: obtaining the historical flight records of each user in the target database, and extracting the user's historical user features from the historical flight records; determining the historical recommended products for each user, determining the historical user features and historical recommended products of each user as a group of training samples, and obtaining multiple groups of training samples; training a preset neural network model based on the multiple groups of training samples to obtain the target model.

[0066] In some examples, Figure 3 is a schematic diagram of the training process of the target model provided in the embodiment of the present application, such as Figure 3 As shown, the historical flight records of all users (i.e., data sets) are obtained, the data sets are divided, and multiple groups of training samples are determined through feature processing and data preprocessing. The preset neural network model (i.e., algorithm) is trained with the multiple groups of training samples to obtain the target model.

[0067] Among them, the preset neural network model may include an embedding layer, a hidden layer, and an output layer. Each layer of the neural network is composed of multiple neurons. The embedding layer is used to receive features, and the number of neurons is consistent with the number of features. The hidden layer is to learn the nonlinear complex relationship between the features, and the output layer is used to output the processing results of the neural network on the input data and convert it into an understandable format. In this embodiment, four hidden layers can be set, each with 64 neurons, and the output layer is one layer, and the number of neurons is set to 32. Use normal distribution to randomly initialize the weights of this part, and the size is limited to [-0.01, 0.01]. In order to prevent the model from overfitting, L1 regularization (Lasso regularization, a penalty term used to control the complexity of the model) can also be added to this part.

[0068] This embodiment trains the target model and then predicts the second recommended product for old users based on the target model, thereby improving the accuracy of product recommendations.

[0069] After determining a plurality of groups of training samples, a target model is trained based on the plurality of groups of training samples. Optionally, in the product recommendation method for an air journey provided in an embodiment of the present application, a preset neural network model is trained based on the plurality of groups of training samples, and obtaining the target model comprises: iteratively training the preset neural network model through the plurality of groups of training samples, wherein each iterative training adjusts the model parameters of the preset neural network model through a preset optimizer; in each iterative training, for each training sample, controlling the preset neural network model to process the historical user features of the training sample input to obtain the output result of the preset neural network model; determining the loss function of the preset neural network model, calculating the loss value based on the output results of the plurality of groups of training samples, the historical recommended products and the loss function, and obtaining the loss value of the preset neural network model after each iterative training; and when the loss value is less than or equal to a loss value threshold, determining the trained preset neural network model corresponding to the loss value as the target model.

[0070] In some examples, first, the parameters of the preset neural network model are initialized. These parameters include but are not limited to weight matrix, bias term, learning rate, etc. Initialization can be random initialization, such as normal distribution or uniform distribution, or the parameters of the pre-trained model can be used. Select a preset optimizer, such as Adam (Adaptive Moment Estimation), SGD (Stochastic Gradient Descent), etc., to adjust the model parameters to minimize the loss function. The settings of the optimizer include learning rate, momentum, decay coefficient, etc.

[0071] For each training sample, historical user features are input into the neural network model, and forward propagation is performed to obtain the output of the model. Define the loss function (such as cross entropy loss, mean square error loss, etc.), and calculate the loss value based on the output of the model, the corresponding historical recommended products, and the loss function. The loss value reflects the gap between the model prediction and the actual historical recommended products. Using the selected optimizer, the gradient of the loss value to the model parameters is calculated through the back propagation algorithm, and then the model parameters are adjusted to reduce the loss value. By adjusting the model parameters, the prediction ability of the model is gradually improved and closer to the actual recommended products.

[0072] After each iteration of training, check whether the loss value is less than or equal to the preset loss value threshold. The loss value threshold is used to determine whether the model has been optimized enough to meet the predetermined performance standards. If after a certain iteration of training, the loss value meets the condition of being less than or equal to the loss value threshold, the preset neural network model trained at this time is used as the target model. After the training is completed, use an independent validation set to evaluate the target model and check the performance of the model, such as accuracy, recall, etc. If the model performance does not meet the requirements, you can adjust the model structure, parameter initialization method, optimizer settings, loss function, etc., and then re-iterate training.

[0073] Through the above steps, this embodiment enables the preset neural network model to learn the complex relationship between user characteristics and recommended products, thereby improving the accuracy and personalization of recommendations.

[0074] In order to facilitate the extraction of user features, the acquired user data needs to be preprocessed. Optionally, in the product recommendation method for an air journey provided in an embodiment of the present application, after acquiring the user information or historical flight records of the target user, the method further includes: removing missing values ​​from the user information or historical flight records to obtain data to be processed; determining outliers from the data to be processed, and removing the outliers from the data to be processed to obtain updated data to be processed; normalizing the updated data to be processed to obtain target data, and executing the step of extracting a first user feature from the user information based on the target data, or executing the step of extracting a second user feature from the historical flight records based on the target data.

[0075] In some examples, the pandas library is used to remove redundant data containing missing values ​​from the original data (i.e. user information or historical flight records), and then outliers in the data are removed. In order to make the data have the same feature measurement scale, the sklearn.preprocessing.StandardScaler() method is used to perform standard normalization on the updated data to be processed. To delete outliers, the mean and median of all feature data can be calculated first. According to the capping principle, if the distance between the data sample and the mean is greater than three times the standard deviation, the data is considered to be an outlier, and the sample data containing the outlier is directly deleted.

[0076] This embodiment can effectively clean and prepare the data by preprocessing the acquired user data, ensuring that the business system or target model can learn and predict based on accurate, consistent and high-quality data, thereby improving the accuracy of recommendations and user satisfaction.

[0077] According to another embodiment of the present application, an optional product recommendation method is also provided. Figure 4is a schematic diagram of an optional product recommendation method provided according to an embodiment of the present application, such as Figure 4 As shown, data is collected from the airline's passengers, and the data is preprocessed to collect passenger requests and determine whether the passenger is a new user. If the passenger is a new user, the cold start strategy is executed to recommend products recommended by the cold start strategy to the passenger. If the passenger is not a new user, the algorithm model is called to output a recommendation list of products based on user characteristics, and products are selected from the recommendation list to be recommended to the passenger.

[0078] This embodiment sets different recommendation strategies according to different users, recommends cards and coupons that meet the needs and preferences of passengers, realizes accurate recommendations, and improves passenger satisfaction and loyalty.

[0079] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0080] The embodiment of the present application also provides a product recommendation device for an air journey. It should be noted that the product recommendation device for an air journey in the embodiment of the present application can be used to execute the product recommendation method for an air journey provided in the embodiment of the present application. The product recommendation device for an air journey provided in the embodiment of the present application is introduced below.

[0081] Figure 5 Schematic diagram of a product recommendation device for an air travel provided according to an embodiment of the present application. Figure 5 As shown, the device comprises:

[0082] An acquisition unit 501 is used to acquire user information of a target user and determine whether the target user is a new user without a flight record based on the user information;

[0083] A first recommendation unit 502 is used to determine a first recommended product based on the similarity between the new user and other users in the target database when the target user is a new user, and recommend the first recommended product to the target user, wherein the other users are users in the target database other than the target user;

[0084] The second recommendation unit 503 is used to obtain the historical flight record of the target user when the target user is not a new user, determine the second recommended product through the historical flight record, and recommend the second recommended product to the target user.

[0085] The product recommendation device for air travel provided in the embodiment of the present application obtains user information of a target user through an acquisition unit 501, and determines whether the target user is a new user without an air travel record based on the user information; the first recommendation unit 502, when the target user is a new user, determines a first recommended product through the similarity between the new user and other users in a target database, and recommends the first recommended product to the target user, wherein the other users are users other than the target user in the target database; the second recommendation unit 503, when the target user is not a new user, obtains the historical flight record of the target user, determines a second recommended product through the historical flight record, and recommends the second recommended product to the target user, thereby solving the problem of inaccurate product recommendations to air passengers in the related art. For a new user, the first recommended product is determined through the similarity between the new user and other users and recommended to the target user; for a target user who is not a new user, the second recommended product is determined through the historical flight record of the target user and recommended to the target user, thereby achieving the effect of improving the accuracy of the products recommended to air passengers.

[0086] Optionally, in the product recommendation device for air travel provided in an embodiment of the present application, the first recommendation unit 502 includes: a first extraction module, used to extract user features from user information to obtain first user features; a matching module, used to match associated users from a target database based on the first user features, wherein the associated user is a user in the target database whose user features have the highest similarity to the first user features; and a first determination module, used to determine the associated user's historically recommended product as the first recommended product.

[0087] Optionally, in the product recommendation device for air travel provided in an embodiment of the present application, the matching module includes: a first determination submodule, used to determine multiple users from a target database and obtain a first user feature of each user, wherein the first user feature includes at least one of the following: travel time, number of trips, passenger type, membership level, and transaction frequency; a calculation submodule, used to calculate, for each user, the cosine similarity between the user's first user feature and the first user feature of the target user to obtain a set of similarities; and a second determination submodule, used to determine the user corresponding to the maximum similarity in a set of similarities as an associated user of the target user.

[0088] Optionally, in the product recommendation device for air travel provided in an embodiment of the present application, the first recommendation unit 502 also includes: a second determination module, used to determine the sales volume of each product within a preset period when no associated user is matched in the target database; a third determination module, used to determine the product corresponding to the maximum sales volume as the first recommended product, and recommend the first recommended product to the target user.

[0089] Optionally, in the product recommendation device for air travel provided in an embodiment of the present application, the second recommendation unit 503 includes: a second extraction module, used to extract user features from historical flight records to obtain second user features; an input module, used to input the second user features into a target model to obtain a second recommended product, wherein the target model is trained by multiple groups of training samples, and each group of training samples includes historical user features and historical recommended products.

[0090] Optionally, in the product recommendation device for air travel provided in the embodiment of the present application, the second recommendation unit 503 also includes: an acquisition module, used to obtain the historical flight records of each user in the target database, and extract the user's historical user features from the historical flight records; a fourth determination module, used to determine the historical recommended products of each user, and determine the historical user features and historical recommended products of each user as a group of training samples to obtain multiple groups of training samples; a training module, used to train a preset neural network model based on the multiple groups of training samples to obtain a target model.

[0091] Optionally, in the product recommendation device for air travel provided in the embodiment of the present application, the training module includes: a training submodule, which is used to iteratively train the preset neural network model through multiple groups of training samples, wherein each iterative training adjusts the model parameters of the preset neural network model through a preset optimizer; a control submodule, which is used to control the preset neural network model to process the historical user characteristics of the training sample input for each training sample during each iterative training, and obtain the output result of the preset neural network model; a third determination submodule, which is used to determine the loss function of the preset neural network model, calculate the loss value based on the output results of the multiple groups of training samples, the historical recommended products and the loss function, and obtain the loss value of the preset neural network model after each iterative training; and a fourth determination submodule, which is used to determine the trained preset neural network model corresponding to the loss value as the target model when the loss value is less than or equal to the loss value threshold.

[0092] Optionally, in the product recommendation device for air travel provided in an embodiment of the present application, the device also includes: a first elimination unit, used to eliminate missing values ​​from user information or historical flight records to obtain data to be processed; a first determination unit, used to determine abnormal values ​​from the data to be processed, and eliminate the abnormal values ​​from the data to be processed to obtain updated data to be processed; a processing unit, used to normalize the updated data to be processed to obtain target data, and execute the step of extracting the first user feature from the user information based on the target data, or, execute the step of extracting the second user feature from the historical flight records based on the target data.

[0093] The product recommendation device for air travel includes a processor and a memory. The acquisition unit 501, the first recommendation unit 502 and the second recommendation unit 503 are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions.

[0094] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the accuracy of the recommended products for airline passengers can be improved by adjusting kernel parameters.

[0095] The 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 RAM, and the memory includes at least one memory chip.

[0096] An embodiment of the present invention provides a computer-readable storage medium having a program stored thereon, and when the program is executed by a processor, a product recommendation method for an air travel is implemented.

[0097] An embodiment of the present invention provides a processor, which is used to run a program, wherein the program executes a product recommendation method for an airline trip when the program is run.

[0098] Figure 6 is a schematic diagram of an electronic device provided according to an embodiment of the present application. Figure 6 As shown, the electronic device 601 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: obtaining user information of the target user, judging whether the target user is a new user without flight records based on the user information; if the target user is a new user, determining the first recommended product by the similarity between the new user and other users in the target database, and recommending the first recommended product to the target user, wherein the other users are users other than the target user in the target database; if the target user is not a new user, obtaining the historical flight record of the target user, determining the second recommended product by the historical flight record, and recommending the second recommended product to the target user. The device in this article may be a server, a PC, a PAD, a mobile phone, etc.

[0099] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining user information of a target user, and determining whether the target user is a new user with no flight record based on the user information; if the target user is a new user, determining a first recommended product through the similarity between the new user and other users in a target database, and recommending the first recommended product to the target user, wherein the other users are users other than the target user in the target database; if the target user is not a new user, obtaining the target user's historical flight record, determining a second recommended product through the historical flight record, and recommending the second recommended product to the target user.

[0100] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0101] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0102] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0104] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0105] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0106] Computer readable media include 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 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.

[0107] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0108] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0109] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A product recommendation method for an air travel, characterized in that: include: Acquire user information of a target user, and determine whether the target user is a new user with no flight record based on the user information; In the case where the target user is the new user, determining a first recommended product according to the similarity between the new user and other users in the target database, and recommending the first recommended product to the target user, wherein the other users are users in the target database other than the target user; In the case that the target user is not the new user, the historical flight record of the target user is obtained, a second recommended product is determined based on the historical flight record, and the second recommended product is recommended to the target user.

2. The method according to claim 1, characterized in that Determining the first recommended product based on the similarity between the new user and other users in the target database includes: Extracting a user feature from the user information to obtain a first user feature; Matching associated users from the target database based on the first user feature, wherein the associated user is a user in the target database whose user feature has the highest similarity with the first user feature; The historically recommended product of the associated user is determined as the first recommended product.

3. The method according to claim 2, characterized in that The associated user is determined in the following manner: Determine a plurality of users from the target database, and obtain a first user feature of each user, wherein the first user feature includes at least one of the following: travel time, number of trips, passenger type, membership level, and transaction frequency; For each user, calculating the cosine similarity between the first user feature of the user and the first user feature of the target user to obtain a set of similarities; A user corresponding to the maximum similarity in the set of similarities is determined as an associated user of the target user.

4. The method according to claim 2, characterized in that: After matching the associated users from the target database based on the first user feature, the method further includes: If the associated user is not matched in the target database, determining the sales volume of each product within a preset period; The product corresponding to the maximum sales volume is determined as the first recommended product, and the first recommended product is recommended to the target user.

5. The method according to claim 1, characterized in that The second recommended product determined by the historical flight records includes: Extracting user features from the historical flight records to obtain second user features; The second user feature is input into a target model to obtain a second recommended product, wherein the target model is trained by multiple groups of training samples, and each group of training samples includes historical user features and historical recommended products.

6. The method according to claim 5, characterized in that The target model is obtained by: Acquire historical flight records of each user in the target database, and extract historical user features of the user from the historical flight records; Determine the historical recommended products for each user, determine the historical user features and historical recommended products of each user as a group of training samples, and obtain multiple groups of training samples; A preset neural network model is trained based on the multiple groups of training samples to obtain the target model.

7. The method according to claim 6, characterized in that Training a preset neural network model based on the multiple sets of training samples to obtain the target model includes: Iteratively training the preset neural network model using the multiple groups of training samples, wherein each iterative training adjusts the model parameters of the preset neural network model using a preset optimizer; In each iteration of training, for each training sample, controlling the preset neural network model to process the historical user features inputted by the training sample to obtain the output result of the preset neural network model; Determine the loss function of the preset neural network model, calculate the loss value based on the output results of the multiple groups of training samples, the historical recommended products and the loss function, and obtain the loss value of the preset neural network model after each iterative training; When the loss value is less than or equal to the loss value threshold, the trained preset neural network model corresponding to the loss value is determined as the target model.

8. The method according to claim 1, characterized in that After obtaining the user information or historical flight records of the target user, the method further includes: Eliminate missing values ​​from the user information or the historical flight records to obtain data to be processed; Determine an outlier from the data to be processed, and remove the outlier from the data to be processed to obtain updated data to be processed; The updated data to be processed is normalized to obtain target data, and based on the target data, a step of extracting a first user feature from the user information is performed, or a step of extracting a second user feature from the historical flight record is performed based on the target data.

9. A product recommendation device for an air travel, characterized in that: include: an acquisition unit, configured to acquire user information of a target user, and determine whether the target user is a new user without a flight record based on the user information; a first recommendation unit, configured to determine a first recommended product according to a similarity between the new user and other users in a target database, and recommend the first recommended product to the target user when the target user is the new user, wherein the other users are users in the target database other than the target user; The second recommendation unit is used to obtain the historical flight record of the target user when the target user is not the new user, determine the second recommended product through the historical flight record, and recommend the second recommended product to the target user.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the product recommendation method for an air journey according to any one of claims 1 to 8 is implemented.

Images