Book intelligent recommendation method and system based on user portrait

Abstract

The invention relates to the technical field of intelligent recommendation, in particular to an intelligent book recommendation method and system based on a user portrait. The intelligent book recommendation system based on the user portrait comprises a life cycle classification module and a recommendation strategy setting module. According to the method, a multi-strategy recommendation architecture based on book life cycle state division is constructed, and matched recommendation strategies are set in a cold start period, a growth period, an outbreak period, a decline period and a settling period respectively in combination with user portrait exploration willingness, group behavior similarity, social influence characteristics and external hot event perception; dynamic switching and accurate putting of book recommendation strategies are realized, so that the recommendation system can maintain relatively high recommendation efficiency and content exposure value in multiple scenes of user interest transfer, book popularity evolution and platform content change, and then the life cycle management capability of books and the overall user satisfaction of the platform are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent recommendation, and particularly relates to a method and system for intelligent book recommendation based on user portraits. Background Art

[0002] In the existing book recommendation technologies, the mainstream methods mostly adopt means such as collaborative filtering, content matching, and deep learning. They mainly start from the user perspective and make recommendations based on their historical behaviors, interest preferences, and similar user portraits. These methods have achieved certain results in aspects such as cold start mitigation and interest capture, but generally have the problem of lacking perception of the life cycle characteristics of the books themselves. Current recommendation systems often regard all books as equivalent candidate items, ignoring the natural evolution process of books from the cold start period, growth period, explosion period to the decline period and precipitation period, and failing to dynamically adjust the recommendation strategy according to the life cycle stage of the books, resulting in problems such as insufficient exposure of cold books, insufficient spread of bestsellers, and inactivity of long-tail content, thereby affecting the balance of the content ecosystem and the continuous optimization of the user recommendation experience. Summary of the Invention

[0003] The present invention constructs a multi-strategy recommendation architecture based on the division of book life cycle states, combines the exploration willingness of user portraits, group behavior similarity, social influence characteristics, and external hot event perception, and sets matching recommendation strategies in the cold start period, growth period, explosion period, decline period, and precipitation period respectively, realizing the dynamic switching and precise placement of book recommendation strategies, enabling the recommendation system to maintain a high recommendation efficiency and content exposure value in multiple scenarios of user interest transfer, book popularity evolution, and platform content change, thereby significantly improving the life cycle management ability of books and the overall user satisfaction of the platform.

[0004] The present invention provides a method for intelligent book recommendation based on user portraits, including: Collecting the book behavior feature data corresponding to all books within a time window at the current monitoring time point, sending the book behavior feature data corresponding to the books into a book life cycle classification model for processing, and outputting the life cycle classification labels corresponding to the books. The life cycle classification labels include the cold start period, growth period, explosion period, decline period, and precipitation period. Based on the life cycle classification labels corresponding to the books, all books are divided into a cold start recommendation library, a growth recommendation library, an explosion recommendation library, a decline recommendation library, and a precipitation recommendation library; For all books in the cold start recommendation library, setting a book recommendation strategy based on the exploration willingness of the user; For all books in the growth recommendation library, setting a book recommendation strategy based on similar users; For all books in the explosion recommendation library, setting a book recommendation strategy based on user influence; Set the same recommendation strategy as the growth recommendation library for all books in the recession recommendation library; For all books in the precipitation recommendation library, construct a recommendation strategy based on hot information.

[0005] As a preferred aspect, for all books in the cold start recommendation library, set a book recommendation strategy based on the exploration willingness of users. Specifically: The book to be recommended is recorded as a cold start recommendation book, and the book feature vector corresponding to the cold start recommendation book is extracted; Traverse all users. For the selected user, obtain the exploration score of the selected user, where the exploration score represents the exploration willingness of the user. Obtain the book feature vector corresponding to the book of the selected user's last purchase behavior, denoted as the interest book feature vector. Calculate the similarity between the interest book feature vector and the book feature vector corresponding to the cold start recommendation book, denoted as the first user similarity score. Then perform an addition operation on the exploration score of the selected user and the first user similarity score to obtain the cold start recommendation score. Arrange all users in descending order of the cold start recommendation score to form a cold start recommendation set, and recommend the cold start recommendation book to the top N users in the cold start recommendation set; The exploration score of the user is calculated as follows: Obtain all the books purchased by the user in history, denoted as historical purchase books, and obtain the book feature vectors corresponding to all historical purchase books. Send the book feature vectors corresponding to the historical purchase books into the style label analysis network for processing to obtain the style labels corresponding to the historical purchase books. Add the reciprocal of the total number of evaluations corresponding to all historical purchase books, the variance corresponding to the book feature vectors of all historical purchase books, and the distribution entropy of all style labels to obtain the exploration score of the user.

[0006] As a preferred aspect, for all books in the growth recommendation library, set a book recommendation strategy based on similar users. Specifically: The book to be recommended is recorded as a growth recommendation book, obtain the book feature vector corresponding to the growth recommendation book, and form a set of users to be recommended who have purchased the growth recommendation book within the time window. For each user in the set of users to be recommended, construct a user portrait feature vector, and perform clustering analysis on all user portrait feature vectors to obtain several clustering clusters, and obtain the cluster center for each clustering cluster; Traverse all users except the set of books to be recommended. For the selected users, obtain the user portrait feature vectors corresponding to the selected users, calculate the similarity between the user portrait feature vectors corresponding to the selected users and the center of each clustering cluster respectively to obtain the simulated recommendation scores corresponding to the selected users, then take the maximum simulated recommendation score as the growth recommendation score corresponding to the selected users, arrange all users except the set of books to be recommended in descending order of the growth recommendation scores to form a growth recommendation set, and recommend the growth recommendation books to the top N users in the growth recommendation set; The specific method for constructing the user portrait feature vector is: based on combining the user's basic information and the book feature vector corresponding to the book for which the user had the last purchase behavior to form the user portrait feature vector.

[0007] As a preferred aspect, for all books in the burst recommendation library, set a book recommendation strategy based on user influence, specifically: Denote the book to be recommended as the burst recommendation book, traverse all users, obtain the interest book feature vectors corresponding to the selected users, calculate the similarity between the interest book feature vectors and the book feature vectors corresponding to the burst recommendation book, denote it as the second user similarity score, then perform an addition operation on the second user similarity score and the user's influence value to obtain the burst recommendation score, arrange all users in descending order of the burst recommendation scores to form a burst recommendation set, and recommend the burst recommendation book to the top N users in the burst recommendation set.

[0008] As a preferred aspect, for all books in the precipitation recommendation library, construct a recommendation strategy based on hot information, specifically: Continuously monitor the hot information and obtain the hot information feature vectors corresponding to the hot information; Denote the books in the precipitation recommendation library as precipitation recommendation books, traverse all precipitation recommendation books in the precipitation recommendation library, calculate the similarity between the hot information feature vectors and each precipitation recommendation book in the precipitation recommendation library respectively, denote it as the precipitation recommendation score corresponding to the precipitation recommendation book, and recommend the precipitation recommendation book corresponding to the maximum precipitation recommendation score to all users.

[0009] As a preferred aspect, the book life cycle classification model is established based on the random forest model, and the training of the book life cycle classification model is specifically carried out as follows: Obtain several book life cycle classification training samples, where the book life cycle classification training samples include the book behavior feature data corresponding to any book, label the book life cycle classification training samples with life cycle classification labels, form a book life cycle classification training set with all the labeled book life cycle classification training samples, and train the book life cycle classification model through the book life cycle classification training set.

[0010] As a preferred aspect, the style label analysis network is established based on a BP neural network, and the style label analysis network is trained, specifically including the following steps: Obtain a number of style label analysis training samples. The style label analysis training samples include the book feature vectors corresponding to the books. The style label analysis training samples are labeled with style labels, and all the labeled style label analysis training samples are combined into a style label analysis training set. The style label analysis network is trained through the style label analysis training set.

[0011] The present invention also provides a book intelligent recommendation system based on a user portrait, including: A life cycle classification module, which is used to collect the book behavior feature data corresponding to all books within a time window at the current monitoring time point, send the book behavior feature data corresponding to the books into a book life cycle classification model for processing, and output the life cycle classification labels corresponding to the books. The life cycle classification labels include a cold start period, a growth period, a burst period, a decline period, and a precipitation period. All books are classified into a cold start recommendation library, a growth recommendation library, a burst recommendation library, a decline recommendation library, and a precipitation recommendation library based on the life cycle classification labels corresponding to the books; A recommendation strategy setting module, which is used to set a book recommendation strategy based on the exploration willingness of the user for all books in the cold start recommendation library, set a book recommendation strategy based on similar users for all books in the growth recommendation library, set a book recommendation strategy based on the user influence for all books in the burst recommendation library, set a recommendation strategy consistent with the growth recommendation library for all books in the decline recommendation library, and construct a recommendation strategy based on hot information for all books in the precipitation recommendation library.

[0012] The present invention has the following advantages: The present invention constructs a multi-strategy recommendation architecture based on the division of the book life cycle state, combines the exploration willingness of the user portrait, group behavior similarity, social influence characteristics, and external hot event perception, and sets matching recommendation strategies in the cold start period, growth period, burst period, decline period, and precipitation period respectively, realizing the dynamic switching and precise placement of the book recommendation strategy, enabling the recommendation system to maintain a high recommendation efficiency and content exposure value in multiple scenarios of user interest transfer, book popularity evolution, and platform content change, thereby significantly improving the life cycle management ability of books and the overall user satisfaction of the platform. Description of the Drawings

[0013] Figure 1 It is a schematic structural diagram of the book intelligent recommendation system based on a user portrait adopted in the embodiment of the present invention. Detailed Embodiments

[0014] To enable those skilled in the art to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention.

[0015] Embodiment 1, a method for intelligent book recommendation based on user portraits, including: Collect the book behavior feature data corresponding to all books within the time window at the current monitoring time point. Here, the time window generally refers to the past 7 days. The book behavior feature data includes the total sales volume, the number of views, the number of evaluations, the sales growth rate, the view growth rate, and the evaluation number growth rate, etc. Among them, the sales growth amount is the growth rate of the total sales volume in the current time window compared with the total sales volume in the past 7 days, the view growth rate is the growth rate of the number of views in the current time window compared with the number of views in the past 7 days, the number of views is the total number of times users click on the book details page on the e-commerce platform, the evaluation number growth rate is the growth rate of the number of evaluations in the current time window compared with the number of evaluations in the past 7 days, and the number of evaluations is the number of comments users make on the book on the e-commerce platform. Send the book behavior feature data corresponding to the book into the book life cycle classification model for processing, and output the life cycle classification label corresponding to the book. The life cycle classification label includes the cold start period, the growth period, the outbreak period, the decline period, and the precipitation period. Among them, the cold start period is when the book is just put on the shelf or there are few comments for a long time, the book behavior data is relatively sparse, and the feedback is less. The growth period is the period when the book behavior feature data of the book is rising but not higher than the threshold set in advance. The outbreak period is the period when the book behavior feature data of the book is higher than the threshold set in advance. The decline period is the period when the book behavior feature data of the book is declining. The precipitation period is the period when the book behavior feature data of the book changes stably; divide all books into the cold start recommendation library, the growth recommendation library, the outbreak recommendation library, the decline recommendation library, and the precipitation recommendation library based on the life cycle classification label corresponding to the book; it should be noted that the time interval between adjacent monitoring time points is generally set to 1 day.

[0016] For all books in the cold start recommendation library, set the book recommendation strategy based on the user's exploration willingness. Specifically: Denote the book to be recommended as the cold start recommendation book, and extract the book feature vector corresponding to the cold start recommendation book. The specific method for extracting the book feature vector here is: use the TF-IDF method to extract keywords from the book title, book subtitle, book introduction, chapter preview, author introduction, and comments, etc., and then perform word embedding operations on the keywords to construct the first keyword vector. Then, perform a weighted summation average operation on all the first keyword vectors and their corresponding TF-IDF values to obtain the book feature vector; Traverse all users, here referring to users registered on the e-commerce platform. For the selected users, obtain the exploration scores of the selected users. The exploration score represents the exploration willingness of the users. Obtain the book feature vector corresponding to the book of the last purchase behavior of the selected users, denoted as the interest book feature vector. Calculate the similarity between the interest book feature vector and the book feature vector corresponding to the cold start recommended book, denoted as the first user similarity score. Then perform an addition operation on the exploration score of the selected users and the first user similarity score to obtain the cold start recommended score. Arrange all users in descending order according to the cold start recommended score to form a cold start recommendation set, and recommend the cold start recommended books to the top N users in the cold start recommendation set; whenever a user has a purchase behavior, the interest book feature vector will be reconstructed; It should be noted that the specific implementation method of recommending the cold start recommended books to the top N users in the cold start recommendation set is to set a recommendation area on the home page interface of the user, and the recommended books are displayed in the recommendation area.

[0017] The exploration score of the user is calculated in the following way: Obtain all the books purchased by the user in history, denoted as the historical purchase books, and obtain the book feature vectors corresponding to all the historical purchase books. Send the book feature vectors corresponding to the historical purchase books into the style label analysis network for processing to obtain the style labels corresponding to the historical purchase books. The style labels generally include science fiction, suspense, and history, etc. Add the reciprocal of the total number of evaluation numbers corresponding to all the historical purchase books, the variance of the book feature vectors corresponding to all the historical purchase books, and the distribution entropy of all the style labels to obtain the exploration score of the user. It should be noted that when calculating the variance of the book feature vectors corresponding to all the historical purchase books, first perform a sum and average operation on the book feature vectors corresponding to all the historical purchase books to obtain the average historical purchase book feature vector F. Calculate the variance D of the book feature vectors corresponding to all the historical purchase books through the following formula D = M -1 ∑ i [sim(F, H i )] 2 , where sim() is the similarity calculation, and H i is the book feature vector corresponding to the i-th historical purchase book, i = 1, 2, 3,..., M, and M is the total number of all the historical purchase books purchased by the user in history; when calculating the distribution entropy of all the style labels, it is realized through the following formula: , where P is the distribution entropy of all the style labels, and p j$f_j$ is the frequency of the $j$-th style tag, where $j = 1, 2, 3, \ldots, J$, and $J$ is the total number of style tags. Since there is little historical purchase data and few reviews during the cold start period, if these historical purchased books are recommended conventionally, users will ignore the recommended information due to lack of trust, resulting in poor recommendation effects. Therefore, we need to find those users who are willing to try new content, that is, users with a high exploration willingness. These users also have a relatively high probability of purchasing books with little historical purchase data and few reviews. Correspondingly, recommending books in the cold start period to these users with a high exploration willingness can achieve better recommendation effects. Once a purchase behavior occurs, it can also increase the historical purchase data and the number of evaluations.

[0018] For all books in the growth recommendation library, set the book recommendation strategy based on similar users. Specifically: Since there are already some purchase records and evaluation numbers for the books in the growth recommendation library, it is possible to find users similar to those recorded in the purchase records for recommendation, thereby improving the conversion rate and diffusion effect. Therefore, the book to be recommended is recorded as a growth recommendation book, and the corresponding book feature vector of the growth recommendation book is obtained. The users who have purchased the growth recommendation book within the time window are grouped into a set of users to be recommended. It should be noted that in the specific implementation process, the set of users to be recommended stores the corresponding user numbers. For each user in the set of users to be recommended, a user portrait feature vector is constructed, and clustering analysis is performed on all user portrait feature vectors. Specifically, the k-means algorithm can be used to obtain several clustering clusters. For each clustering cluster, the center of the clustering cluster is obtained. It should be noted that during the execution of the k-means algorithm, the similarity between two user portrait feature vectors is used as the distance between the two user portrait feature vectors, and the center of the clustering cluster is the average value of the entire clustering cluster. Traverse all users except the set of users to be recommended. For the selected users, obtain the corresponding user portrait feature vectors of the selected users, and calculate the similarity between the corresponding user portrait feature vectors of the selected users and the center of each clustering cluster respectively to obtain the simulated recommendation scores corresponding to the selected users. The similarity calculation generally uses the cosine similarity calculation method. Then, take the maximum simulated recommendation score as the growth recommendation score corresponding to the selected user. Arrange all users except the set of users to be recommended in descending order of the growth recommendation scores to form a growth recommendation set, and recommend the growth recommendation books to the top $N$ users in the growth recommendation set.

[0019] The specific method for constructing the user portrait feature vector is: based on combining the user's basic information and the book feature vector corresponding to the book for which the user had the last purchase behavior to form the user portrait feature vector. Here, the basic information includes age group, gender, region, etc.; whenever a user has a purchase behavior, the user portrait feature vector will be reconstructed.

[0020] For all books in the outbreak recommendation library, set the book recommendation strategy based on user influence. Specifically: For books in the outbreak recommendation library that already have sufficient purchase records and evaluation numbers and have sufficient influence, the recommendation strategy at this time should be to continue to expand the recommendation effect. Users with sufficient influence are good recommendation targets, and the influence of users can be used to further improve the recommendation effect. Therefore, the books to be recommended are recorded as outbreak recommendation books. Traverse all users, obtain the interest book feature vectors corresponding to the selected users, calculate the similarity between the interest book feature vectors and the book feature vectors corresponding to the outbreak recommendation books, which is recorded as the second user similarity score, and then perform an addition operation on the second user similarity score and the user's influence value to obtain the outbreak recommendation score. Here, the user's influence value is generally determined by the total number of all accounts associated with the user. Arrange all users in descending order of the outbreak recommendation score to form an outbreak recommendation set, and recommend the outbreak recommendation books to the top N users in the outbreak recommendation set; it should be noted that when actually performing recommendations for all books in the outbreak recommendation library, reward measures can be set for the users who perform the recommendations.

[0021] For all books in the decline recommendation library, set the same recommendation strategy as the growth recommendation library. Since the purchase data of books in the decline recommendation library shows a downward trend, it indicates that most users with a desire to purchase have already performed purchase behaviors. At this time, it is possible to recommend to users similar to those recorded in the return and purchase records.

[0022] For all books in the precipitation recommendation library, construct a recommendation strategy based on hot information. Specifically: Continuously monitor hot information. Here, the hot information can be hot search content, highly praised articles, or descriptions of popular TV dramas, etc. Obtain the hot information feature vectors corresponding to the hot information. The specific acquisition method is to extract keywords from the hot information using the TF-IDF method, then perform a word embedding operation on the keywords to construct a second keyword vector, and then perform a weighted sum and average operation on all the second keyword vectors and their corresponding TF-IDF values to obtain the hot information feature vectors; Record the books in the precipitation recommendation library as precipitation recommendation books. Traverse all the precipitation recommendation books in the precipitation recommendation library, calculate the similarity between the hot information feature vectors and each precipitation recommendation book in the precipitation recommendation library respectively, which is recorded as the precipitation recommendation score corresponding to the precipitation recommendation book, and recommend the precipitation recommendation book corresponding to the largest precipitation recommendation score to all users; it should be noted that the recommendation conversion rate of books in the precipitation recommendation library is relatively low. Therefore, recommending similar books through the latest hot information can significantly improve the recommendation conversion rate of books in the precipitation recommendation library.

[0023] The book life cycle classification model is established based on the random forest model. The training of the book life cycle classification model includes the following steps: Obtain several book life cycle classification training samples. The book life cycle classification training samples include book behavior feature data corresponding to any book. The book behavior feature data here is obtained based on actual monitoring. The book life cycle classification training samples are labeled with life cycle classification labels. The labeled life cycle classification labels here are manually implemented by operators. All labeled book life cycle classification training samples form a book life cycle classification training set. The book life cycle classification model is trained with the book life cycle classification training set. The training target is the labeled life cycle classification labels. It is judged whether the accuracy of the book life cycle classification model meets expectations. If the accuracy of the book life cycle classification model meets expectations, the trained book life cycle classification model is output. Otherwise, the book life cycle classification model continues to be trained with the book life cycle classification training set.

[0024] The style label analysis network is established based on the BP neural network. The training of the style label analysis network includes the following steps: Obtain several style label analysis training samples, which include book feature vectors corresponding to books. The book feature vectors here are obtained by operators based on actual books. Label the style label analysis training samples with style labels. The labeled style labels here are manually divided by operators based on actual reading. All labeled style label analysis training samples are combined into a style label analysis training set. The style label analysis network is trained with the style label analysis training set. The training target is the labeled style labels. It is determined whether the accuracy of the style label analysis network meets expectations. If the accuracy of the style label analysis network meets expectations, the trained style label analysis network is output. Otherwise, the style label analysis network continues to be trained with the style label analysis training set.

[0025] This application constructs a multi-strategy recommendation architecture based on the division of book life cycle states, combines user portrait exploration willingness, group behavior similarity, social influence characteristics and external hot event perception, and sets matching recommendation strategies in the cold start period, growth period, outbreak period, decline period and sedimentation period respectively, to achieve dynamic switching and precise delivery of book recommendation strategies, so that the recommendation system can maintain high recommendation efficiency and content exposure value in multiple scenarios such as user interest transfer, book popularity evolution and platform content changes, thereby significantly improving the book life cycle management capabilities and the overall user satisfaction of the platform.

[0026] Example 2, a book intelligent recommendation system based on user portrait, see Figure 1 ,include: A life cycle classification module is used to collect the book behavior characteristic data corresponding to all books within a time window at the current monitoring time point. Here, the time window generally refers to the past 7 days. The book behavior characteristic data includes the total sales volume, page view volume, number of evaluations, sales growth rate, page view growth rate, and evaluation number growth rate, etc. Among them, the sales volume increase is the growth rate of the total sales volume in the current time window compared to the total sales volume in the past 7 days. The page view growth rate is the growth rate of the page view volume in the current time window compared to the page view volume in the past 7 days. The page view volume is the total number of times users click on the book detail page on the e-commerce platform. The evaluation number growth rate is the growth rate of the number of evaluations in the current time window compared to the number of evaluations in the past 7 days. The number of evaluations is the number of comments users make on the book on the e-commerce platform. The book behavior characteristic data corresponding to the book is sent into the book life cycle classification model for processing, and the life cycle classification label corresponding to the book is output. The life cycle classification label includes the cold start period, growth period, explosion period, decline period, and precipitation period. Among them, the cold start period is when the book is just put on the shelf or there are few comments for a long time, and the book behavior data is relatively sparse and the feedback is less. The growth period is the period when the book behavior characteristic data of the book is rising but not higher than the pre-set threshold. The explosion period is the period when the book behavior characteristic data of the book is all higher than the pre-set threshold. The decline period is the period when the book behavior characteristic data of the book is declining. The precipitation period is the period when the book behavior characteristic data of the book changes stably; all books are divided into a cold start recommendation library, a growth recommendation library, an explosion recommendation library, a decline recommendation library, and a precipitation recommendation library based on the life cycle classification label corresponding to the book; it should be noted that the time interval between adjacent monitoring time points is generally set to 1 day; A recommendation strategy setting module is used to set book recommendation strategies for all books in the cold start recommendation library based on the exploration willingness of users, set book recommendation strategies for all books in the growth recommendation library based on similar users, set book recommendation strategies for all books in the explosion recommendation library based on user influence, set the same recommendation strategy as the growth recommendation library for all books in the decline recommendation library, and build a recommendation strategy based on hot information for all books in the precipitation recommendation library.

[0027] It should be understood that for those of ordinary skill in the art, improvements or transformations can be made according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention. The parts not described in detail in this specification belong to the prior art well-known to those of ordinary skill in the art.

Claims

1. A book intelligent recommendation method based on user portraits, characterized in that, Including: Collect the book behavior feature data corresponding to all books within the time window at the current monitoring time point, send the book behavior feature data corresponding to the books into the book life cycle classification model for processing, and output the life cycle classification labels corresponding to the books. The life cycle classification labels include the cold start period, the growth period, the outbreak period, the decline period, and the precipitation period. Divide all books into the cold start recommendation library, the growth recommendation library, the outbreak recommendation library, the decline recommendation library, and the precipitation recommendation library based on the life cycle classification labels corresponding to the books; For all books in the cold start recommendation library, set the book recommendation strategy based on the exploration willingness of the users; For all books in the growth recommendation library, set the book recommendation strategy based on similar users; For all books in the outbreak recommendation library, set the book recommendation strategy based on user influence; For all books in the decline recommendation library, set the same recommendation strategy as the growth recommendation library; For all books in the precipitation recommendation library, construct the recommendation strategy based on hot information.

2. The intelligent book recommendation method based on user portraits according to claim 1, wherein, For all books in the cold start recommendation library, set the book recommendation strategy based on the exploration willingness of the users. Specifically: Record the book to be recommended as the cold start recommendation book, and extract the book feature vector corresponding to the cold start recommendation book; Traverse all users. For the selected user, obtain the exploration score of the selected user. The exploration score represents the exploration willingness of the user. Obtain the book feature vector corresponding to the book that the selected user purchased last time, denoted as the interest book feature vector. Calculate the similarity between the interest book feature vector and the book feature vector corresponding to the cold start recommendation book, denoted as the first user similarity score. Then perform an addition operation on the exploration score of the selected user and the first user similarity score to obtain the cold start recommendation score. Arrange all users in descending order according to the cold start recommendation score to form a cold start recommendation set, and recommend the cold start recommendation book to the top N users in the cold start recommendation set; The exploration score of the user is calculated in the following way: Obtain all the books purchased by the user in history, denoted as the historical purchase books, and obtain the book feature vectors corresponding to all the historical purchase books. Send the book feature vectors corresponding to the historical purchase books into the style label analysis network for processing to obtain the style labels corresponding to the historical purchase books. Add the reciprocal of the total number of evaluations corresponding to all the historical purchase books, the variance corresponding to the book feature vectors of all the historical purchase books, and the distribution entropy of all the style labels to obtain the exploration score of the user.

3. The intelligent book recommendation method based on user portraits according to claim 2, wherein, For all books in the growth recommendation library, set the book recommendation strategy based on similar users. Specifically: Record the book to be recommended as the growth recommendation book, obtain the book feature vector corresponding to the growth recommendation book, and form a set of users to be recommended with the users who purchased the growth recommendation book within the time window. For each user in the set of users to be recommended, construct a user portrait feature vector, and perform clustering analysis on all the user portrait feature vectors to obtain several clustering clusters. Obtain the cluster center for each clustering cluster; Traverse all users except the set of users to be recommended. For the selected users, obtain the user portrait feature vectors corresponding to the selected users, calculate the similarity between the user portrait feature vectors corresponding to the selected users and the center of each clustering cluster respectively, obtain the simulated recommendation scores corresponding to the selected users, then take the maximum simulated recommendation score as the growth recommendation score corresponding to the selected users, arrange all users except the set of users to be recommended in descending order of the growth recommendation scores to form a growth recommendation set, and recommend the growth recommendation books to the top N users in the growth recommendation set; The specific method for constructing the user portrait feature vector is: based on the user's basic information and the book feature vector corresponding to the book purchased by the user the last time, form the user portrait feature vector.

4. The intelligent book recommendation method based on user portraits according to claim 3, characterized in that, For all books in the burst recommendation library, set a book recommendation strategy based on user influence, specifically: Record the book to be recommended as a burst recommendation book, traverse all users, obtain the interest book feature vector corresponding to the selected users, calculate the similarity between the interest book feature vector and the book feature vector corresponding to the burst recommendation book, record it as the second user similarity score, then perform an addition operation on the second user similarity score and the user's influence value to obtain the burst recommendation score, arrange all users in descending order of the burst recommendation scores to form a burst recommendation set, and recommend the burst recommendation book to the top N users in the burst recommendation set.

5. The intelligent book recommendation method based on user portraits according to claim 4, characterized in that For all books in the precipitation recommendation library, construct a recommendation strategy based on hot information, specifically: Continuously monitor hot information and obtain the hot information feature vector corresponding to the hot information; Record the books in the precipitation recommendation library as precipitation recommendation books, traverse all precipitation recommendation books in the precipitation recommendation library, calculate the similarity between the hot information feature vector and each precipitation recommendation book in the precipitation recommendation library respectively, record it as the precipitation recommendation score corresponding to the precipitation recommendation book, and recommend the precipitation recommendation book corresponding to the maximum precipitation recommendation score to all users.

6. The intelligent book recommendation method based on user portraits according to claim 5, characterized in that The book life cycle classification model is established based on the random forest model. Training the book life cycle classification model specifically includes the following steps: Obtain several book life cycle classification training samples. The book life cycle classification training samples include the book behavior feature data corresponding to any book. Label the book life cycle classification training samples with the life cycle classification label, form the book life cycle classification training set with all the labeled book life cycle classification training samples, and train the book life cycle classification model with the book life cycle classification training set.

7. The intelligent book recommendation method based on user portraits according to claim 6, characterized in that The style label analysis network is established based on the BP neural network. Training the style label analysis network specifically includes the following steps: Obtain several style label analysis training samples. The style label analysis training samples include the book feature vector corresponding to the book. Label the style label analysis training samples with the style label, form the style label analysis training set with all the labeled style label analysis training samples, and train the style label analysis network with the style label analysis training set.

8. A book intelligent recommendation system based on user portraits, characterized in that, The system applies the intelligent book recommendation method based on user portraits described in any one of claims 1-7 above, including: A life cycle classification module, which is used to collect the book behavior feature data corresponding to all books within a time window at the current monitoring time point, send the book behavior feature data corresponding to the books into a book life cycle classification model for processing, and output the life cycle classification labels corresponding to the books. The life cycle classification labels include the cold start period, the growth period, the explosion period, the decline period, and the precipitation period. All books are divided into a cold start recommendation library, a growth recommendation library, an explosion recommendation library, a decline recommendation library, and a precipitation recommendation library based on the life cycle classification labels corresponding to the books; A recommendation strategy setting module, which is used to set book recommendation strategies based on the exploration willingness of users for all books in the cold start recommendation library, set book recommendation strategies based on similar users for all books in the growth recommendation library, set book recommendation strategies based on user influence for all books in the explosion recommendation library, set the same recommendation strategies as those in the growth recommendation library for all books in the decline recommendation library, and construct recommendation strategies based on hot information for all books in the precipitation recommendation library.

Images