Cold start user recommendation method, system and equipment based on heterogeneous transfer learning

Abstract

The invention discloses a cold start user recommendation method, system and equipment based on heterogeneous transfer learning, and the method comprises the steps: combining data of a source domain and a target domain into an enhanced matrix through employing a feature enhancement technology, and then mapping the enhanced matrix to a potential feature space through employing an auto-encoder, and performing feature matching and distribution alignment processing in the space, and finally reconstructing user features. And then a solution model is established according to the reconstructed user features, that is, the solution can be solved through matrix feature decomposition, so that the recommendation precision of the cold start user is improved. In addition, the model is expanded to the neural network, the effectiveness of the model is verified through experiments on a plurality of real world data sets, and the performance indexes are excellent in performance. According to the method, the auxiliary source domain information can be effectively utilized to assist the recommendation task of the target domain, so that more accurate and personalized recommendation is provided for the cold start user.

Description

Technical Field

[0001] This application relates to the field of recommendation technologies, and in particular, to a cold-start user recommendation method, system, and device based on heterogeneous transfer learning. Background Art

[0002] Recommendation systems play a crucial role in e-commerce, social media, and content distribution platforms. They provide personalized recommendations by analyzing users' behaviors and preferences, thus enhancing the user experience. However, a long-standing challenge is how to handle cold-start users, that is, new users who have no historical interaction data in the system.

[0003] In related technologies, cold-start users lack historical interaction data, which makes it difficult for the recommendation model to learn their preferences. This limits the ability of the recommendation system to serve new users. At the same time, user characteristics come from different domains (such as social networks and purchase histories), and there may be distribution differences between these domains, making it difficult to directly fuse them. This heterogeneity increases the complexity of extracting useful information from different data sources. In addition, existing models are difficult to quickly adapt and provide recommendations when new users or new items appear. This indicates that existing systems have limitations in dealing with dynamically changing user preferences and item characteristics. Summary of the Invention

[0004] The objective of this application is to provide a cold-start user recommendation method, system, and device based on heterogeneous transfer learning to solve the problem of inaccurate recommendations for cold-start users in related recommendation technologies.

[0005] To achieve the above objective, this application provides the following technical solutions:

[0006] In a first aspect, a cold-start user recommendation method based on heterogeneous transfer learning provided by this application includes:

[0007] Obtain the first information data of a user, where the first information data of the user includes: social relationship data of warm users and cold-start users, and historical interaction data of warm users;

[0008] Perform feature enhancement processing on the first information data of the user to obtain the second information data of the user;

[0009] Perform space mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain the third information data of the user;

[0010] Construct a target recommendation model according to the third information data of the user;

[0011] Generate a recommendation list for cold-start users according to the target recommendation model.

[0012] Further, the feature enhancement processing of the first information data of the user to obtain the second information data of the user includes:

[0013] Data feature enhancement is performed by zero-padding means, and the formula is as follows:

[0014]

[0015] where X is the second information data of the user, U w is the social relationship data of warm users, H is the historical interaction data of warm users, U c is the social relationship data of cold start users.

[0016] Further, the second information data of the user is sequentially subjected to spatial mapping processing, feature matching processing, and distribution alignment processing using an autoencoder to obtain the third information data of the user, specifically as follows:

[0017] Use the autoencoder to map X of the second information data of the user from the original space to a low-dimensional latent space, and then obtain the unknown interactions of cold start users through the reconstruction of the autoencoder. Among them, the loss function in the construction of the spatial mapping process:

[0018]

[0019] s.t., W T W = I

[0020]

[0021] where I is the identity matrix, X is the second information data of the user, ∥·∥ F is the Euclidean norm, W is the mapping matrix, W T is the transpose of W, and Tr(·) is the trace of the mapping matrix;

[0022] Construct the loss function in the feature matching process:

[0023]

[0024] where is the loss function in feature matching, W s and W t are mapping matrices, and are the transposes of the mapping matrices, m is the number of items, and d is the feature dimension of the user's social relationship;

[0025] ​​

[0026]

[0027] n = n w + n c

[0028] wherein, is the loss function for distribution alignment processing, u i and h j are the i-th and j-th vectors of U and H respectively, n is the number of users, n w is the number of warm users, n c is the number of cold start users.

[0029] Furthermore, the target recommendation model is constructed according to the third information data of the user, and the specific formula is as follows:

[0030]

[0031] A = λB + γM - XX T

[0032] f(W) = Tr(W T AW) - Tr(Φ T (W T W - I))

[0033]

[0034] where λ and γ are hyperparameters used to balance the weights between the three loss functions; is the overall loss function, f(W) is the Lagrangian function, and Φ is a symmetric matrix.

[0035] Furthermore, the second information data of the user is subjected to space mapping processing, feature matching processing, and distribution alignment processing in sequence based on the autoencoder of the neural network to obtain the third information data of the user, specifically as follows:

[0036] Obtain the reconstructed data in the low-dimensional space The formula is as follows:

[0037]

[0038]

[0039] Y (l+1) = σ(W (l+1) Y (l) )

[0040]

[0041] X (0) = X

[0042] Among them, σ(·) is the non-linear ReLU activation function, is the network parameter between the (l) and (l+1) layers, X (l) and X (l+1) are the representations in the (l) and (l+1) layers respectively; q(·) is the encoder; W (l+1) is the network parameter between the (l) and (l+1) layers; p(·) is the decoder;

[0043] Construct the loss function in the spatial mapping processing, and the formula is as follows:

[0044]

[0045] Among them, g1(·,·) is the mean square error;

[0046] Construct the loss function in the feature matching processing, and the formula is as follows:

[0047]

[0048] Among them, g2(·,·) is the mean square error;

[0049] Construct the loss function in the distribution alignment processing, and the formula is as follows:

[0050]

[0051] Among them, g3(·,·) is the MMD loss.

[0052] Furthermore, construct the target recommendation model according to the third information data of the user, specifically as follows:

[0053] Construct the overall loss function of the target recommendation model, and the formula is as follows:

[0054]

[0055] Among them, is the overall loss function;

[0056] Calculate the gradient of the overall loss function; then update the parameters through the optimizer.

[0057] Furthermore, the method further includes:

[0058] Evaluate the results of the recommendation list, and the evaluation metrics include: calculation accuracy, recall rate, normalized discounted cumulative gain, and hit rate.

[0059] Further, evaluating the results of the recommendation list, the evaluation metrics include: calculation precision, recall rate, normalized discounted cumulative gain, and hit rate. The specific calculation formulas are as follows:

[0060] The calculation precision evaluation formula is as follows:

[0061]

[0062] where N is the number of cold start users, and R i is the true preferred item of the i-th cold start user, and is the top-K item list recommended for the i-th cold start user;

[0063] The recall rate evaluation formula is as follows:

[0064]

[0065] The normalized discounted cumulative gain evaluation formula is as follows:

[0066]

[0067] where r i is the relevance score of the item at position i, and IDCG is set to ensure that a perfect ranking results in a normalized discounted cumulative gain value of 1;

[0068] The hit rate evaluation formula is as follows:

[0069]

[0070] where HR@K is defined as the total number of correctly recommended items divided by the total number of all users' preferred items.

[0071] In a second aspect, the present application also provides a cold start user recommendation system based on heterogeneous transfer learning, including:

[0072] A data acquisition module, configured to acquire the first information data of the user, where the first information data of the user includes: social relationship data of warm users and cold start users, and historical interaction data of warm start users;

[0073] A feature matching module, configured to perform feature enhancement processing on the first information data of the user to obtain the second information data of the user;

[0074] A data processing module, configured to perform space mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain the third information data of the user;

[0075] A construction module for constructing a target recommendation model according to the third information data of the user;

[0076] A recommendation list generation module for generating a recommendation list for cold start users according to the target recommendation model.

[0077] In a third aspect, the present application also provides a computer electronic device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the cold start user recommendation method based on heterogeneous transfer learning described in any one of the above.

[0078] The beneficial effects of the cold start user recommendation method, system, and device based on heterogeneous transfer learning provided by the present application are as follows:

[0079] 1. By redefining the cold start user recommendation problem as a heterogeneous domain transfer learning problem and combining feature matching and distribution alignment processing methods, the proposed model effectively solves the cold start user recommendation problem and improves the performance and robustness of the recommendation system;

[0080] 2. The present application verifies the superior performance of the model in metrics such as precision, recall, NDCG, and hit rate on multiple real datasets, providing more accurate and personalized recommendation services for users.

[0081] 3. The present application verifies the superior performance of the model in metrics such as precision, recall, NDCG, and hit rate on multiple real datasets, providing more accurate and personalized recommendation services for users. Description of the Drawings

[0082] Figure 1 is a schematic flowchart of a cold start user recommendation method based on heterogeneous transfer learning in an embodiment of the present application;

[0083] Figure 2 is a schematic structural diagram of a cold start user recommendation system based on heterogeneous transfer learning in an embodiment of the present application;

[0084] Figure 3 is a schematic structural diagram of a computer electronic device in an embodiment of the present application. Detailed Embodiments

[0085] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the protection scope of the present application.

[0086] It should be noted that when an element is referred to as "fixed to" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. On the contrary, when an element is referred to as being "directly on" another element, there is no intermediate element. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0087] In this application, unless otherwise clearly defined and limited, the terms "installed", "connected", "coupled", "fixed" and other terms should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0088] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of this application, "a plurality of" means two or more unless otherwise specifically defined.

[0089] The terms used in one or more embodiments of this application are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this application. The singular forms of "a", "the" and "said" used in one or more embodiments of this application are also intended to include the plural forms unless the context clearly indicates otherwise.

[0090] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used in the specification of this template are for the purpose of describing specific embodiments only and are not intended to limit this application. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0091] It should be understood that although the terms first, second, etc. may be used in one or more embodiments of the present application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while".

[0092] Currently, traditional recommendation systems mainly rely on collaborative filtering and content-based filtering techniques. These methods perform well when dealing with users with rich historical data. However, for cold-start users, due to the lack of sufficient historical interaction information, it is difficult for these systems to generate accurate recommendations. In addition, existing systems also face challenges when dealing with heterogeneous data sources, such as social relationships and purchase histories, because they usually require a large amount of labeled data and complex feature engineering.

[0093] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. These specific embodiments below can be combined with each other, and the same or similar concepts or processes will not be repeated in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0094] Please refer to Figure 1 , a cold-start user recommendation method based on heterogeneous transfer learning provided by an embodiment of the present application includes at least the following steps:

[0095] S10. Obtain the first information data of the user, where the first information data of the user includes: social relationship data of warm users and cold-start users, and historical interaction data of warm users.

[0096] It can be understood that in this embodiment, the first information data of the user includes warm user data and cold-start user data. As a preferred embodiment, in this embodiment, 90% of the users are randomly selected for training, that is, warm users, and 10% of the users are randomly selected for testing, that is, cold-start users.

[0097] It should be noted that heterogeneous transfer learning aims to use knowledge from other source domains to assist the learning tasks of the target domain. The main challenge of domain adaptation is to reduce the domain gap between the source domain and the target domain. Existing work tends to use methods such as maximum mean discrepancy, optimal transport, and adversarial loss to estimate the distribution difference between the two domains.

[0098] S20. Perform feature enhancement processing on the first information data of the user to obtain the second information data of the user.

[0099] Specifically, in this example, data feature enhancement is performed by means of zero padding, and the formula is as follows:

[0100]

[0101] Among them, X is the second information data of the user, U w is the social relationship data of warm users, H is the historical interaction data of warm users, U c is the social relationship data of cold start users.

[0102] S30. Perform space mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain the third information data of the user.

[0103] Specifically, in the process of data processing, this application includes two processing methods, which are as follows:

[0104] First, use an autoencoder to perform space mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence to obtain the third information data of the user. The specific processing process is as follows:

[0105] S301. Space mapping processing: Use an autoencoder to map the feature-enhanced data from the original space to a low-dimensional latent space, and then obtain the unknown interactions of cold start users through the reconstruction of the autoencoder. Specifically: to obtain the unknown interactions of cold start users. Specifically:

[0106]

[0107] Among them, X is the feature-enhanced data, U w is the social relationship of warm users; H is the historical interaction of warm users, U c is the social relationship of cold start users, m is the number of items, d is the feature dimension of user social relationships, and n is the number of users.

[0108] First, define the mapping matrix: The goal is to learn a mapping matrix to obtain the reconstructed [[ID=4)]Therefore, indirectly obtain the unknown interactions of cold start users, and these values are regarded as the ratings of users for items, and recommendations are made based on these ratings.

[0109] Secondly, construct the reconstruction loss function: The goal is to minimize the square of the Euclidean (Frobenius) norm between the reconstructed and the original to retain the rich information of warm users in the

[0110]

[0111] s.t., W T W = I

[0112] where I is the identity matrix, X is the second information data of the user, ∥·∥ F is the Euclidean norm, and W is the mapping matrix, W T is the transpose of W; it should be noted that there are two reasons for imposing the orthogonality constraint on W. First, this constraint can avoid the degeneracy phenomenon, such as all transformed samples collapsing to the origin. Second, this constraint ensures that there is no redundant information between the dimensions of the transformed data.

[0113] Finally, optimize the loss function for reconstructing X: According to the definition of the Euclidean (Frobenius) 2-norm, the loss function for reconstructing X can be optimized as:

[0114]

[0115] where Tr(·) is the trace of the mapping matrix.

[0116] S302. Feature matching processing: In the latent space, perform feature matching on the social relationships and historical interactions of warm users to reduce the differences between the two representations.

[0117] First, the definition of feature matching is as follows: For the social relationships and historical interactions of warm users, although they are heterogeneous, it is necessary to seek to minimize the differences between them in the latent space;

[0118] Second, define two transformation mappings: The goal is to use two mapping matrices and to map the social relationships U w of warm users and their historical interactions H into a potential common space to make them similar to each other;

[0119] Immediately afterwards, construct the feature matching loss function: The goal is to minimize the square of the Euclidean (Frobenius) norm in the latent space of the social relationships and historical interactions transformed by W s and W t Therefore, we have the following feature matching loss function:

[0120]

[0121] Finally, optimize the feature matching loss function: First, combine W s and W t into W, and then according to the definition of the Euclidean (Frobenius) 2-norm, the feature matching loss function can be optimized as:

[0122]

[0123] Among them, is the loss function in feature matching, and W s and W t are mapping matrices. and is the transpose of the mapping matrix.

[0124] S303. Distribution alignment processing: Use the Maximum Mean Discrepancy (MMD) method to estimate the distribution difference between the source domain (user social relationship U) and the target domain (user historical interaction H), and minimize these differences to reduce the gap between domains.

[0125] First, the definition of distribution alignment is as follows: For the user social relationship U (including warm users U w and cold start users U c ), and the user historical interaction H are representations from two different feature spaces, and they exhibit heterogeneous properties. Therefore, regard the user social relationship U as the source domain and the user historical interaction H as the target domain. Use MMD for heterogeneous transfer learning, in other words, reduce the distribution difference between the source domain and the target domain in the latent space.

[0126] Second, define two mapping matrices: The goal is to use two mapping matrices and respectively to minimize the distribution difference between the user social relationship U (source domain) and its historical interaction H (target domain) in the latent space and make them closer.

[0127] Immediately afterwards, construct the distribution alignment loss function: The goal is to use MMD to minimize the distribution difference between the user social relationship U (source domain) and the historical interaction H (target domain) in the latent space. Therefore, the distribution alignment loss function is as follows:

[0128]

[0129] where u i and h j are the i-th and j-th vectors of U and H respectively, n is the number of users, and n w is the number of warm users.

[0130] Finally, optimize the distribution alignment loss function: First combine W s and W t into W, and then according to the definition of the Euclidean (Frobenius) 2-norm, the distribution alignment loss function can be optimized to:

[0131]

[0132] Among them, is the loss function for distribution alignment processing.

[0133] Second, use the autoencoder based on neural network to perform spatial mapping processing, feature matching processing and distribution alignment processing on the second information data of the user in sequence to obtain the third information data of the user. The specific processing process is as follows:

[0134] S311. The spatial mapping process is as follows: First, encode the feature-expanded X through a layer of neural network, as follows:

[0135]

[0136] where σ(·) is the non-linear ReLU activation function, is the network parameter between the (l) and (l + 1) layers, X (l) and X (l+1) are the representations in the (l) and (l + 1) layers respectively, and X (0) = X.

[0137] Secondly, use a multi-layer neural network to encode the feature-expanded X to obtain a latent representation Y with dimension K, as follows:

[0138]

[0139] where q(·) is an encoder including multiple layers of S6.1;

[0140] Immediately afterwards, decode the latent representation Y through a layer of neural network, as follows:

[0141] Y (l+1) = σ(W (l+1) Y (l) )

[0142] where, W (l+1) is the network parameter between the (l) and (l + 1) layers.

[0143] Then, use a multi-layer neural network to decode the latent representation Y, and finally obtain the reconstruction

[0144]

[0145] where p(·) is a decoder including multiple layers of S6.3.

[0146] Finally, make the reconstructed data as similar as possible to the original data X, and the reconstruction loss function is defined as follows:

[0147]

[0148] where g1(·, ·) is the mean squared error.

[0149] S312. The feature matching is processed as follows: Calculate the feature matching loss based on the latent representation Y, so that the social relationship U w of warm users and their historical interactions H in the latent representation are similar, and the implementation is as follows:

[0150]

[0151] where g2(·, ·) is the mean squared error.

[0152] S313. The distribution alignment is processed as follows: Calculate the distribution alignment loss based on the latent representation Y by computing MMD, so that the distribution difference between the heterogeneous user social relationship U (source domain) and its historical interaction H (target domain) in the latent representation is reduced, and the implementation is as follows:

[0153]

[0154] where g3(·, ·) is the MMD loss.

[0155] S40. Construct a target recommendation model according to the third information data of the user.

[0156] It can be understood that in this embodiment, the third information of the user includes two types: The first type is the third information data of the user corresponding to the data processing using the autoencoder in step S30, and the details are as follows:

[0157] S401. Overall objective function and optimization: Ensure the performance of the recommendation system when processing cold-start users, combine the autoencoder conversion, feature matching, and distribution alignment into an overall objective function, and solve the overall objective function through Lagrange;

[0158] Overall loss function: In the comprehensive steps, design the overall loss function of the model:

[0159]

[0160] where λ and γ are hyperparameters used to balance the weights between the three loss functions;

[0161] S402. Rewrite the overall loss function: Combine XX T after feature augmentation, B in feature matching, and M in distribution alignment into A, and it can be rewritten as:

[0162]

[0163] where A = λB + γM - XX T ;

[0164] S403. Transform into the Lagrangian function: By introducing the symmetric matrix Φ as the Lagrange multiplier, the Lagrangian function is as follows:

[0165] f(W) = Tr(W T AW) - Tr(Φ T (W T W - I))

[0166] S404. Take the derivative of the Lagrangian function: Take the derivative of the Lagrangian function and set its derivative to zero

[0167]

[0168] S405. Transform into an eigenvalue problem: The equation AW = WΦ indicates that the column vectors of W are the eigenvectors of matrix A, and Φ is related to the eigenvalues. Therefore, the problem is transformed into solving the k smallest eigenvectors of matrix A.

[0169] The second is to process the third information data of the corresponding user by using the neural network-based autoencoder in step S30, specifically as follows:

[0170] S411. Construct the overall loss function:

[0171]

[0172] Among them, is the overall loss function.

[0173] S412. Calculate the gradient of the overall loss function;

[0174] S413. Update the parameters through an optimizer (such as Adam, SGD, etc.), and then obtain the local optimal solution.

[0175] S50. Generate a recommendation list for cold-start users according to the target recommendation model.

[0176] In this embodiment, a recommendation list for cold-start users is generated based on the autoencoder (FDMA) and the neural network-based autoencoder (D-FDMA) respectively.

[0177] Exemplarily, taking the autoencoder to generate a recommendation list as an example:

[0178] S501. Obtain the reconstructed data (i.e., the third information data of the user), and the model learns the mapping matrix W, specifically as follows

[0179]

[0180] Among them, and respectively represent the reconstructed social relationships of warm users, the reconstructed social relationships of cold-start users, and the historical interactions of warm users.

[0181] S502. Obtain the ratings of items by cold-start users: Obtain the ratings of items by cold-start users through the sigmoid activation function:

[0182]

[0183] S503. Generate a recommendation list: Obtain the ratings of items by cold-start users for generating top-k recommendations, and each cold-start user gets the corresponding k items with the highest ratings.

[0184] In one embodiment of the present application, the cold-start user recommendation method based on heterogeneous transfer learning further includes:

[0185] S60. Evaluate the results of the recommendation list, and the evaluation metrics include: computing precision, recall rate, normalized discounted cumulative gain, and hit rate.

[0186] Specifically, the following metrics are used to evaluate the performance of the recommendation system:

[0187] S601. Precision: The proportion of correctly recommended items (i.e., the preferred items existing in the recommendation list) in the total number of all recommended items, averaged over all samples:

[0188]

[0189] where N is the number of cold-start users, R i is the true preferred item of the i-th cold-start user, and is the list of top-K items recommended for the i-th cold-start user;

[0190] S602. Recall rate: The recall rate is the proportion of the number of correctly recommended items in the total number of items that should be recommended, that is, the number of preferred items, and can be calculated by the following formula:

[0191]

[0192] S603. Normalized Discounted Cumulative Gain (NDCG): A recommendation quality metric that considers the item ranking. For NDCG, we calculate NDCG@K as:

[0193]

[0194] where r iis the relevance score of the item at position i. IDCG is set to ensure that a perfect ranking results in an NDCG (Normalized Discounted Cumulative Gain) value of 1. Specifically, we set the r of the preferred item i to 1 and the others to 0;

[0195] S604. Hit Rate: The proportion that the recommended list contains at least one item actually preferred by the user. HR@K is defined as the total number of correctly recommended items divided by the total number of all user-preferred items:

[0196]

[0197] It should be noted that this application uses the Flickr dataset for the user cold-start recommendation problem. Flickr contains data collected from a popular personal photo management and sharing website. Users in this system can add tags to photos and subscribe to photos using tags. In our experiment, 195 interest tags were selected as items, and 8000 users were selected to construct the learning task. Similar to BlogCatalog, user features are generated from their social networks;

[0198] Compare the Top-10 prediction results in Flickr with JCTN (Tingting Liang, Congying Xia, Yuyu Yin, and Philip S Yu. 2020. Joint Training Capsule Network for Cold Start Recommendation. In ACM SIGIR Conference on Research and Development in Information Retrieval. 1769–1772.), AADA (Hanrui Wu, Jinyi Long, Nuosi Li, Dahai Yu, and Michael K Ng. 2023. Adversarial Auto-encoder Domain Adaptation for Cold-start Recommendation with Positive and Negative Hypergraphs. ACM Transactions on Information Systems 41, 2 (2023), 1–25.), and FMM (Hanrui Wu, Nuosi Li, Ka Ho Kwok, Xuheng Cai, Jia Zhang, Jinyi Long, and Michael K. Ng. 2024. Feature Matching Machine for Cold-Start Recommendation. IEEE Transactions on Services Computing 17, 1 (2024), 98–112.);

[0199] The experimental results are shown in Table 1. Algorithm 1 and Algorithm 2 respectively correspond to the verification results FDMA and D-FDMA of the algorithm proposed in the present invention; Algorithm 3 corresponds to the verification result of JCTN; Algorithm 4 corresponds to the verification result of AADA; Algorithm 5 corresponds to the verification result of FMM. It can be seen from Table 1 that the present invention is better than other algorithms in most evaluation metrics by considering the label-related performance;

[0200] Table 1

[0201] Precision Recall NDCG Hit Rate Algorithm 1 19.42±0.4 61.39±1.3 51.74±1.1 61.47±1.2 Algorithm 2 19.12±0.2 59.89±0.5 50.51±0.8 60.20±0.8 Algorithm 3 11.12±0.3 36.82±1.4 29.30±0.7 35.19±1.0 Algorithm 4 14.91±0.3 47.94±0.9 37.53±1.6 47.10±1.0 Algorithm 5 16.56±0.4 52.16±1.1 41.22±1.0 52.10±1.4

[0202] A cold-start user recommendation method for heterogeneous transfer learning provided by this application has the beneficial effects that:

[0203] 1. By redefining the cold-start user recommendation problem as a heterogeneous domain transfer learning problem and combining feature matching and distribution alignment processing methods, the proposed model effectively solves the cold-start user recommendation problem and improves the performance and robustness of the recommendation system.

[0204] 2. This application verifies the superior performance of the model in metrics such as precision, recall, NDCG, and hit rate on multiple real-world datasets, providing more accurate and personalized recommendation services for users.

[0205] 3. This application verifies the superior performance of the model in metrics such as precision, recall, NDCG, and hit rate on multiple real-world datasets, providing more accurate and personalized recommendation services for users.

[0206] Please refer to Figure 2 , this application also provides a cold-start user recommendation system 200 based on heterogeneous transfer learning, including:

[0207] A data acquisition module 201, configured to acquire the first information data of a user, where the first information data of the user includes: social relationship data of warm users and cold-start users, and historical interaction data of warm-start users;

[0208] A feature matching module 202, configured to perform feature enhancement processing on the first information data of the user to obtain second information data of the user;

[0209] A data processing module 203, configured to perform space mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain third information data of the user;

[0210] A construction module 204, configured to construct a target recommendation model according to the third information data of the user;

[0211] A recommendation list generation module 205, configured to generate a recommendation list for cold-start users according to the target recommendation model.

[0212] Please refer to Figure 3 , an embodiment of this application also provides a computer electronic device 300, including a memory 303 and a processor 302. The memory 303 stores a computer program, and when the processor executes the computer program, it implements the steps of the cold-start user recommendation method based on heterogeneous transfer learning described in any one of the above.

[0213] Specifically, the electronic device 300 includes a transceiver 301, a bus interface, and a processor 302. The processor 302 is configured to obtain first information data of a user. The first information data of the user includes social relationship data of warm users and cold start users and historical interaction data of warm users. The processor 302 performs feature enhancement processing on the first information data of the user to obtain second information data of the user. The processor 302 performs spatial mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain third information data of the user. The processor 302 constructs a target recommendation model according to the third information data of the user. The processor 302 generates a recommendation list for cold start users according to the target recommendation model.

[0214] In an embodiment of the present application, the electronic device 300 further includes a memory 303. In Figure 3 this case, the bus architecture may include any number of interconnected buses and bridges. Specifically, various circuits of one or more processors represented by the processor 302 and the memory represented by the memory 303 are linked together. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art. Therefore, they will not be further described herein. The bus interface provides an interface. The transceiver 301 may be multiple components, that is, including a transmitter and a receiver, and provides a unit for communicating with various other devices on a transmission medium. The processor 302 is responsible for managing the bus architecture and general processing. The memory 303 may store data used by the processor 302 when performing operations.

[0215] In this embodiment, the computer-readable storage medium may be a non-volatile storage medium or a volatile storage medium. For example, the computer storage medium may include, but is not limited to, a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc., which are various media that can store program codes.

[0216] In all the examples shown and described herein, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0217] It should be noted that like reference numerals and letters denote like items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0218] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the structure diagram and / or flowchart, as well as the combination of blocks in the structure diagram and / or flowchart, can be implemented by a dedicated hardware-based system that executes the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0219] In addition, in each embodiment of this application, each functional module or unit can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0220] If the above functions are implemented in the form of software function modules and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a terminal device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of this application.

[0221] As described above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application.

Claims

1. A cold start user recommendation method based on heterogeneous transfer learning, characterized in that Including: Obtain the first information data of the user, where the first information data of the user includes: social relationship data of warm users and cold start users, and historical interaction data of warm users; Perform feature enhancement processing on the first information data of the user to obtain the second information data of the user; Perform spatial mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain the third information data of the user; Construct a target recommendation model based on the third information data of the user; Generate a recommendation list for cold start users according to the target recommendation model.

2. The cold start user recommendation method based on heterogeneous transfer learning according to claim 1, characterized in that, The performing feature enhancement processing on the first information data of the user to obtain the second information data of the user includes: Perform data feature enhancement by means of zero padding, and the formula is as follows: Among them, X is the second information data of the user, U w is the social relationship data of the warm user, and H is the historical interaction data of the warm user, U c is the social relationship data of the cold start user.

3. The cold start user recommendation method based on heterogeneous transfer learning according to claim 1, characterized in that The performing spatial mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder to obtain the third information data of the user is specifically as follows: Use an autoencoder to map the X of the second information data of the user from the original space to a low-dimensional latent space, and then reconstruct it through the autoencoder to obtain the unknown interactions of cold-start users. Among them, the loss function in the construction of the space mapping process is as follows: s.t., where I is the identity matrix, X is the second information data of the user, ∥·∥ F is the Euclidean norm, W is the mapping matrix, W T is the transpose of W, and Tr(·) is the trace of the mapping matrix; Construct the loss function in the feature matching process: Among them, is the loss function in feature matching, W s and W t are mapping matrices, and is the transpose of the mapping matrix, m is the number of items, and d is the feature dimension of the user's social relationship; Construct the loss function in the distribution alignment process: n = n w +n c Among them, is the loss function for distribution alignment processing, u i and h j are the i-th and j-th vectors of U and H respectively, n is the number of users, n w is the number of warm users, n c is the number of cold start users.

4. The cold start user recommendation method based on heterogeneous transfer learning according to claim 3, characterized in that, The constructing a target recommendation model based on the third information data of the user is specifically as follows: Among them, λ and γ are hyperparameters used to balance the weights among the three loss functions; is the overall loss function, f(W) is the Lagrangian function, and Φ is a symmetric matrix.

5. The cold start user recommendation method based on heterogeneous transfer learning according to claim 1, characterized in that The performing spatial mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder based on a neural network to obtain the third information data of the user is specifically as follows: Obtain the reconstructed data in the low-dimensional space The formula is as follows: Y (l+1) = σ(W (l+1) Y (l) ) X (0) = X where, σ(·) is the non-linear ReLU activation function, are the network parameters between layers (l) and (l+1), X (l) and X (l+1) are the representations in layers (l) and (l+1) respectively; q(·) is the encoder; W (l+1) are the network parameters between layers (l) and (l+1); p(·) is the decoder; Construct the loss function in the spatial mapping process, and the formula is as follows: where g1(·,·) is the mean square error; Construct the loss function in the feature matching process, and the formula is as follows: where g2(·,·) is the mean square error; Construct the loss function in the distribution alignment process, and the formula is as follows: where g3(·,·) is the MMD loss.

6. The cold start user recommendation method based on heterogeneous transfer learning according to claim 5, characterized in that The constructing a target recommendation model based on the third information data of the user is specifically as follows: Construct the overall loss function of the target recommendation model, and the formula is as follows: l(q,p) = g1 + λg2 + γg3 where l(q,p) is the overall loss function; Calculate the gradient of the overall loss function; then update the parameters through an optimizer.

7. The cold start user recommendation method based on heterogeneous transfer learning according to claim 1, wherein, The method further includes: Evaluate the result of the recommendation list, and the evaluation metrics include: computational precision, recall rate, normalized discounted cumulative gain, and hit rate.

8. The cold start user recommendation method based on heterogeneous transfer learning according to claim 7, characterized in that The evaluating the result of the recommendation list, and the evaluation metrics include: computational precision, recall rate, normalized discounted cumulative gain, and hit rate, and the specific calculation formulas are as follows: The computational precision evaluation formula is as follows: Among them, N is the number of cold start users, and R i is the true preference item of the i-th cold start user, while is the list of top-K items recommended for the i-th cold start user; The recall rate evaluation formula is as follows: The normalized discounted cumulative gain evaluation formula is as follows: where r i is the relevance score of the item at position i, and IDCG is set to ensure that a perfect ranking results in a normalized discounted cumulative gain value of 1; The hit rate evaluation formula is as follows: where HR@K is defined as the total number of correctly recommended items divided by the total number of all user-preferred items.

9. A cold-start user recommendation system based on heterogeneous transfer learning, characterized in that, Including: A data acquisition module for obtaining the first information data of the user, where the first information data of the user includes: social relationship data of warm users and cold start users, and historical interaction data of warm start users; A feature matching module, configured to perform feature enhancement processing on the first information data of the user to obtain the second information data of the user; A data processing module, configured to perform spatial mapping processing, feature matching processing, and distribution alignment processing on the second information data of the user in sequence using an autoencoder or an autoencoder based on a neural network to obtain the third information data of the user; A construction module, configured to construct a target recommendation model according to the third information data of the user; A recommendation list generation module, configured to generate a recommendation list for a cold start user according to the target recommendation model.

10. A computer electronic device, characterized in that, It includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the cold start user recommendation method based on heterogeneous transfer learning according to any one of claims 1-8 are implemented.

Images