A personalized recommendation method and system for library books

Abstract

The present application relates to the technical field of book recommendation and discloses a personalized recommendation method and system for library books, comprising: obtaining book RFID tag flow data and user behavior data during a user's book borrowing process; dividing the library into grids according to a preset grid size, analyzing the user's interest level in books in each grid area, and constructing a book interest heat map; extracting hot spots with interest level values ​​greater than a preset interest level threshold based on the book interest heat map, screening recommended books in the hot spots using a preset book recommendation model, and obtaining a first recommended book set; optimizing the recommended books using a preset recommendation optimization model based on the user's borrowing status of the first recommended book set, and obtaining a second recommended book set; the present application can adapt to dynamic changes in user interests in the book recommendation process, improve the accuracy and timeliness of recommended books, and achieve personalized recommendations for users.

Description

Technical Field

[0001] The present invention relates to the technical field of book recommendation, and in particular to a personalized recommendation method and system for library books. Background Art

[0002] With the rapid development of information technology, libraries, as important places for knowledge dissemination, need to combine users' interests when providing personalized book recommendations to users. Users' interests are time-sensitive. Currently, when recommending books to users, they are mainly based on book categories, which has poor recommendation effects and cannot achieve personalized recommendations for each user.

[0003] The existing technology has the following problems: analyzing user interests based on user borrowing data or book content attributes, the data is single and cannot comprehensively and accurately analyze the user's interest preferences; only analyzing and recommending based on book categories or content ignores the logical placement of books in the library, and directly analyzes each book for recommendation, which makes the analysis process complicated and the recommendation efficiency low; after recommending books, there is a lack of analysis and prediction of real-time changes in user interests, and it is impossible to recommend books based on the user's interest change trends. To solve at least one of the above problems, the present invention proposes a personalized recommendation method and system for library books. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the main purpose of the present invention is to provide a personalized recommendation method and system for library books, which can effectively solve the problems in the background technology. The specific technical solutions of the present invention are as follows:

[0005] A personalized recommendation method for library books, comprising:

[0006] Obtain book RFID tag circulation data and user behavior data during the user's book borrowing process;

[0007] Combining the book RFID tag flow data and user behavior data, the library is divided into grids according to a preset grid size, and the user interest in the books in each grid area is analyzed to construct a book interest heat map;

[0008] Extracting hotspot areas with interest levels greater than a preset interest level threshold based on the book interest heat map, and filtering recommended books using a preset book recommendation model based on the hotspot areas to obtain a first recommended book set;

[0009] Based on the borrowing status of the first recommended book set by the user, the recommended books are optimized through a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books.

[0010] Specifically, the library is divided into grids according to a preset grid size by combining the book RFID tag flow data and user behavior data. The user's interest in the books in each grid area is analyzed to construct a book interest heat map, including:

[0011] Divide the library into grids according to a preset grid size to obtain multiple grid areas;

[0012] Combining book RFID tag flow data and user behavior data, the user's interest in books in each grid area is analyzed through a preset interest analysis model to obtain the interest level value of each grid area;

[0013] Taking the grid area where the book borrowed by the user is located as the center, regional diffusion is performed to the surrounding grid areas in combination with the interest level value to construct a book interest heat map.

[0014] Specifically, the method combines the book RFID tag flow data and the user behavior data, analyzes the user's interest in the books in each grid area through a preset interest analysis model, and obtains the interest level value of each grid area, including:

[0015] Based on the book RFID tag flow data, the first grid area where the user's borrowed books are located is extracted, and the time difference between book borrowing and returning is analyzed to obtain the book borrowing duration;

[0016] Based on user behavior data, the user's book browsing behavior is analyzed through a preset interest analysis model to obtain user behavior analysis results;

[0017] Determining the interest level value of the first grid area based on the book borrowing duration and the user behavior analysis result;

[0018] The interest level value of each grid area is calculated by combining the content similarity between the books in the grid areas with the interest level value of the first grid area.

[0019] Specifically, the book interest heat map is constructed by taking the grid area where the book borrowed by the user is located as the center and performing regional diffusion to the surrounding grid areas in combination with the interest level value, including:

[0020] The first grid area where the books borrowed by the user are located is used as the central grid area;

[0021] Calculating the thermal value of each grid area based on the interest level value of each grid area and the distance between the grid area and the central grid area, wherein when there are multiple central grid areas, the thermal values ​​calculated for each central grid area are superimposed;

[0022] Taking the central grid area as the center, regional diffusion is performed to the surrounding grid areas according to the thermal value to construct a book interest heat map.

[0023] Specifically, based on the book interest heat map, hot spots with interest levels greater than a preset interest level threshold are extracted, and recommended books are screened out using a preset book recommendation model based on the hot spots to obtain a first recommended book set, including:

[0024] According to the book interest heat map, extract the second grid areas where the interest level is greater than a preset interest level threshold, and merge adjacent second grid areas to obtain multiple hot spots;

[0025] In each hotspot area, the user's interest in different books is analyzed through a preset book recommendation model to filter recommended books and obtain a first area recommended book set;

[0026] By analyzing the book correlation between the hotspot area and the non-hotspot area, recommended books are screened in the non-hotspot area to obtain a set of recommended books for the second area;

[0027] The first region recommended book set and the second region recommended book set are combined to obtain a first recommended book set.

[0028] Specifically, in each hotspot area, the user's interest in different books is analyzed by a preset book recommendation model to filter recommended books, and a first area recommended book set is obtained, including:

[0029] In each hotspot area, the attributes of each book are analyzed to obtain the book attribute characteristics;

[0030] Based on the user behavior data and the book attribute characteristics, the user's interest matching degree for each book is calculated through a preset book recommendation model;

[0031] According to the interest matching degree from high to low, a preset first number of recommended books are screened out to obtain a first region recommended book set.

[0032] Specifically, the method analyzes the book association between the hotspot area and the non-hotspot area, filters the recommended books in the non-hotspot area, and obtains the second area recommended book set, including:

[0033] The book similarity is obtained by calculating the similarity between the book attribute features of the books in the non-hotspot area and the book attribute features of the books in the hotspot area;

[0034] The correlation between non-hotspot areas and hotspot areas is calculated by combining the distance between them and the corresponding book similarities.

[0035] Screening out a preset second number of non-hotspot areas according to the correlation degree from high to low to obtain a set of candidate areas;

[0036] In each non-hotspot area in the candidate area set, a preset third number of recommended books are screened out according to the user's interest matching degree for books in the non-hotspot area from high to low to obtain a second area recommended book set.

[0037] Specifically, based on the borrowing status of the first recommended book set by the user, the recommended books are optimized by a preset recommendation optimization model to obtain the second recommended book set, so as to make personalized recommendations for library books, including:

[0038] According to the user's borrowing of books in the first recommended book set, the user's interest changes are analyzed to obtain an interest vector;

[0039] In combination with the interest vector, the recommended books are optimized through a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books.

[0040] Specifically, combining the interest vector, optimizing the recommended books through a preset recommendation optimization model to obtain a second set of recommended books, so as to make personalized recommendations for library books, including:

[0041] Based on the interest vector and user behavior data, the user's interest migration is predicted through a preset recommendation optimization model to obtain an interest migration prediction vector;

[0042] Based on the interest migration prediction vector, screening out books that match the interest migration prediction vector as an optimized book set;

[0043] The recommended books are optimized by optimizing the book set to obtain a second recommended book set, so as to make personalized recommendations for library books.

[0044] A personalized recommendation system for library books, used to implement the personalized recommendation method for library books, comprising:

[0045] The data acquisition module obtains the book RFID tag flow data and user behavior data during the user's book borrowing process;

[0046] The book interest heat map construction module combines the book RFID tag flow data and user behavior data, divides the library into grids according to a preset grid size, analyzes the user's interest in the books in each grid area, and constructs a book interest heat map;

[0047] a book recommendation module, which extracts hotspot areas having interest levels greater than a preset interest level threshold based on the book interest heat map, and filters out recommended books using a preset book recommendation model based on the hotspot areas to obtain a first set of recommended books;

[0048] The recommended book optimization module optimizes the recommended books based on the user's borrowing status of the first recommended book set through a preset recommendation optimization model to obtain a second recommended book set to make personalized recommendations for library books.

[0049] Compared with the prior art, this application has the following beneficial effects:

[0050] This application combines book RFID tag flow data and user behavior data, divides the library into grids according to a preset grid size, analyzes the interest levels of users in different areas, constructs a book interest heat map, and recommends and optimizes corresponding books based on the book interest heat map and the changes in user interests; combined with multi-dimensional data, it can accurately analyze the user's activity trajectory and interest preferences in the library, combine user interests with the library's spatial structure, and consider the spatial propagation characteristics of user interests and the correlation between the placement of library books. This can improve the efficiency of book recommendations, expand the scope of recommendations, and improve the comprehensiveness of recommended books. By optimizing recommended books, it can adapt to the dynamic changes of user interests and improve the accuracy and timeliness of recommended books. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a workflow diagram of a personalized recommendation method for library books in Example 1 of the present invention;

[0052] Figure 2 This is a schematic diagram of a book interest heat map in Example 1 of the present invention;

[0053] Figure 3 This is a schematic diagram of the hotspot area screening of the book interest heat map in Example 1 of the present invention;

[0054] Figure 4 This is a structural diagram of a personalized recommendation system for library books in Example 2 of the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0058] Example 1

[0059] This embodiment provides a personalized recommendation method for library books, such as Figure 1 As shown, a personalized recommendation method for library books includes:

[0060] S101, obtaining book RFID tag circulation data and user behavior data during the user's book borrowing process;

[0061] S102: combining book RFID tag flow data and user behavior data, dividing the library into grids according to a preset grid size, analyzing user interest in books within each grid area, and constructing a book interest heat map;

[0062] S103, extracting hot spots with interest levels greater than a preset interest level threshold based on the book interest heat map, and filtering recommended books using a preset book recommendation model based on the hot spots to obtain a first recommended book set;

[0063] S104: Based on the borrowing status of the first recommended book set by the user, the recommended books are optimized by a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books.

[0064] With the continuous increase in the number of library books and the improvement of users' demand for personalized services, the existing book recommendation methods are difficult to meet users' needs and changes in interests. This technical solution combines the book RFID tag flow data and user behavior data during the user's book borrowing process, divides the library into grids according to a preset grid size, calculates the interest level value of each grid area based on the acquired data, and constructs a book interest heat map. Based on the book interest heat map, hot spots with interest level values ​​greater than a preset threshold are extracted. Recommended books are screened from hot spots and non-hot spots respectively through a preset book recommendation model to form a first recommended book set. Based on the user's borrowing of the first recommended book set, the user's interest changes are analyzed, and the recommended books are optimized through a preset recommendation optimization model to obtain a second recommended book set to make personalized recommendations for library books.

[0065] This embodiment integrates multi-dimensional data to perform user interest analysis, which can obtain accurate user interest analysis results, recommend books that match user interests, and improve the accuracy and relevance of recommendations; combined with the library space area division results and the book interest heat map, it can quickly screen out books that users are interested in, and analyze changes in user interests based on the user's borrowing of recommended books, and optimize the recommended books in real time, which can improve the matching degree between recommended books and user interests, realize real-time personalized recommendation of books, attract users to borrow more books, and improve the circulation of library books.

[0066] In this embodiment, first, the book RFID tag flow data and user behavior data during the user's book borrowing process are obtained. The book RFID tag flow data records the location changes and borrowing status of the book in the library, and the user behavior data records the user's activities in the library, including the user's stay time in various areas, search records and other data; by obtaining multi-dimensional data, it is possible to fully understand the interaction between users and books, including the book borrowing flow, user activity trajectory and behavioral preferences in the library, providing data support for analyzing books that users are interested in.

[0067] For example, RFID readers are installed at library entrances, exits, and shelves. When a book with an RFID tag passes by the reader, the reader reads the tag information, including the book ID and the time it was moved. This information is then transmitted in real time to the library management system's database. For example, when a user borrows a book, the RFID reader at the entrance records the borrowing time and book number. When a book is returned to the shelf, the RFID reader on the shelf records the return time and shelf location.

[0068] For example, sensors are installed in the library, including infrared sensors, Wi-Fi probes and other sensors. The infrared sensor can detect the time a user stays in a certain area. When a user enters or leaves the area where the infrared sensor is installed, the sensor will record the corresponding time point and calculate the length of time the user stays in the area; the Wi-Fi probe can detect the MAC address of the user's device and analyze the movement trajectory of the user's device in the library. At the same time, the library's management system can also record behavioral data such as the keywords entered by users when searching for books and the book pages browsed.

[0069] Specifically, the library space is divided into grids according to a preset grid size to obtain multiple grid areas. The book RFID tag flow data and user behavior data are combined to analyze the user's interest in the books in each grid area, and a book interest heat map is constructed according to the interest level. By analyzing indicators such as the borrowing frequency of books in each grid area and the user's residence time, the user's interest level in the books in the grid area is calculated, and a book interest heat map is constructed according to different interest level values. Areas with high interest levels are displayed as darker areas on the heat map, and areas with low interest levels are lighter in color. By constructing a book interest heat map, the distribution of users' interests in different areas of the library can be intuitively displayed, and when recommending books to users, the book areas that users are interested in can be quickly searched.

[0070] Specifically, according to the interest level value of each grid area, grid areas with interest level values ​​greater than a preset interest level threshold are extracted from the book interest heat map as hot spots. Recommended books are screened out through a preset book recommendation model in combination with the hot spots to obtain a first recommended book set. First, books related to books borrowed by users are analyzed in the hot spots, and recommended books are screened out from the hot spots. By analyzing the distance and book association between the hot spots and non-hot spots, recommended books are screened out from the non-hot spots. The recommended books in the hot spots and non-hot spots are combined to form a first recommended book set. By extracting hot spots and screening out recommended books in the hot spots, the user's high-interest areas can be quickly located. Extracting recommended books in the high-interest areas can improve the efficiency of the book recommendation process. Screening out recommended books in non-hot spots can improve the comprehensiveness of recommended books, improve the categories of recommended books, and improve the matching degree between recommended books and user interests.

[0071] Preferably, based on the user's borrowing of books in the first recommended book set, the recommended books are optimized through a preset recommendation optimization model to obtain a second recommended book set; by analyzing the user's borrowing of recommended books and analyzing the user's interest changes, the books in the first recommended book set are adjusted and optimized according to the user's interest changes, removing books with low user interest, and adding books that meet the user's real-time interests in combination with the user's interest changes to obtain a second recommended book set, thereby achieving personalized book recommendations; by analyzing the user's borrowing feedback on recommended books, the recommended books are optimized, and the recommendation strategy can be continuously improved in combination with the user's real-time interests, and the recommended books can be adjusted in time to improve the accuracy and effectiveness of the recommendations, thereby providing users with personalized recommendations that better meet current needs.

[0072] This application combines book RFID tag flow data and user behavior data, divides the library into grids according to a preset grid size, analyzes the interest levels of users in different areas, constructs a book interest heat map, and recommends and optimizes corresponding books based on the book interest heat map and the changes in user interests; combined with multi-dimensional data, it can accurately analyze the user's activity trajectory and interest preferences in the library, combine user interests with the library's spatial structure, and consider the spatial propagation characteristics of user interests and the correlation between the placement of library books. This can improve the efficiency of book recommendations, expand the scope of recommendations, and improve the comprehensiveness of recommended books. By optimizing recommended books, it can adapt to the dynamic changes of user interests and improve the accuracy and timeliness of recommended books.

[0073] Furthermore, by combining book RFID tag flow data and user behavior data, the library is divided into grids according to a preset grid size, and the user interest in the books in each grid area is analyzed to construct a book interest heat map, including:

[0074] S201, dividing the library into grids according to a preset grid size to obtain multiple grid areas;

[0075] S202: Analyze the user's interest in the books in each grid area using a preset interest analysis model by combining the book RFID tag flow data and the user behavior data to obtain an interest level value for each grid area;

[0076] S203 , taking the grid area where the book borrowed by the user is located as the center, regional diffusion is performed to the surrounding grid areas in combination with the interest level value to construct a book interest heat map.

[0077] In this embodiment, first, the library is grid-divided according to a preset grid size to obtain multiple grid areas; based on the actual area of ​​the library, the layout of the bookshelves, and the convenience of data processing, the side length of the grid is determined, and the library space is divided into multiple grid areas according to the corresponding grid size. Through grid division, the books in each grid area can be analyzed, thereby improving the efficiency of book category analysis and book recommendation.

[0078] Specifically, based on the book RFID tag circulation data and user behavior data, the preset interest analysis model is used to analyze the degree of user interest in the books in each grid area, and the interest level value of each grid area is obtained; the collected book RFID tag circulation data and user behavior data are sorted according to the grid area, and the total number of book borrowings, average borrowing time, total stay time of users in the area, average number of books browsed, and other data in each grid area are counted respectively; the interest level value of each grid area is calculated by combining the content similarity between the grid area where the user borrows the book and other grid areas; by calculating the interest level value, the user's interest in the books in each grid area can be quickly analyzed, the reliability and accuracy of the analysis results can be improved, and the calculated interest level value provides data support for book recommendations.

[0079] Preferably, the grid area where the books borrowed by the user are located is taken as the central grid area, and the interest level value of each grid area is diffused to the surrounding areas to construct a book interest heat map; the grid area where the books borrowed by the user are located is taken as the center, and the thermal value of each grid area is calculated in combination with the interest level value of each grid area and the distance between the grid area and the central grid area. When there is an overlapping area in the diffusion area of ​​the central grid area, the thermal values ​​of the overlapping areas are superimposed to obtain the thermal value of each grid area. With the central grid area as the center, regional diffusion is performed to the surrounding grid areas according to the thermal value to obtain a book interest heat map; a thermal value range is set, and thermal values ​​in different ranges are mapped to different colors. For example, thermal values ​​between 0-10 are set to blue, between 10-20 are set to green, between 20-30 are set to yellow, and above 30 are set to red to generate a book interest heat map; constructing a book interest heat map through regional diffusion can comprehensively display the distribution of user interests in the library space, and when recommending books to users, the user's area of ​​interest can be quickly located.

[0080] Furthermore, by combining the book RFID tag flow data and user behavior data, the user's interest in the books in each grid area is analyzed through a preset interest analysis model to obtain the interest level value of each grid area, including:

[0081] S301: Extract the first grid area where the book borrowed by the user is located based on the book RFID tag flow data, analyze the time difference between the borrowing and returning of the book, and obtain the book borrowing duration;

[0082] S302: Analyze the user's book browsing behavior using a preset interest analysis model based on the user behavior data to obtain a user behavior analysis result;

[0083] S303: Determine the interest level value of the first grid area based on the book borrowing time and the user behavior analysis result;

[0084] S304: Calculate the interest level value of each grid area by combining the content similarity between the books in the grid areas with the interest level value of the first grid area.

[0085] In this embodiment, based on the book RFID tag flow data, the grid area where the book borrowed by the user is located is taken as the first grid area. The time difference between the operation time recorded by the RFID reading and writing device when the book is borrowed and returned is calculated to obtain the book borrowing time. By calculating the borrowing time, the user's investment in the book can be reflected. The longer the borrowing time, the higher the user's interest in the book, which provides a data basis for analyzing the user's interest in the books in the first grid area.

[0086] Specifically, the user's book browsing situation is analyzed according to the user behavior data, and the user's book search keywords, browsing history, etc. are analyzed through the preset interest analysis model to obtain the user behavior analysis results, which can be used to determine the type of books that the user is interested in; according to the library's business needs and understanding of user behavior, corresponding user interest judgment rules are set in the preset behavior analysis model. For example, if the user searches for books on the same topic more than 3 times, it means that the user is interested in books on that topic; the sorted user behavior data is input into the behavior analysis model, and analyzed according to the preset interest judgment rules to generate corresponding user behavior analysis results for each user; by analyzing the user behavior data, the user's interested books can be analyzed from the user's operating behavior. Compared with analysis based only on book borrowing data, this solution can more comprehensively understand the user's activities and interest needs in the library, thereby improving the accuracy and pertinence of book recommendations.

[0087] Specifically, the book borrowing time and the user behavior analysis results are combined to calculate the interest level value of the first grid area; according to the book interest analysis requirements, corresponding weights are set for the book borrowing time and the user behavior analysis results, and the book borrowing time and the user behavior analysis results are weighted and summed to calculate the interest level value of the first grid area; the book borrowing time reflects the user's interest in the book from a time perspective, and the user behavior analysis results reflect the user's interest from a behavioral perspective. Combining the two can more comprehensively and accurately analyze the user's interest level in the books in the first grid area.

[0088] Preferably, after calculating the interest level value of the first grid area, the interest levels of books in other grid areas can be analyzed using the interest level value of the first grid area by calculating the content similarity between books in the grid areas. Text analysis techniques, including TF-IDF algorithm, cosine similarity algorithm, etc., are used to analyze the content of books between different grid areas and calculate the content similarity between books in different grid areas. Specifically, text information such as book titles, abstracts, and keywords is vectorized, and then the cosine value of the angle between the vectors of books in the grid areas is calculated using the cosine similarity algorithm. The closer the value is to 1, the higher the similarity of the book content. At the same time, a weight for content similarity in the propagation of interest level values ​​is set according to the distance between the grid areas of the library. For each grid area, the content similarity and the set propagation weight are weighted and summed to calculate the interest level value of each grid area. The interest level value of each grid area is calculated by analyzing the content similarity between books in the grid areas, which comprehensively considers the relevance of book content and the distance between grid areas, avoids isolated analysis of each grid area, can comprehensively evaluate the distribution of users' interest in books in each grid area of ​​the library, and provide an accurate interest level reference for book recommendations.

[0089] Furthermore, taking the grid area where the user's borrowed book is located as the center, the interest level value is combined with regional diffusion to the surrounding grid areas to construct a book interest heat map, including:

[0090] S401: The first grid area where the book borrowed by the user is located is used as the central grid area;

[0091] S402: Calculate the heat value of each grid area based on the interest level value of each grid area and the distance between the grid area and the central grid area. When there are multiple central grid areas, the heat values ​​calculated for each central grid area are superimposed.

[0092] S403: Taking the central grid area as the center, regional diffusion is performed to the surrounding grid areas according to the heat value to construct a book interest heat map.

[0093] In this embodiment, the first grid area where the books borrowed by the user are located is used as the central grid area. The first grid area concentrates on the user's current reading interest. Setting the first grid area as the central grid area can spread to the surrounding areas with the books of interest as the core. When the books in the library are placed, the books in adjacent grid areas have certain relevance in terms of subject and category. Spreading to the surrounding areas with the first grid area as the core can make the analysis of the user's books of interest more targeted, avoid blindly analyzing the entire library grid area, and be able to associate areas related to the user's current interests, thereby improving the efficiency of user interest analysis and providing a basis for constructing a heat map.

[0094] Specifically, the heat value of each grid area is calculated based on the interest level value of each grid area and the distance between each grid area and the central grid area. When a grid area is affected by multiple central grid areas, the heat value calculated for each central grid area is superimposed to calculate the heat value of the corresponding grid area. The interest level value reflects the user's interest in the books in the grid area, and the distance between the grid area and the central grid area reflects the degree of spatial association between different areas. The closer the distance between the grid area and the central grid area, the higher the user's interest in the books in the area, and the more likely the user is to browse the books in the area during the book borrowing process. The more times a book is browsed, the thermal value of each grid area is calculated by comprehensively considering the interest level value and the distance from the central grid area, taking the inverse of the distance and performing weighted summation with the interest level value. When there are multiple central grid areas, the thermal value calculated for each central grid area is added together to obtain the thermal value of each grid area. By calculating the thermal value based on the comprehensive interest level value and distance, the user's interest intensity and spatial correlation factors are taken into account. The calculated thermal value can more accurately reflect the distribution of user interests in the library space, improve the comprehensiveness and accuracy of the thermal value calculation, and provide a data basis for constructing a book interest heat map.

[0095] Preferably, the thermal value calculated based on each grid area is diffused to the surrounding grid areas with the central grid area as the center, and a corresponding relationship between the thermal value and the color is established; for example, the thermal value is set to blue in the range of 0-20, green in the range of 21-40, yellow in the range of 41-60, orange in the range of 61-80, and red in the range of 81-100, and a book interest heat map is constructed based on the thermal value and the corresponding color; by constructing a book interest heat map, the distribution of user interests can be intuitively displayed, and based on the book interest heat map, the book area of ​​interest can be quickly screened out, thereby improving the efficiency of book recommendation.

[0096] like Figure 2As shown in the figure, the thickness of the area box represents the depth of the heat map color, where the thicker the area box, the darker the color in the corresponding area. In the figure, with the central grid area as the center, red area, orange area and yellow area are drawn as an example to represent the regional diffusion of the book interest heat map.

[0097] Furthermore, based on the book interest heat map, hot spots with interest levels greater than a preset interest level threshold are extracted, and recommended books are screened out using a preset book recommendation model based on the hot spots to obtain a first recommended book set, including:

[0098] S501: extracting second grid areas whose interest levels are greater than a preset interest level threshold according to the book interest heat map, and merging adjacent second grid areas to obtain multiple hotspot areas;

[0099] S502: In each hotspot area, analyzing the user's interest in different books by using a preset book recommendation model to filter recommended books and obtain a first area recommended book set;

[0100] S503, by analyzing the book relevance between the hotspot area and the non-hotspot area, screening recommended books in the non-hotspot area to obtain a second area recommended book set;

[0101] S504: Combine the first region recommended book set and the second region recommended book set to obtain a first recommended book set.

[0102] like Figure 3 As shown, this embodiment extracts the second grid area with an interest level value greater than a preset interest level threshold based on the book interest heat map, and obtains multiple hot spots. The area in the dotted box is screened out as the hot spot area in the figure. In each hot spot area, the preset book recommendation model is used to analyze the user's interest in different books, and the books that meet the conditions are screened out to obtain a first area recommended book set. The correlation between the books between the hot spot area and the non-hot spot area is analyzed. Combined with the user's interest books, books with potential interest are screened out in the non-hot spot area to obtain a second area recommended book set. The first area recommended book set and the second area recommended book set are deduplicated and combined to obtain a first recommended book set. By combining the hot spot area and the non-hot spot area to screen the recommended books, the user's book interest preference can be accurately grasped, and books that are highly matched with the user's interest can be recommended, reducing the recommendation of irrelevant books and improving the accuracy of the recommendation. At the same time, by analyzing the correlation between books between regions, potential interest books in the non-hot spot area are screened, enriching the types and range of recommended books, providing users with diverse reading choices, and stimulating users' reading interest.

[0103] In this embodiment, first, in the book interest heat map, the second grid area whose interest level value is greater than the preset interest level threshold is screened out, and the adjacent second grid areas are merged to obtain the hot spot area. The higher the interest level value, the higher the user's interest in the books in the grid area. By screening out the areas with high user interest, the areas where the user's interest is concentrated can be quickly located, making the book recommendation process more targeted and improving the efficiency and accuracy of book recommendations.

[0104] Specifically, in each hotspot area, the preset book recommendation model is used to analyze the user's interest in the books in the hotspot area. The model comprehensively considers interest factors such as book borrowing history, user's historical borrowing behavior, book ratings, content tags, etc., analyzes the degree of match between each book and user's interests, and screens out the books with the highest user interest to form the first area recommended book set; by screening books in the hotspot area, it is possible to fully analyze the user's interest needs in the interest-concentrated area and improve the match between the recommended books and the user's interests. Compared with random recommendations or indiscriminate recommendations, this solution can more accurately meet the user's reading interest needs and improve the user's acceptance and borrowing rate of recommended books.

[0105] Preferably, the books in the non-hotspot area and the hotspot area are related in content. By analyzing the book correlation between the hotspot area and the non-hotspot area, recommended books are screened out in the non-hotspot area to obtain a second area recommended book set; by screening recommended books in the non-hotspot area, taking into account the various correlations between books in the library, such as content correlation, subject correlation, author correlation, etc., there will also be correlations between books in areas with different popularity. By analyzing the correlation between books in the hotspot area and the non-hotspot area, potential books related to user interests in the non-hotspot area can be screened out; for example, a certain type of book is popular in the hotspot area, and books with the same author or subject in the non-hotspot area will also meet the user's interests. Screening these books to form the second area recommended book set can enrich the recommended content, expand the user's reading selection range, and increase the probability of recommended books being borrowed by users.

[0106] Specifically, the first area recommended book set and the second area recommended book set are combined to obtain a first recommended book set, the first area recommended book set includes books in hot spot areas that are highly matched with user interests, and the second area recommended book set includes books in non-hot spot areas that users may be interested in. Combining the two can integrate book resources in areas where users' interests are concentrated and potential related areas to obtain a richer and more comprehensive first recommended book set. While ensuring the accuracy of recommended books, the breadth of recommendations is expanded to provide users with diverse reading choices; the books in the first area recommended book set and the second area recommended book set are deduplicated and then merged, and the recommended books are sorted from high to low according to the interest level values ​​of the grid areas where the recommended books are located to obtain the first recommended book set. By combining the hot spot area recommended books and the non-hot spot area recommended books, it can not only meet the user's current interest needs, but also guide the user to discover potential reading interests, enhance the user's reading experience in the library, and realize personalized book recommendations for users.

[0107] Furthermore, in each hotspot area, the user's interest in different books is analyzed by a preset book recommendation model to filter recommended books, thereby obtaining a first area recommended book set including:

[0108] S601: Analyze the attributes of each book in each hotspot area to obtain book attribute characteristics;

[0109] S602: Calculate the user's interest matching degree for each book using a preset book recommendation model based on the user behavior data and the book attribute characteristics;

[0110] S603: Filter out a preset first number of recommended books according to interest matching from high to low, and obtain a first region recommended book set.

[0111] In this embodiment, in each hotspot area, the attributes of each book are analyzed to obtain book attribute characteristics. The book attributes include book category, author, publication year, content summary, keywords, reader rating, number of borrowing times and other information. The book attribute information of each book is quantified according to a preset quantification standard to obtain the book attribute characteristics of each book. By analyzing the book attributes to obtain the book attribute characteristics and quantitatively analyzing the book information, the book characteristics can be analyzed from multiple dimensions, thereby improving the accuracy and relevance of book recommendations.

[0112] Specifically, based on user behavior data and book attribute characteristics, the user's interest matching degree for each book is calculated through a preset book recommendation model, and the user behavior data is sorted and analyzed. The book recommendation model includes but is not limited to a content-based recommendation algorithm. According to the user's interest preferences, corresponding weights are set for the user behavior data and book attribute characteristics. For example, the borrowing history weight is set to 0.4, the search keyword weight is set to 0.3, the book rating weight is set to 0.2, and the borrowing times weight is set to 0.1; the sorted user behavior data and book attribute characteristics are input into the recommendation model, and the model calculates the user's interest matching degree for each book by analyzing the correlation between the user's interest and the book attributes; for example For example, for user A and the book "Ordinary World", user A has borrowed the book "Ordinary World". Based on user A's borrowing history and search keywords, combined with the attribute characteristics of the book "Ordinary World", the similarity scores between other books and "Ordinary World" are calculated according to the set weights, thereby calculating the user's interest matching degree for each book; by combining user behavior data and book attribute characteristics to calculate the interest matching degree, it is possible to deeply analyze the user's interest preferences and screen out books that match the user's interests. Compared with the single-dimensional recommendation method, this solution combines multiple dimensions to improve the accuracy of book recommendations and the user's acceptance of recommended books, which helps to improve the user's reading experience.

[0113] Preferably, the books are sorted from high to low according to the calculated interest matching degree, and among the sorted books, the top-ranked books are selected according to the preset first quantity to form the first area recommended book set; for example, the system sets the preset first quantity to 5 books according to the user's book borrowing situation, and according to the sorting result, the top 5 books with the highest interest matching degree are selected as the first area recommended book set; filtering the recommended books according to the interest matching degree can quickly focus on the books that match the user's interests, making the book recommendation process more personalized, improving the efficiency and practicality of recommending books, and enhancing the user's satisfaction with the recommendation service.

[0114] Furthermore, by analyzing the book association between the hotspot area and the non-hotspot area, recommended books are screened in the non-hotspot area to obtain a second area recommended book set, including:

[0115] S701, obtaining book similarity by calculating the similarity between the book attribute features of the books in the non-hotspot area and the book attribute features of the books in the hotspot area;

[0116] S702, calculating the correlation between the non-hotspot area and the hotspot area based on the distance between the non-hotspot area and the hotspot area and the corresponding book similarity;

[0117] S703: Filter out a preset second number of non-hotspot areas according to the correlation degree from high to low to obtain a set of candidate areas;

[0118] S704: In each non-hotspot area in the candidate area set, a preset third number of recommended books are screened out according to the user's interest matching degree for books in the non-hotspot area from high to low to obtain a second area recommended book set.

[0119] In this embodiment, first, the book similarity is calculated by analyzing the similarity between the book attribute features of books in the hotspot area and the non-hotspot area, and the book attribute features of the books are extracted from the hotspot area and the non-hotspot area respectively. The book similarity is obtained by calculating the cosine similarity between the book attribute features. For each book in the non-hotspot area, its similarity with all the books in the hotspot area is calculated respectively, and the average is calculated to obtain the book similarity between each book in the non-hotspot area and the books in the hotspot area; by calculating the book similarity, the connection between the books in the non-hotspot area and the books in the hotspot area can be analyzed from the book attributes, thereby avoiding the situation of only focusing on the hotspot area and ignoring the valuable books in the non-hotspot area, and broadening the scope of recommended books.

[0120] Preferably, the correlation between the non-hotspot area and the hotspot area is analyzed based on the calculated book similarity and the distance between the non-hotspot area and the hotspot area; considering that in the spatial layout of the library, the distance between the non-hotspot area and the hotspot area affects the user's attention to the books in the non-hotspot area to a certain extent, the closer the distance, the greater the probability that the user will be exposed to the books in the area; at the same time, the book similarity reflects the degree of correlation of the book content, the book similarity of each book in the non-hotspot area is averaged, the distance between the non-hotspot area and the hotspot area is inverted, and the weighted sum is performed with the averaged book similarity to obtain the correlation, which can comprehensively consider spatial factors and content factors, more comprehensively analyze the correlation between the non-hotspot area and the hotspot area, quickly screen out the non-hotspot areas that are closely related to the hotspot area, and improve the pertinence and accuracy of book recommendations.

[0121] Specifically, according to the calculated correlation, the non-hotspot areas are sorted from high to low according to the correlation, and according to the preset second number, the non-hotspot areas with the highest ranking are selected to obtain a set of candidate areas; by screening the set of candidate areas, the recommendation range of the non-hotspot areas can be quickly narrowed down, avoiding blind screening of books in a large number of non-hotspot areas, improving the recommendation efficiency, and by screening out high-correlation areas, the recommended books are more in line with the user's interests, thereby improving the accuracy and effectiveness of book recommendations. In the candidate area set, the user's interest matching degree for each book is calculated through the book recommendation model, and the interest matching degree is sorted from high to low to screen out a preset third number of books. For example, the system sets the preset third number to 2 books based on the user's book borrowing situation, and selects the top 2 books with the highest interest matching degree in the candidate non-hotspot areas to form the second area recommended book set; by screening recommended books in the candidate areas, the user's reading choices are improved.

[0122] Furthermore, based on the borrowing status of users for the first recommended book set, the recommended books are optimized by a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books, including:

[0123] S801: Analyze changes in user interest based on the user's borrowing of books in the first recommended book set to obtain an interest vector;

[0124] S802: Optimize the recommended books using a preset recommendation optimization model in combination with the interest vector to obtain a second set of recommended books, so as to make personalized recommendations for library books.

[0125] In this embodiment, based on the user's borrowing of books in the first recommended book set, the user's interest changes are analyzed to obtain an interest vector; by analyzing the book borrowing situation, corresponding quantitative standards are set, and corresponding values ​​are set for each interest dimension. When a user borrows a certain type of book with a longer borrowing time and more borrowing times, the interest value corresponding to this type of book is higher; based on the user's book borrowing situation, the user's interest changes are analyzed to obtain an interest vector.

[0126] For example, user B borrows a literature book for 15 days, and this is the second time borrowing this type of book. The value of the literature dimension can be set to 8 (the full score is 10); when the user has not borrowed a certain type of book, the value of this dimension is set to 0, and the interest vector of user B is [8, 0, 0, 0, 0], indicating that user B has a high interest in literature books and a low interest in other categories. By constructing an interest vector, the user's abstract interest changes are converted into quantitative data, providing a data basis for optimizing book recommendations. According to the interest vector, the book recommendation strategy can be adjusted in time to improve the pertinence and timeliness of book recommendations and meet the user's ever-changing reading needs.

[0127] Specifically, based on the interest vector, the books in the first recommended book set are adjusted and optimized through a preset recommendation optimization model combined with the library's book resource information. Taking into account factors such as the degree of matching between the book and the interest vector, the popularity of the book, and the inventory status, books that are more in line with the user's current interest changes are screened out, and books that do not match the user's interests are removed to obtain a second recommended book set; the book recommendation results are dynamically adjusted according to changes in user interests, the matching degree between the recommended books and the user's interests is improved, and users are provided with personalized recommendations that are more in line with their interests, which helps to improve the book borrowing rate and the utilization efficiency of library resources.

[0128] Furthermore, the recommended books are optimized by combining the interest vectors with a preset recommendation optimization model to obtain a second set of recommended books, so as to make personalized recommendations for library books, including:

[0129] S901: Based on the interest vector and user behavior data, a preset recommendation optimization model is used to predict the user's interest migration to obtain an interest migration prediction vector;

[0130] S902: Based on the interest transfer prediction vector, select books that match the interest transfer prediction vector as an optimized book set;

[0131] S903: Optimize the recommended books by optimizing the book set to obtain a second recommended book set, so as to make personalized recommendations for library books.

[0132] In this embodiment, the user's interest migration is predicted based on the interest vector combined with the user behavior data through a preset recommendation optimization model to obtain an interest migration prediction vector; considering that the user's reading interest is not fixed and will change with factors such as reading behavior and life experience, the user's interest vector is integrated with the user behavior data, and the recommendation optimization model includes models such as time series analysis algorithm, association rule algorithm or deep learning-based prediction algorithm. The recommendation optimization model in this embodiment is a recurrent neural network model, which is trained with a large amount of historical data to obtain a pre-trained recurrent neural network model. The integrated data is input into the pre-trained recurrent neural network model, and the model outputs a predicted interest migration prediction vector; by predicting the user's interest migration prediction vector, it is possible to analyze the development trend of the user's interests in advance, recommend books of interest to the user in advance, improve the accuracy and foresight of the recommendation, and stimulate the user's new reading interest.

[0133] Specifically, based on the calculated interest migration prediction vector, by matching the attribute characteristics of the book with the interest migration prediction vector, the books that meet the interest migration prediction vector are screened out. First, the matching rules between the book attribute characteristics and the interest migration prediction vector are determined, and the book categories and book keywords are matched separately. The books whose book keywords have a preset proportion (such as 30%) of overlap with the user's recent search keywords or the keywords related to the interest book category in the interest vector are screened out as the optimized book collection; optimizing book screening based on the interest migration prediction vector can screen out books that meet the user's interest trends, improve recommendation efficiency, and meet the user's reading interest needs.

[0134] Preferably, the screened optimized book set is used to adjust and optimize the books in the first recommended book set to obtain a second recommended book set; the matching degree between each book in the first recommended book set and the interest migration prediction vector is calculated, and books with a matching degree greater than a preset ratio are retained in the first recommended book set, and books with a matching degree less than or equal to the preset ratio are removed. At the same time, the books in the optimized book set that are not in the first recommended book set are added to the recommendation list in order from high to low according to the matching degree to obtain a second recommended book set, and the books in the second recommended book set are sorted from high to low according to the matching degree, so that users can see the books that best suit their interests first; the recommended books are optimized by optimizing the book set, so as to realize dynamic updating and personalized adjustment of the recommended books, and the optimized recommended books are more in line with the user's changing interest needs, thereby improving the quality and effect of book recommendations.

[0135] Example 2

[0136] In this embodiment, if Figure 4 , provides a personalized recommendation system for library books, which is used to implement a personalized recommendation method for library books, including:

[0137] The data acquisition module obtains the book RFID tag flow data and user behavior data during the user's book borrowing process;

[0138] The book interest heat map construction module combines book RFID tag flow data and user behavior data to divide the library into grids according to the preset grid size, analyze the user's interest in the books in each grid area, and construct a book interest heat map;

[0139] The book recommendation module extracts hot spots with interest levels greater than a preset interest level threshold based on the book interest heat map, and selects recommended books based on the hot spots using a preset book recommendation model to obtain a first set of recommended books.

[0140] The recommended book optimization module optimizes the recommended books based on the user's borrowing situation of the first recommended book set through a preset recommendation optimization model to obtain a second recommended book set to make personalized recommendations for library books.

[0141] In this embodiment, the data acquisition module collects data generated during the user's book borrowing process from the library management system and related equipment, including book RFID tag flow data and user behavior data. The book RFID tag flow data records information such as book borrowing, returning, and movement trajectory, and the user behavior data includes behavioral data such as the user's stay time in the library, browsing habits, and search operations. Data collection provides data support for user interest analysis and book recommendations. The book interest heat map construction module divides the library space into multiple grid areas according to a pre-set grid size based on the data collected by the data acquisition module. Combined with the book RFID tag flow data and user behavior data, it analyzes the user's interest level in the books in each grid area through a preset interest analysis model, calculates the corresponding interest level value, and constructs a book interest heat map with the grid area where the user borrows the book as the center and the interest level value spreads to the surrounding areas, which can intuitively present the distribution trend of user interests.

[0142] Specifically, the book recommendation module is based on the book interest heat map and, through the set interest level threshold, screens out hot spots where user interests are concentrated. In the hot spots, the preset book recommendation model is used to analyze the user's interest preferences for different books, screens out books that meet the user's interests, and forms a first recommended book set. The module also analyzes the correlation between books in the hot spots and non-hot spots, screens out potential recommended books in the non-hot spots, and adds them to the recommended book set, providing users with a richer, more comprehensive, and more interest-oriented book recommendation list. The recommended book optimization module analyzes the dynamic changes of user interests based on the user's borrowing feedback on the first recommended book set, obtains the user's interest vector, and combines the user behavior data to predict the user's interest migration trend through the preset recommendation optimization model to obtain the interest migration prediction vector. Based on the interest migration prediction vector, matching books are screened out to form an optimized book set. The first recommended book set is optimized, unmatched books are removed, and books that match the user's interest change trend are added to obtain the second recommended book set, thereby realizing personalized library book recommendations and improving the user's borrowing experience and the utilization efficiency of library resources.

[0143] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A personalized recommendation method for library books, characterized in that: include: Obtain book RFID tag circulation data and user behavior data during the user's book borrowing process; Combining the book RFID tag flow data and user behavior data, the library is divided into grids according to a preset grid size, and the user interest in the books in each grid area is analyzed to construct a book interest heat map; Extracting hotspot areas with interest levels greater than a preset interest level threshold based on the book interest heat map, and filtering recommended books using a preset book recommendation model based on the hotspot areas to obtain a first recommended book set; Based on the borrowing status of the first recommended book set by the user, the recommended books are optimized by a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books; Combining the book RFID tag flow data and user behavior data, the library is divided into grids according to a preset grid size, and the user interest in the books in each grid area is analyzed to construct a book interest heat map, including: Divide the library into grids according to a preset grid size to obtain multiple grid areas; Combining book RFID tag flow data and user behavior data, the user's interest in books in each grid area is analyzed through a preset interest analysis model to obtain the interest level value of each grid area; Taking the grid area where the user's borrowed book is located as the center, the interest level value is combined with regional diffusion to the surrounding grid areas to construct a book interest heat map; The method combines the book RFID tag flow data and the user behavior data to analyze the user's interest in the books in each grid area through a preset interest analysis model to obtain the interest level value of each grid area, including: Based on the book RFID tag flow data, the first grid area where the user's borrowed books are located is extracted, and the time difference between book borrowing and returning is analyzed to obtain the book borrowing duration; Based on user behavior data, the user's book reading situation is analyzed through the preset interest analysis model to obtain the user behavior analysis results; Determining the interest level value of the first grid area based on the book borrowing duration and the user behavior analysis result; The interest level value of each grid area is calculated by combining the content similarity between the books in the grid areas with the interest level value of the first grid area.

2. A personalized recommendation method for library books according to claim 1, characterized in that: The method of constructing a book interest heat map by taking the grid area where the book borrowed by the user is located as the center and combining the interest level value to diffuse to the surrounding grid areas includes: The first grid area where the books borrowed by the user are located is used as the central grid area; Calculating the thermal value of each grid area based on the interest level value of each grid area and the distance between the grid area and the central grid area, wherein when there are multiple central grid areas, the thermal values ​​calculated for each central grid area are superimposed; Taking the central grid area as the center, regional diffusion is performed to the surrounding grid areas according to the thermal value to construct a book interest heat map.

3. A personalized recommendation method for library books according to claim 1, characterized in that: Extracting hotspot areas with interest levels greater than a preset interest level threshold based on the book interest heat map, and filtering out recommended books using a preset book recommendation model based on the hotspot areas to obtain a first recommended book set, including: According to the book interest heat map, extract the second grid areas where the interest level is greater than a preset interest level threshold, and merge adjacent second grid areas to obtain multiple hot spots; In each hotspot area, the user's interest in different books is analyzed through a preset book recommendation model to filter recommended books and obtain a first area recommended book set; By analyzing the book correlation between the hotspot area and the non-hotspot area, recommended books are screened in the non-hotspot area to obtain a set of recommended books for the second area; The first region recommended book set and the second region recommended book set are combined to obtain a first recommended book set.

4. A personalized recommendation method for library books according to claim 3, characterized in that: In each hotspot area, the user's interest in different books is analyzed by a preset book recommendation model to filter recommended books, thereby obtaining a first area recommended book set, including: In each hotspot area, the attributes of each book are analyzed to obtain the book attribute characteristics; Based on the user behavior data and the book attribute characteristics, the user's interest matching degree for each book is calculated through a preset book recommendation model; According to the interest matching degree from high to low, a preset first number of recommended books are screened out to obtain a first region recommended book set.

5. A personalized recommendation method for library books according to claim 3, characterized in that: The method of analyzing the book association between the hotspot area and the non-hotspot area, screening recommended books in the non-hotspot area, and obtaining a second area recommended book set includes: The book similarity is obtained by calculating the similarity between the book attribute features of the books in the non-hotspot area and the book attribute features of the books in the hotspot area; The correlation between non-hotspot areas and hotspot areas is calculated by combining the distance between them and the corresponding book similarities. Screening out a preset second number of non-hotspot areas according to the correlation degree from high to low to obtain a set of candidate areas; In each non-hotspot area in the candidate area set, a preset third number of recommended books are screened out according to the user's interest matching degree for books in the non-hotspot area from high to low to obtain a second area recommended book set.

6. A personalized recommendation method for library books according to claim 1, characterized in that: The method of optimizing the recommended books based on the borrowing status of the first recommended book set by the user through a preset recommendation optimization model to obtain a second recommended book set to make personalized recommendations for library books includes: According to the user's borrowing of books in the first recommended book set, the user's interest changes are analyzed to obtain an interest vector; In combination with the interest vector, the recommended books are optimized through a preset recommendation optimization model to obtain a second recommended book set, so as to make personalized recommendations for library books.

7. A personalized recommendation method for library books according to claim 6, characterized in that: In combination with the interest vector, the recommended books are optimized by a preset recommendation optimization model to obtain a second set of recommended books, so as to make personalized recommendations for library books, including: Based on the interest vector and user behavior data, the user's interest migration is predicted through a preset recommendation optimization model to obtain an interest migration prediction vector; Based on the interest migration prediction vector, screening out books that match the interest migration prediction vector as an optimized book set; The recommended books are optimized by optimizing the book set to obtain a second recommended book set, so as to make personalized recommendations for library books.

8. A personalized recommendation system for library books, characterized by: A method for implementing a personalized recommendation method for library books as claimed in any one of claims 1 to 7, comprising: The data acquisition module obtains the book RFID tag flow data and user behavior data during the user's book borrowing process; The book interest heat map construction module combines the book RFID tag flow data and user behavior data, divides the library into grids according to a preset grid size, analyzes the user's interest in the books in each grid area, and constructs a book interest heat map; a book recommendation module, which extracts hotspot areas having interest levels greater than a preset interest level threshold based on the book interest heat map, and filters out recommended books using a preset book recommendation model based on the hotspot areas to obtain a first set of recommended books; The recommended book optimization module optimizes the recommended books based on the user's borrowing status of the first recommended book set through a preset recommendation optimization model to obtain a second recommended book set to make personalized recommendations for library books.

Images