Personalized recommendation method for exposure fair representation learning

Abstract

The invention relates to a personalized recommendation method for exposure fair representation learning, and belongs to the technical field of recommendation systems and deep learning. The method comprises the following steps: acquiring e-commerce data, constructing an e-commerce data set, and dividing the e-commerce data set into a training set and a test set; constructing a personalized recommendation model of exposure fairness representation learning, wherein the model comprises an initial embedding module, an encoder learning module and an exposure fairness module; training a personalized recommendation model of exposure fair representation learning through training centralized e-commerce data; optimizing the model in the training process by using a loss function and an Adam optimizer to obtain a trained model; and inputting the e-commerce data in the test set into the trained model to obtain a personalized recommendation result. According to the method, personalized recommendation can be provided, the deviation in a recommendation system can be remarkably reduced, and the user experience and the exposure fairness of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the field of recommendation systems and deep learning technologies, and specifically relates to a personalized recommendation method for exposure fair representation learning. Background Art

[0002] With the rapid development of Internet technology, the surge in data volume has made information overload an inevitable challenge. In order to help users efficiently find the content they are interested in among the massive amount of information, recommendation systems have emerged. These systems use various algorithms to extract the most relevant information from huge amounts of data and recommend it to users to improve user experience and satisfaction. In recent years, many recommendation algorithms have been continuously innovated to meet the needs of different scenarios, including item-based collaborative filtering, neural network-based, graph models, variational autoencoders, and contrastive learning methods. The continuous advancement of these technologies has significantly improved the recommendation effect, making personalized recommendation systems widely used.

[0003] However, with the development and application of recommendation systems, fairness issues have gradually become prominent. In traditional recommendation systems, algorithms often focus on optimizing the accuracy of recommendations, while ignoring the fairness of recommendation results between user groups and items. This unfairness may lead to the phenomenon of information cocoons, limiting users' access to diverse information, thereby inhibiting their discovery of potential interests. In addition, the imbalance in exposure between items may affect the profits of underlying merchants, thereby having a negative impact on the platform's ecosystem. Therefore, how to ensure the exposure fairness of the system while improving the recommendation effect has become an important topic in the research of recommendation systems.

[0004] To address this challenge, exposure-fair representation learning has gradually become a research hotspot. This method reduces bias in the recommendation process by learning exposure-fair representations of users and items. However, most existing recommendation methods do not pay attention to item exposure fairness and lack consideration of exposure fairness requirements. In fact, different items need to be exposed more fairly, so in practical applications, recommendation systems need to strike a balance between personalization and exposure fairness.

[0005] In addition, recommendation systems also face the problem of data sparsity when achieving exposure fairness goals. This problem is particularly evident in the recommendation of unpopular items or minority users, as traditional methods are difficult to ensure fair exposure for these groups. In recent years, some studies have attempted to study the exposure fairness of recommendation systems by using exposure fairness indicators. However, these indicators are usually not differentiable, which limits their application in model training. Summary of the invention

[0006] In order to solve the above problems, the present invention provides a personalized recommendation method for learning exposure fair representation.

[0007] In order to achieve the above object, the present invention is implemented through the following technical solutions: The present invention provides a personalized recommendation method for exposure fair representation learning, comprising the following steps: S1. Obtain e-commerce data, build an e-commerce dataset, and divide it into training set and test set; S2. construct a personalized recommendation model for learning exposure fairness representation, the model comprising an initial embedding module, an encoder learning module and an exposure fairness module; train the personalized recommendation model for learning exposure fairness representation using e-commerce data in a training set; S3. Use the loss function and the Adam optimizer to optimize the model during the training process to obtain a trained model; S4. Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

[0008] Furthermore, step S1 specifically includes: The items interacted by users in the e-commerce dataset are divided according to the leave-one-out method, and the last interaction of each user is retained as a test set, and the rest of the data is used as a training set; The e-commerce data set includes M users constituting a user set U, N items constituting an item set I, and the interaction behaviors between the two constitute an interaction matrix Y.

[0009] Furthermore, the initial embedding module in step S2 specifically includes: In the training set, the position labels of users in set U and items in set I are set as user ID and item ID respectively, and the user ID and item ID are mapped into one-hot vector matrices respectively. and , each user and item is represented by a d-dimensional vector, and the user vector and item vector are obtained. The user vector and item vector are spliced ​​vertically in sequence to obtain the user embedding matrix and the item embedding matrix , and use the Xavier method to embed the user matrix and the item embedding matrix Initialize parameters; One-hot vector matrix mapped according to the user ID and the user embedding matrix , get the user's initial embedding, the calculation formula is as follows: , in, Representation Matrix The one-hot vector of the u-th user, Represents the dot product operation between a matrix and a vector, , represents the embedding dimension; One-hot vector matrix mapped according to the item ID and the item embedding matrix , get the initial embedding of the item, the calculation formula is as follows: , in, Representation Matrix The one-hot vector of the ith item, Represents the dot product operation between a matrix and a vector, .

[0010] Furthermore, the encoder learning module in step S2 specifically includes: The user's embedding at each layer is obtained through the graph encoder, and the formula is as follows: , in, Indicates that the user embeds The graph convolution result of the layer, , is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the item is embedded in The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the user embeddings of each layer to get the final user embedding , the formula is as follows: , in, is the total number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer; The embedding of the object at each layer is obtained through the graph encoder, and the formula is as follows: , in, Items embedded in The graph convolution result of the layer, is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the user embeds The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the item embeddings of each layer to get the final item embedding , the formula is as follows: , in, is the number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer.

[0011] Furthermore, the exposure fairness module in step S2 specifically includes: In the exposure fairness module, "pre-recommendation" is performed, through the final user embedding and item embedding After the vector dot multiplication operation, we get the preference score of user u for item i. The calculation formula is as follows: , in, Represents a vector dot product operation, Indicates the preference rating of user u for item i; When making a "pre-recommendation", the formula Calculate the k+1th largest score given by user u among all items. The item with a score greater than the k+1th largest score is the recommended item, where: Indicates the user's preference for all items. It means to find the kth largest value from the set; Map the user's rating to the item's exposure to get the exposure of item i to user u. The formula is as follows: , in, represents the rectified linear unit function, represents the sigmoid function, k represents the selection of the first k items to be used in calculating the index, represents the temperature coefficient; The exposure of the item i to the user u is summed up to obtain the total exposure of the item, which is expressed as follows: , in, Indicates the total exposure of the item; The exposure fairness index is calculated based on the total exposure of the item. The formula is as follows: , in, Represents the total exposure of all items. represents any differentiable exposure fairness indicator calculation function.

[0012] Furthermore, step S3 specifically includes: The total loss optimization model is used to obtain the trained model. The formula is as follows: , , , in, represents the BPR loss, represents the training set, , Indicates that the user is observed and items The set of interactions that exist between Indicates user Sampling unobserved items in the interaction set , Represents the Sigmoid activation function; represents fairness loss, , The set of derivable exposure fairness index calculation functions that represent the larger the function value, the more fair the recommendation system is, and the smaller the function value, the fairer the recommendation system is. Represents the calculated function value of the exposure fairness indicator function fair of the recommendation system; represents the total loss, Represents a hyperparameter.

[0013] The present invention also provides a system for applying the above-mentioned personalized recommendation method for learning exposure fair representation, comprising: Data acquisition module: used to acquire e-commerce data, build e-commerce data sets, and divide them into training sets and test sets; Model building module: used to build a personalized recommendation model for learning exposure fairness representation, the model includes an initial embedding module, an encoder learning module and an exposure fairness module; and train the personalized recommendation model for learning exposure fairness representation using e-commerce data in the training set; Model optimization module: used to optimize the model in the training process using the loss function and the Adam optimizer to obtain a trained model; Recommendation module: Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

[0014] The advantages of the present invention are: The present invention proposes a personalized recommendation method for learning exposure-fair representation, which consists of an encoder learning module and a fairness module. First, the encoder is used to learn the user embedding and item embedding in each behavior. On this basis, the user's preference for items is calculated, and the fairness loss that can be derived is calculated according to these preferences, which is used as part of the final loss to train the model. Finally, the user-item embedding becomes fairer after optimization, and this embedding is more exposure-fair for subsequent recommendation tasks. Therefore, the present invention can not only provide personalized recommendations, but also significantly reduce the deviation in the recommendation system, improve user experience and exposure fairness of the system. By introducing exposure fairness constraints in the process of user and item embedding, the over-recommendation of items is constrained, ensuring that the recommendation system reduces the unfair impact of exposure on different items without sacrificing the recommendation effect. The method uses a deep neural network to learn the embedding representation of users and items, and introduces exposure fairness constraints in the embedding process to ensure that the model has exposure-fair representation capabilities for each item when processing user behavior data. The embedding representation adjusted by exposure fairness is used for recommendation prediction, which effectively improves the overall performance and exposure fairness of the recommendation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0016] Figure 1 is a flow chart of the method of the present invention; Figure 2 is the exposure fairness and performance balance diagram of the present invention; Figure 3 To guide the comparison of different exposure fairness indicators of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Example 1 In this embodiment, Figure 1 As shown, the present invention provides a personalized recommendation method for learning exposure fair representation, and the specific steps include: S1. Obtain e-commerce data, build an e-commerce dataset, and divide it into training set and test set; Specifically, the items interacted by users in the e-commerce dataset are divided according to the leave-one-out method, the last interaction of each user is retained as a test set, and the remaining data is used as a training set; the e-commerce dataset includes M users constituting a user set U, N items constituting an item set I, and the interaction behaviors between the two constitute an interaction matrix Y.

[0019] S2. construct a personalized recommendation model for learning exposure fairness representation, the model comprising an initial embedding module, an encoder learning module and an exposure fairness module; train the personalized recommendation model for learning exposure fairness representation using e-commerce data in a training set; Specifically, the initial embedding module: In the training set, the position labels of users in set U and items in set I are set as user ID and item ID respectively, and the user ID and item ID are mapped into one-hot vector matrices respectively and , each user and item is represented by a d-dimensional vector, and the user vector and item vector are obtained. The user vector and item vector are spliced ​​vertically in sequence to obtain the user embedding matrix and the item embedding matrix , and use the Xavier method to embed the user matrix and the item embedding matrix Initialize parameters; One-hot vector matrix mapped according to the user ID and the user embedding matrix , get the user's initial embedding, the calculation formula is as follows: , in, Representation Matrix The one-hot vector of the u-th user, Represents the dot product operation between a matrix and a vector, , represents the embedding dimension; One-hot vector matrix mapped according to the item ID and the item embedding matrix , get the initial embedding of the item, the calculation formula is as follows: , in, Representation Matrix The one-hot vector of the ith item, Represents the dot product operation between a matrix and a vector, .

[0020] Specifically, the encoder learning module: The user’s embedding at each layer is obtained through the graph encoder, and the formula is as follows: , in, Indicates that the user embeds The graph convolution result of the layer, , is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the item is embedded in The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the user embeddings of each layer to get the final user embedding , the formula is as follows: , in, is the total number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer; The embedding of the object at each layer is obtained through the graph encoder, and the formula is as follows: , in, Items embedded in The graph convolution result of the layer, is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the user embeds The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the item embeddings of each layer to get the final item embedding , the formula is as follows: , in, is the number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer.

[0021] Specifically, the exposure fairness module: the “pre-recommendation” is performed in the exposure fairness module, and the final user embedding and item embedding After the vector dot multiplication operation, we get the preference score of user u for item i. The calculation formula is as follows: , in, Represents a vector dot product operation, Indicates the preference rating of user u for item i; When making a "pre-recommendation", the formula Calculate the k+1th largest score given by user u among all items. The item with a score greater than the k+1th largest score is the recommended item, where: Indicates the user's preference for all items. It means to find the kth largest value from the set; Map the user's rating to the item's exposure to get the exposure of item i to user u. The formula is as follows: , in, represents the rectified linear unit function, represents the sigmoid function, k represents the selection of the first k items to be used in calculating the index, represents the temperature coefficient; The exposure of the item i to the user u is summed up to obtain the total exposure of the item, which is expressed as follows: , in, Indicates the total exposure of the item; The exposure fairness index is calculated based on the total exposure of the item. The formula is as follows: , in, Represents the total exposure of all items. represents any differentiable exposure fairness indicator calculation function.

[0022] S3. Use the loss function and the Adam optimizer to optimize the model during the training process to obtain a trained model; Specifically, the total loss optimization model is used to obtain the trained model, and the formula is as follows: , , , in, represents the BPR loss, represents the training set, , Indicates that the user is observed and items The set of interactions that exist between Indicates user Sampling unobserved items in the interaction set , Represents the Sigmoid activation function; represents fairness loss, , The set of derivable exposure fairness index calculation functions that represent the larger the function value, the more fair the recommendation system is, and the smaller the function value, the fairer the recommendation system is. Represents the calculated function value of the exposure fairness indicator function fair of the recommendation system; represents the total loss, Represents a hyperparameter.

[0023] S4. Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

[0024] Example 2 This embodiment provides a system for applying the above-mentioned personalized recommendation method for learning exposure fair representation, including: Data acquisition module: used to acquire e-commerce data, build e-commerce data sets, and divide them into training sets and test sets; Model building module: used to build a personalized recommendation model for learning exposure fairness representation, the model includes an initial embedding module, an encoder learning module and an exposure fairness module; and train the personalized recommendation model for learning exposure fairness representation using e-commerce data in the training set; Model optimization module: used to optimize the model in the training process using the loss function and the Adam optimizer to obtain a trained model; Recommendation module: Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

[0025] Example 3 In this embodiment, we further introduce the Ml-1m dataset, one of the most classic datasets in the field of recommendation systems. First, the Ml-1m dataset is divided into a training set and a test set, and the data of the training set is used to construct the interaction matrix and the initial user-item embedding. Secondly, the encoder applied to the interaction matrix performs graph convolution on the initial user-item embedding, and the result of the graph convolution is used to obtain the embedding directly used to recommend user items. Then, the embedding is used to calculate the recommendation loss and exposure fairness loss respectively, and the model is optimized using the gradient descent method to obtain the final embedding. Finally, the final embedding is used to make recommendations and generate a recommendation list.

[0026] like Figure 2 and Figure 3, taking the LightGCN model as an example, we will conduct extensive parameter experiments on the LightGCN after introducing the module, and compare it with the original LightGCN model. In the two figures, REL and FAIR represent the accuracy evaluation index and exposure fairness index respectively, where {↑ / ↓}{type}@{k} represents the final result of the type index in the first k recommended items to the user, and the arrow indicates how the index changes better. The type of type can be hit for hit rate, precision for precision, recall for recall, ndcg for normalized discounted cumulative gain, Jain for Jain fairness index, Ent for entropy, Gini for Gini coefficient, and VoCD for customer voice data, where hit, precision, recall, and ndcg are accuracy indicators, and Jain, Ent, Gini, and VoCD are exposure fairness indicators. For example, ↑hit@10 represents the calculation result of the hit index in the first 10 recommended items to the user, and the larger the index, the better. There are three colors in the figure, namely red, green and gray. Red represents performance degradation, green represents performance improvement, and gray represents the loss improved by adding the module using that exposure fairness indicator.

[0027] Figure 2 In the figure, we introduce our module into LightGCN (taking Ent as an example) and conduct extensive parameter experiments. We select four representative experimental results, which are the best accuracy without considering the fairness of the model (i.e., the highest accuracy), the best fairness when ensuring 100% accuracy (ensuring 100% accuracy), the best fairness when ensuring 90% accuracy (ensuring 90% accuracy), and the best fairness when ensuring 80% accuracy (ensuring 80% accuracy). We conclude that by changing the training parameters, we can change the balance between exposure fairness and accuracy, thereby meeting different practical needs.

[0028] Figure 3 Different indicators (VoCD, Gini, Ent, Jain) are used to improve the performance of the LightGCN model. When we use the module, we will also improve the accuracy.

[0029] In summary, compared with the original LightGCN model, the present invention not only has a 2% improvement in accuracy and different degrees of exposure fairness in the best case, but also can meet the exposure fairness requirements of different situations, such as ensuring 90% recommendation performance to greatly improve exposure fairness. In addition, the present invention is well portable and can be transplanted to other models to improve exposure fairness, including but not limited to BPR, NeuMF, SGL, MultiVAE, SLIM and other models.

[0030] Example 4 In this example, we assume that we have a movie recommendation platform where users watch movies and interact with the movies. We first collect the user's viewing history and pre-process the data.

[0031] Then, we create user-item embeddings and design an encoder to capture the interaction information between users and movies, and use the exposure fairness module to ensure that user-item embeddings learn exposure information. For example, if Movie 1, Movie 2, and Movie 3 are recommended 5, 3, and 1 times respectively, Movie 3 is suffering from unfair exposure. Through our module, we can guide the model to give more exposure to Movie 3.

[0032] By optimizing the model, the exposure fairness loss is minimized to ensure that the recommendation results are more fair, while maximizing the information between the user's embedding and their movie preferences to ensure the accuracy of the recommendation results. The model is trained using the training dataset, the parameters of the sensitive attribute encoder and the interest encoder are optimized, and the performance of the model is further improved through cross-validation and hyperparameter tuning.

[0033] Finally, based on the user's exposure fairness embedding, the similarity between him and all the movies on the platform is calculated, and a personalized movie recommendation list is generated based on the similarity. For example, "Movie 1", "Movie 2" and "Movie 3" are recommended. Through the above steps, we can achieve a personalized movie recommendation based on exposure fairness representation learning, ensuring the fairness and accuracy of the recommendation results.

[0034] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A personalized recommendation method based on exposure fair representation learning, It is characterized in that The following steps are involved: S1. Obtain e-commerce data, build an e-commerce dataset, and divide it into training set and test set; S2. construct a personalized recommendation model for learning exposure fairness representation, the model comprising an initial embedding module, an encoder learning module and an exposure fairness module; train the personalized recommendation model for learning exposure fairness representation using e-commerce data in a training set; S3. Use the loss function and the Adam optimizer to optimize the model during the training process to obtain a trained model; S4. Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

2. A personalized recommendation method for learning exposure-fair representation according to claim 1, It is characterized in that Step S1 specifically includes: The items interacted by users in the e-commerce dataset are divided according to the leave-one-out method, and the last interaction of each user is retained as a test set, and the rest of the data is used as a training set; The e-commerce data set includes M users constituting a user set U, N items constituting an item set I, and the interaction behaviors between the two constitute an interaction matrix Y.

3. The personalized recommendation method for learning exposure fair representation according to claim 2, It is characterized in that The initial embedding module in step S2 specifically includes: In the training set, the position labels of users in set U and items in set I are set as user ID and item ID respectively, and the user ID and item ID are mapped into one-hot vector matrices respectively. and , each user and item is represented by a d-dimensional vector, and the user vector and item vector are obtained. The user vector and item vector are spliced ​​vertically in sequence to obtain the user embedding matrix and the item embedding matrix , and use the Xavier method to embed the user matrix and the item embedding matrix Initialize parameters; One-hot vector matrix mapped according to the user ID and the user embedding matrix , get the user's initial embedding, the calculation formula is as follows: ， in, Representation Matrix The one-hot vector of the u-th user, Represents the dot product operation between a matrix and a vector, , represents the embedding dimension; One-hot vector matrix mapped according to the item ID and the item embedding matrix , get the initial embedding of the item, the calculation formula is as follows: ， in, Representation Matrix The one-hot vector of the ith item, Represents the dot product operation between a matrix and a vector, .

4. The personalized recommendation method for learning exposure-fair representation according to claim 3, It is characterized in that The encoder learning module in step S2 specifically includes: The user's embedding at each layer is obtained through the graph encoder, and the formula is as follows: ， in, Indicates that the user embeds The graph convolution result of the layer, , is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the item is embedded in The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the user embeddings of each layer to get the final user embedding , the formula is as follows: ， in, is the total number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer; The embedding of the object at each layer is obtained through the graph encoder, and the formula is as follows: ， in, Items embedded in The graph convolution result of the layer, is the number of current graph convolution layers, L is the total number of graph convolution layers, Indicates that the user embeds The graph convolution result of the layer, Represents user A collection of interactive items. represents the set of users who interact with item i, Indicates the size of the collection; Aggregate the item embeddings of each layer to get the final item embedding , the formula is as follows: ， in, is the number of graph convolution layers, Indicates that the user embeds Graph convolution result of the layer.

5. The personalized recommendation method for learning exposure-fair representation according to claim 4, It is characterized in that The exposure fairness module in step S2 specifically includes: In the exposure fairness module, "pre-recommendation" is performed, through the final user embedding and item embedding After the vector dot multiplication operation, we get the preference score of user u for item i. The calculation formula is as follows: ， in, Represents a vector dot product operation, Represents the preference rating of user u for item i; When making a "pre-recommendation", the formula Calculate the k+1th largest score given by user u among all items. The item with a score greater than the k+1th largest score is the recommended item, where: Indicates the user's preference for all items. It means to find the kth largest value from the set; Map the user's rating to the item's exposure to get the exposure of item i to user u. The formula is as follows: ， in, represents the rectified linear unit function, represents the sigmoid function, k represents the selection of the first k items to be used in calculating the index, represents the temperature coefficient; The exposure of the item i to the user u is summed up to obtain the total exposure of the item, which is expressed as follows: ， in, Indicates the total exposure of the item; The exposure fairness index is calculated based on the total exposure of the item. The formula is as follows: ， in, Represents the total exposure of all items. represents any differentiable exposure fairness indicator calculation function.

6. The personalized recommendation method for learning exposure-fair representation according to claim 5, It is characterized in that Step S3 specifically includes: The total loss optimization model is used to obtain the trained model. The formula is as follows: ， ， ， in, represents the BPR loss, represents the training set, , Indicates that the user is observed and items The set of interactions that exist between Indicates user Sampling unobserved items in the interaction set , Represents the Sigmoid activation function; represents fairness loss, , The set of derivable exposure fairness index calculation functions that represent the larger the function value, the more fair the recommendation system is, and the smaller the function value, the fairer the recommendation system is. Represents the calculated function value of the exposure fairness indicator function fair of the recommendation system; represents the total loss, Represents a hyperparameter.

7. A system using the personalized recommendation method for learning exposure fair representation according to claim 1, It is characterized in that include: Data acquisition module: used to acquire e-commerce data, build e-commerce data sets, and divide them into training sets and test sets; Model building module: used to build a personalized recommendation model for learning exposure fairness representation, the model includes an initial embedding module, an encoder learning module and an exposure fairness module; and train the personalized recommendation model for learning exposure fairness representation using e-commerce data in the training set; Model optimization module: used to optimize the model in the training process using the loss function and the Adam optimizer to obtain a trained model; Recommendation module: Input the e-commerce data in the test set into the trained model to obtain personalized recommendation results.

Images