A book recommendation method and system based on hybrid recommendation algorithm

Abstract

The present invention relates to a book recommendation method and system based on a hybrid recommendation algorithm, and relates to the field of data processing technology, including: based on a book title set and a user portrait label, obtaining user short-term interest recommended books, user long-term interest recommended books, hot spot recommended books, and content balanced recommended books; according to the user portrait label, obtaining the user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight, and content balanced recommendation weight; according to the user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight, and content balanced recommendation weight, obtaining a book fusion recommendation value; according to the book fusion recommendation value, obtaining a book sorting result; sending the book title set to a user-side visualization interface for display according to the book sorting result. The technical problem that the lack of effective fusion of various types of book recommendations in the recommendation results in the prior art affects the user experience is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a book recommendation method and system based on a hybrid recommendation algorithm. Background Art

[0002] With the development of information technology, book recommendation systems have been widely used in scenarios such as libraries, e-book platforms, and social reading websites. Traditional recommendation systems mainly rely on collaborative filtering, content-based recommendations, and rule-based recommendation methods. When recommending books, these methods often use algorithms to calculate the books that users may be interested in based on information such as users' historical borrowing records, rating data, and browsing habits. However, a single algorithm has certain limitations in dealing with changes in user interests and information diversity, and the accuracy and personalization of recommendation results are low. To solve this problem, hybrid recommendation algorithms have gradually become a research hotspot in the field of book recommendations, combining the advantages of multiple recommendation methods to achieve more accurate and personalized recommendations.

[0003] Existing book recommendation methods usually have the following shortcomings: First, a single recommendation algorithm is difficult to effectively respond to diverse user needs, and it is easy to have a single or biased recommendation result. Second, the traditional recommendation system does not clearly distinguish between short-term and long-term interests, and cannot effectively capture the dynamic changes of user interests. Third, there is a lack of balance between hot books and personalized recommendations in the recommendation results, which can easily lead to poor user experience. In addition, the recommendation system has deficiencies in the response speed to user feedback and the dynamic adjustment of recommendation weights, and fails to update and optimize the recommendation strategy in a timely manner. Therefore, the current recommendation system still has a lot of room for improvement in terms of accuracy, diversity, real-time and user experience. Summary of the invention

[0004] The present invention aims to solve the technical problem in the prior art that various types of book recommendations in the recommendation results lack effective integration, which affects the user experience. A book recommendation method and system based on a hybrid recommendation algorithm are provided to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] In a first aspect, the present invention provides a book recommendation method based on a hybrid recommendation algorithm, comprising:

[0007] Obtain a collection of book titles through the library database;

[0008] Receive user portrait tags through the user end;

[0009] Based on the book title set and the user portrait label, multiple algorithms are executed in parallel to recommend books to obtain books recommended by users’ short-term interests, books recommended by users’ long-term interests, books recommended by hot topics, and books recommended by balanced content;

[0010] According to the user portrait tag, multi-algorithm user feedback information is collected, and algorithm weight analysis is performed to obtain user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights;

[0011] According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain a book fusion recommendation value;

[0012] According to the book fusion recommendation value, the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books are serialized and adjusted from large to small to obtain a book ranking result;

[0013] The book title set is sent to a visualization interface of a user terminal for display according to the book sorting result.

[0014] In a second aspect, the present invention provides a book recommendation system based on a hybrid recommendation algorithm, comprising:

[0015] The book title acquisition module is used to obtain a collection of book titles through the library database;

[0016] A user portrait acquisition module is used to receive user portrait tags through the user end;

[0017] A multi-algorithm recommendation module, for executing multi-algorithm parallel recommendation based on the book title set and the user portrait tag, to obtain books recommended by the user's short-term interests, books recommended by the user's long-term interests, hot-spot recommended books and content-balanced recommended books;

[0018] An algorithm weight analysis module is used to collect multi-algorithm user feedback information according to the user portrait label, perform algorithm weight analysis, and obtain user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight and content balance recommendation weight;

[0019] A weight frequency evaluation module is used to perform weight frequency evaluation on the books recommended by the user's short-term interests, the books recommended by the user's long-term interests, the books recommended by the hot spots, and the books recommended by the content balance according to the weight of the recommendation, the weight of the recommendation, and the weight of the content balance recommendation, so as to obtain a book fusion recommendation value;

[0020] A book sorting module is used to perform serial adjustment on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books from large to small according to the book fusion recommendation value to obtain a book sorting result;

[0021] The book ranking recommendation module is used to send the book title set to the visual interface of the user end for display according to the book ranking result.

[0022] In a third aspect, the present application provides a non-transitory computer-readable storage medium, in which a computer software program is stored. When the computer software program is executed by a processor, the method for recommending books based on a hybrid recommendation algorithm described in the first aspect is implemented.

[0023] The beneficial effects of the present invention are as follows: a set of book titles is obtained through a library database, and a user portrait tag is received through a user terminal. Based on the book title set and the user portrait tag, multiple algorithms are performed in parallel to recommend books, and books recommended by users' short-term interests, books recommended by users' long-term interests, books recommended by hot spots, and books recommended by content balance are obtained. Combined with the user portrait tag, user feedback information of multiple algorithms is collected, and algorithm weight analysis is performed to obtain the weights of various recommended books. Then, a weight frequency evaluation is performed on these recommended books, the book fusion recommendation value is calculated, and various books are sorted and displayed on the user terminal according to the fusion recommendation value. Through multiple algorithm parallel recommendation, a combination of short-term interest, long-term interest, hot spots and content balance recommendation is achieved, which solves the problem that a single algorithm is difficult to meet the diverse needs of users, increases the diversity of recommendation results, performs a weight frequency evaluation on the recommended books, calculates the book fusion recommendation value, and performs serialization adjustment on various books, so that the recommendation results are displayed according to the weight, ensuring the balance between hot books and personalized recommendations, and optimizing the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart of a book recommendation method based on a hybrid recommendation algorithm provided by the present invention;

[0025] Figure 2 A flowchart of the steps for recommending hot book recommendations in a book recommendation method based on a hybrid recommendation algorithm provided by the present invention;

[0026] Figure 3 A structural schematic diagram of a book recommendation system based on a hybrid recommendation algorithm provided by the present invention;

[0027] Figure 4 A schematic diagram of the structure of a computer-readable storage medium provided by the present invention.

[0028] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0029] Book title acquisition module 1000, user portrait acquisition module 2000, multi-algorithm recommendation module 3000, algorithm weight analysis module 4000, weight frequency evaluation module 5000, book sorting module 6000, book sorting recommendation module 7000, computer readable storage medium 600, computer program 611. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0031] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0032] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or description". Any embodiment described as "for example" in the present invention is not necessarily to be interpreted as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that one of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details.

[0033] In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the present invention with unnecessary detail. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope consistent with the principles and features disclosed in the present invention.

[0034] Embodiment 1:

[0035] like Figure 1 As shown, the embodiment of the present invention provides a book recommendation method based on a hybrid recommendation algorithm, comprising the steps of:

[0036] S10: Obtain a collection of book titles through a library database;

[0037] By connecting to the library database, we extract all the recommended book titles. These titles include basic information about the books, such as title, author, category, year of publication, etc. Obtaining a complete set of book titles provides a basic data source for the subsequent recommendation algorithm, which is the starting point of the entire recommendation process. This step ensures that the recommendation system can access the latest and most comprehensive book information, laying a data foundation for subsequent personalized recommendations.

[0038] S20: receiving a user portrait tag through the user terminal;

[0039] User portrait tags are collected through the user end. These tags include basic information and behavioral characteristics of users, such as age, gender, occupation, education, hobbies, reading history and preferences. User portrait tags are used to describe the personalized characteristics of users and are the core basis for personalized calculations of recommendation algorithms. By obtaining these tags, the system can more accurately analyze the interests of users, thereby achieving targeted recommendations in the subsequent recommendation process, improving the relevance of recommendations and user satisfaction.

[0040] S30: Based on the book category set and the user portrait label, executing multi-algorithm parallel recommendation to obtain books recommended by the user's short-term interests, books recommended by the user's long-term interests, hot-spot recommended books, and content-balanced recommended books;

[0041] By using the book title collection and user portrait tags, multiple recommendation algorithms are executed in parallel to obtain multiple types of recommended books. Specifically, the multi-algorithm parallel recommendation includes: the short-term interest recommendation algorithm recommends books related to the current interests based on the user's recent reading behavior and interest changes; the long-term interest recommendation algorithm recommends books related to stable preferences based on the user's long-term reading history; the hot spot recommendation algorithm recommends books with wide attention to users by analyzing the current popular books; the content balanced recommendation algorithm balances different types of recommendations based on the diversity of book content to avoid the singleness of recommendations.

[0042] Through the combination of these algorithms, multi-dimensional recommendations are achieved to meet the needs of users in different time periods and different interests, and enhance the diversity and targeting of recommendations.

[0043] S40: According to the user portrait tag, collect multi-algorithm user feedback information, perform algorithm weight analysis, and obtain user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight, and content balance recommendation weight;

[0044] Based on user portrait tags, user feedback information generated by various recommendation algorithms is collected, such as user behavior data such as clicks, borrowing, and ratings of recommended books. By analyzing this feedback information, the system can evaluate the actual effect of each recommendation algorithm and perform algorithm weight analysis. The core of weight analysis is to calculate the performance of various recommendation algorithms, and then assign weights to short-term interest recommendations, long-term interest recommendations, hot spot recommendations, and content balanced recommendations, which are stored as user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balanced recommendation weights. These weights reflect the applicability and priority of different recommendation algorithms in the current user scenario, which helps to optimize the recommendation combination, ensure that the recommendation results are more in line with the actual needs and interests of users, and improve the accuracy of recommendations and user satisfaction.

[0045] S50: According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain a book fusion recommendation value;

[0046] Use the previously obtained user short-term interest recommendation weights, long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights to perform a weighted frequency evaluation on the corresponding recommended books. Specifically, the system calculates the comprehensive book fusion recommendation value based on the weight and frequency of each recommended book in different recommendation types. This fusion recommendation value is obtained by weighted calculation of the contribution of each type of recommended book, reflecting the overall applicability of the book after integrating multiple recommendation algorithms. Through weighted frequency evaluation, the system can effectively integrate different recommendation results, avoid the bias of a single algorithm, and ensure that the final recommended books integrate multiple interests and content balance considerations, thereby providing more accurate and personalized recommendations and optimizing user experience.

[0047] S60: According to the book fusion recommendation value, the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books are serialized and adjusted from large to small to obtain a book ranking result;

[0048] According to the previously calculated book fusion recommendation values, all recommended books (including books recommended by users' short-term interests, long-term interests, hot topics, and balanced content) are sorted. The system first uses the fusion recommendation values ​​of the books to sort the recommended books from large to small, ensuring that books with high recommendation values ​​are displayed first. This sorting process ensures that the final recommendation results can both highlight the most relevant books for the user and maintain the diversity and balance of the content, thereby optimizing the user's reading experience and improving the overall effect of the recommendation system.

[0049] S70: Sending the book title set to a visual interface of a user terminal for display according to the book sorting result.

[0050] Furthermore, the step of recommending books based on the user's short-term interests or books based on the user's long-term interests includes S31: Step S31 includes the steps of:

[0051] S311: Obtaining a sample collection time zone, wherein when recommending a user's short-term interests, the sample collection time zone is the past six months, and when recommending a user's long-term interests, the sample collection time zone is the past two years;

[0052] S312: According to the user portrait tag and the sample collection time zone, collect a first book title borrowing frequency set and a first book title borrowing duration set of the book title set by users with similar portraits;

[0053] S313: performing normalization processing on the first book title borrowing frequency set to obtain a first book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a first weight;

[0054] S314: performing normalization processing on the first book title borrowing time set to obtain a first book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a second weight;

[0055] S315: According to the first weight and the second weight, the first book title borrowing frequency normalization parameter set and the first book title borrowing duration normalization parameter set are added to obtain a first book recommendation value of the book title set;

[0056] S316: Sort the first book recommendation values ​​from large to small, select a preset number of books with head numbers, and obtain the recommended books for the user's short-term interests or the recommended books for the user's long-term interests.

[0057] First, the recommendation algorithm steps for user short-term interest recommendation and user long-term interest recommendation are the same. Specifically, when recommending books for user short-term interests, the past six months are selected as the sample collection time zone, and when recommending books for user long-term interests, the past two years are selected as the sample collection time zone to distinguish between users' recent interest changes and long-term stable preferences. Then, other users with similar profiles to the target user are selected based on user profile tags, and the borrowing frequency and borrowing duration data of these similar users in the book title set are collected.

[0058] After normalization, these data are used to generate a borrowing frequency parameter with a first weight and a borrowing duration parameter with a second weight. By weighted calculation of these two parameters, the first book recommendation value of each book is obtained. Then, the system sorts the books from large to small according to the recommendation value, and selects the top-ranked books as the recommendation results. In this way, by analyzing the reading behavior of similar users and combining weight evaluation, the system can accurately recommend books that meet the user's short-term and long-term interests, thereby improving the relevance of the recommendation and user experience.

[0059] Further, according to the user portrait tag, a book title borrowing frequency set and a book title borrowing duration set of the book title set by users with similar portraits are collected. Step S312 includes the following steps:

[0060] S3121: The user portrait tag includes age, education, gender, and occupation, wherein the age has an age matching weight, the education has an education matching weight, the gender has a gender matching weight, and the occupation has an occupation matching weight, and the sum of the age matching weight, the education matching weight, the gender matching weight, and the occupation matching weight is equal to 1;

[0061] S3122: constructing a portrait similarity evaluation model according to the age, the education level, the gender, and the occupation, in combination with the age matching weight, the education level matching weight, the gender matching weight, and the occupation matching weight, wherein the portrait similarity evaluation model is constructed based on a portrait similarity evaluation function;

[0062] S3123: According to the portrait similarity evaluation model, extract users whose portrait similarity coefficient is greater than or equal to a portrait similarity coefficient threshold as the similar portrait users.

[0063] First, user portrait labels include basic features such as age, education, gender and occupation. These features are assigned corresponding matching weights, such as age matching weight, education matching weight, gender matching weight and occupation matching weight, and the sum of all weights is equal to 1 to ensure the rationality of the overall weight distribution.

[0064] Next, based on the user's age, education, gender, occupation and other labels, and combined with their respective matching weights, the system built a portrait similarity evaluation model, which operates based on a specific portrait similarity evaluation function. Through this model, the system can calculate the similarity between users and screen out users whose portrait similarity coefficient is greater than or equal to the portrait similarity coefficient threshold (the threshold is pre-set by the manager), defining them as similar portrait users. The borrowing frequency and borrowing duration data of these similar portrait users will be used for further recommendation calculations to ensure that the recommendation results can reflect the group behavior that is most similar to the target user's interests, and improve the accuracy and personalization of the recommendations.

[0065] Furthermore, the portrait similarity evaluation function is: in, Characterizes the image similarity coefficient, Characterize the first user portrait label and weight parameter, Characterize the second user portrait label and weight parameter, Representing age matching weights, Represents the academic qualification matching weight, Represents the gender matching weight, Characterizes the occupation matching weight, Characterizes the age of the first user, Characterizes the first user's education level, Characterizes the gender of the first user, Characterize the first user occupation, Characterize the second user's age, Represents the second user's academic qualifications, Characterizes the second user's gender, Characterizes the second user's occupation.

[0066] Further, such as Figure 2 As shown, the step of recommending hot recommended books includes step S32, and step S32 includes the steps of:

[0067] S321: Collecting borrowing record data of the set of book titles in a preset time zone;

[0068] S322: According to the borrowing record data in the preset time zone, traverse the book title set to perform statistics, and obtain a second book title borrowing frequency set and a second book title borrowing duration set;

[0069] S323: performing normalization processing on the second book title borrowing frequency set to obtain a second book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a third weight;

[0070] S324: performing normalization processing on the second book title borrowing time set to obtain a second book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a fourth weight;

[0071] S325: According to the third weight and the fourth weight, the second book title borrowing frequency normalization parameter set and the second book title borrowing duration normalization parameter set are added to obtain a second book recommendation value of the book title set;

[0072] S326: Sort the second book recommendation values ​​from large to small, select a preset number of books with head serial numbers, and obtain the hot recommended books.

[0073] First, the system collects the borrowing record data of the book title set in the preset time zone, and these data are used to analyze the popularity of books in the current time period. Then, the system traverses and counts the collected borrowing data to form a second book title borrowing frequency set and a borrowing duration set, which respectively reflect the number of borrowing times and borrowing duration of the book. Subsequently, the two sets are normalized to generate a borrowing frequency normalization parameter with a third weight and a borrowing duration normalization parameter with a fourth weight. By weighted summing the two normalized parameter sets, the system calculates the second book recommendation value of each book, which reflects the comprehensive popularity of the book in the current time zone. Finally, the system sorts the books from large to small according to the recommendation value, and selects several books with high rankings as hot recommended books. This process effectively captures the most popular books in the current time period, enabling the recommendation system to quickly respond to users' needs for hot content, improving the timeliness of recommendations and user interest stickiness.

[0074] Furthermore, the step of recommending books with balanced content includes step S33, and step S33 includes the steps of:

[0075] S331: performing cluster analysis on the set of book titles according to book categories to obtain multiple clusters of book titles;

[0076] S332: traversing the multiple clusters of book titles and performing hotspot recommendation analysis respectively to obtain multiple groups of sub-hotspot recommended books, wherein the multiple groups of sub-hotspot recommended books have multiple groups of third recommendation values;

[0077] S333: dividing the preset number of recommended books based on the number of clusters to obtain the number of recommendations for a single cluster;

[0078] S334: According to the number of single cluster recommendations, traverse the multiple groups of sub-hotspot recommended books and sort them according to the multiple groups of third recommendation values ​​from large to small to obtain multiple clusters of selected recommended books, wherein the multiple clusters of selected recommended books have multiple clusters of selected third recommendation values;

[0079] S335: According to the third recommendation values ​​selected by the multiple clusters, the recommended books selected by the multiple clusters are serially adjusted from large to small to obtain the content-balanced recommended books.

[0080] First, the system performs cluster analysis on the book title set according to the book category, and divides the books into multiple category clusters to ensure the diversity of recommendations. Next, the system traverses these category clusters and performs hot spot recommendation analysis on each cluster, thereby obtaining multiple groups of sub-hot spot recommended books. Each group of sub-hot spot recommended books has its own third recommendation value, indicating their popularity in their respective categories. Then, the preset number of recommendations is divided according to the number of category clusters, and the number of recommendations for each cluster is determined to ensure that each category has an appropriate number of book recommendations. The system then sorts the multiple groups of sub-hot spot recommended books from large to small according to the third recommendation values ​​of each group, selects multiple clusters of selected recommended books, and assigns multiple clusters of selected third recommendation values ​​to these books.

[0081] Finally, the system serializes and adjusts the selected recommended books according to these third recommendation values ​​to form the final content-balanced recommended book set. This step avoids the problem of recommendations being concentrated on a certain type of book by balancing the popularity and quantity among categories, thereby achieving diversification and balance of recommended content and improving user experience.

[0082] Further, according to the user portrait tag, multi-algorithm user feedback information is collected, and algorithm weight analysis is performed to obtain user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights. Step S40 includes the following steps:

[0083] S41: obtaining first user feedback information of a user short-term interest recommendation algorithm, wherein the first user feedback information includes a plurality of first user borrowing status information, wherein any one of the first user borrowing status information is a borrowing status or a non-borrowing status;

[0084] S42: Counting the borrowing status proportions of the borrowing status information of the plurality of first users, and setting the proportions as weight factors of the user short-term interest recommendation algorithm;

[0085] S43: traverse the user long-term interest recommendation algorithm, the hot spot recommendation algorithm and the content balance recommendation algorithm to obtain the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the content balance recommendation algorithm;

[0086] S44: Add the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm, and the weight factor of the content balance recommendation algorithm to obtain a sum of the weight factors;

[0087] S45: Calculate the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balance recommendation algorithm respectively, and the ratio of the sum of the weight factors to obtain the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balance recommendation weight.

[0088] First, the system obtains the first user feedback information of the user short-term interest recommendation algorithm. These feedback information include several borrowing status information, indicating whether the user has actually borrowed the recommended books. By counting the proportion of borrowing status in these feedback information, the system calculates the weight factor of the user short-term interest recommendation algorithm. Subsequently, the system traverses the user long-term interest recommendation algorithm, the hot spot recommendation algorithm and the content balance recommendation algorithm, and obtains their respective weight factors in the same way, including the user long-term interest recommendation algorithm weight factor, the hot spot recommendation algorithm weight factor and the content balance recommendation algorithm weight factor. Next, these four weight factors are added together to obtain the sum of the weight factors. Finally, the system calculates the ratio of the weight factor of each algorithm to the sum result, thereby obtaining the final user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight and content balance recommendation weight. Through this process, the system can dynamically adjust the influence of each recommendation algorithm in the overall recommendation, optimize the recommendation strategy according to the actual feedback of the user, and thus improve the accuracy of the recommendation and user experience.

[0089] Further, according to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, the weight frequency of the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books are evaluated to obtain the book fusion recommendation value. Step S50 includes the steps of:

[0090] S51: Obtain a first recommended book according to the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books, wherein the first recommended book has a user short-term interest recommended serial number tag and / or a user long-term interest recommended serial number tag and / or a hot spot recommended serial number tag and / or a content balanced recommended serial number tag;

[0091] S52: multiplying the user short-term interest recommendation weight and the inverse of the user short-term interest recommendation sequence number label to obtain a first fusion parameter;

[0092] S53: multiplying the user's long-term interest recommendation weight and the inverse of the user's long-term interest recommendation sequence number label to obtain a second fusion parameter;

[0093] S54: multiplying the hotspot recommendation weight by the inverse of the hotspot recommendation sequence number label to obtain a third fusion parameter;

[0094] S55: multiplying the content balanced recommendation weight and the inverse of the content balanced recommendation sequence number label to obtain a fourth fusion parameter;

[0095] S56: Add the first fusion parameter, the second fusion parameter, the third fusion parameter and the fourth fusion parameter to obtain a first recommended book fusion recommendation value, and add the first recommended book fusion recommendation value to the book fusion recommendation value.

[0096] First, the system determines the first recommended book based on the user's short-term interest recommended books, long-term interest recommended books, hot spot recommended books and content balanced recommended books. The first recommended book may contain multiple recommendation serial number labels, such as short-term interest, long-term interest, hot spot and content balanced serial number labels. Then, the system uses the inverse of these serial number labels to multiply their respective weights. For example, the user's short-term interest recommendation weight is multiplied by the inverse of its serial number label to obtain the first fusion parameter, the long-term interest recommendation weight is multiplied by the inverse of the long-term interest serial number label to obtain the second fusion parameter, the hot spot recommendation weight is multiplied by the inverse of the hot spot serial number label to obtain the third fusion parameter, and the content balance recommendation weight is multiplied by the inverse of the content balance serial number label to obtain the fourth fusion parameter. Finally, the system adds all the fusion parameters to obtain the first recommended book fusion recommendation value of the book, and adds it to the overall book fusion recommendation value. In this way, the system effectively combines the weights and rankings of multiple recommendation results, realizes the precise fusion of book recommendations, optimizes the recommendation order finally displayed to users, and makes the recommended content more in line with the interests and needs of users.

[0097] The book recommendation method based on the hybrid recommendation algorithm provided by the embodiment of the present invention has at least the following technical effects:

[0098] The book title set is obtained through the library database, and the user portrait label is received through the user end. Based on the book title set and the user portrait label, multiple algorithms are performed in parallel to recommend books based on user short-term interests, books based on user long-term interests, books based on hot topics, and books based on balanced content. Combined with the user portrait label, user feedback information of multiple algorithms is collected, and algorithm weight analysis is performed to obtain the weights of various recommended books. Then, the weight frequency of these recommended books is evaluated, the book fusion recommendation value is calculated, and various books are sorted and displayed on the user end according to the fusion recommendation value. Through the parallel recommendation of multiple algorithms, the combination of short-term interests, long-term interests, hot topics, and balanced content recommendations is realized, which solves the problem that a single algorithm is difficult to meet the diverse needs of users, increases the diversity of recommendation results, evaluates the weight frequency of recommended books, calculates the book fusion recommendation value, and performs serialization adjustment on various books, so that the recommendation results are displayed according to the weight, ensuring the balance between hot books and personalized recommendations, and optimizing the user experience.

[0099] Embodiment 2:

[0100] like Figure 3 As shown, based on the same inventive concept as the book recommendation method based on the hybrid recommendation algorithm provided in the first embodiment, the embodiment of the present invention further provides a book recommendation system based on the hybrid recommendation algorithm, including:

[0101] The book title acquisition module 1000 is used to obtain a book title set through a library database;

[0102] The user portrait obtaining module 2000 is used to receive the user portrait tag through the user terminal;

[0103] The multi-algorithm recommendation module 3000 is used to perform multi-algorithm parallel recommendation based on the book title set and the user portrait tag to obtain books recommended by the user's short-term interests, books recommended by the user's long-term interests, hot-spot recommended books and content-balanced recommended books;

[0104] The algorithm weight analysis module 4000 is used to collect multi-algorithm user feedback information according to the user portrait label, perform algorithm weight analysis, and obtain user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight and content balance recommendation weight;

[0105] The weight frequency evaluation module 5000 is used to perform weight frequency evaluation on the books recommended by the user's short-term interests, the books recommended by the user's long-term interests, the books recommended by the hot spots, and the books recommended by the content balance according to the weight of the recommendation, the weight of the recommendation, and the weight of the content balance recommendation, so as to obtain a book fusion recommendation value;

[0106] The book ranking module 6000 is used to perform serial adjustment on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books from large to small according to the book fusion recommendation value to obtain a book ranking result;

[0107] The book ranking recommendation module 7000 is used to send the book title set to the visual interface of the user end for display according to the book ranking result.

[0108] Furthermore, the steps of recommending books of user's short-term interests or books of user's long-term interests by the multi-algorithm recommendation module 3000 include:

[0109] Obtaining a sample collection time zone, wherein when recommending a user's short-term interests, the sample collection time zone is the past six months, and when recommending a user's long-term interests, the sample collection time zone is the past two years;

[0110] According to the user portrait label and the sample collection time zone, a first book title borrowing frequency set and a first book title borrowing duration set of the book title set by users with similar portraits are collected;

[0111] Normalizing the first book title borrowing frequency set to obtain a first book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a first weight;

[0112] Normalizing the first book title borrowing time set to obtain a first book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a second weight;

[0113] According to the first weight and the second weight, the first book title borrowing frequency normalization parameter set and the first book title borrowing duration normalization parameter set are added to obtain a first book recommendation value of the book title set;

[0114] According to the first book recommendation value, the books are sorted from large to small, and a preset number of books with head serial numbers are selected to obtain the user's short-term interest recommended books or the user's long-term interest recommended books.

[0115] Furthermore, the multi-algorithm recommendation module 3000 collects a set of book title borrowing frequencies and a set of book title borrowing durations of the book title set by users with similar profiles according to the user profile tags, including:

[0116] The user portrait tag includes age, education, gender, and occupation, wherein the age has an age matching weight, the education has an education matching weight, the gender has a gender matching weight, and the occupation has an occupation matching weight, and the sum of the age matching weight, the education matching weight, the gender matching weight, and the occupation matching weight is equal to 1;

[0117] According to the age, the education level, the gender, and the occupation, a portrait similarity evaluation model is constructed in combination with the age matching weight, the education level matching weight, the gender matching weight, and the occupation matching weight, wherein the portrait similarity evaluation model is constructed based on a portrait similarity evaluation function;

[0118] According to the portrait similarity evaluation model, users whose portrait similarity coefficient is greater than or equal to a portrait similarity coefficient threshold are extracted as the similar portrait users.

[0119] Furthermore, the portrait similarity evaluation function is: in, Characterizes the image similarity coefficient, Characterize the first user portrait label and weight parameter, Characterize the second user portrait label and weight parameter, Representing age matching weights, Represents the academic qualification matching weight, Represents the gender matching weight, Characterizes the occupation matching weight, Characterizes the age of the first user, Characterizes the first user's education level, Characterizes the gender of the first user, Characterize the first user occupation, Characterize the second user's age, Represents the second user's academic qualifications, Characterizes the second user's gender, Characterizes the second user's occupation.

[0120] Furthermore, the recommendation step of the hot recommended books of the multi-algorithm recommendation module 3000 includes:

[0121] Collecting borrowing record data of the set of book titles in a preset time zone;

[0122] According to the borrowing record data of the preset time zone, the book title set is traversed and counted to obtain a second book title borrowing frequency set and a second book title borrowing duration set;

[0123] Normalizing the second book title borrowing frequency set to obtain a second book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a third weight;

[0124] Normalizing the second book title borrowing time set to obtain a second book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a fourth weight;

[0125] According to the third weight and the fourth weight, the second book title borrowing frequency normalization parameter set and the second book title borrowing duration normalization parameter set are added to obtain a second book recommendation value of the book title set;

[0126] According to the second book recommendation value, the books are sorted from large to small, and a preset number of books with head serial numbers are selected to obtain the hot recommended books.

[0127] Furthermore, the content-balanced book recommendation step of the multi-algorithm recommendation module 3000 includes:

[0128] According to the book categories, cluster analysis is performed on the book title set to obtain multiple clusters of book titles;

[0129] Traversing the multiple clusters of book titles and performing hot spot recommendation analysis respectively to obtain multiple groups of sub-hot spot recommended books, wherein the multiple groups of sub-hot spot recommended books have multiple groups of third recommendation values;

[0130] Divide the preset number of recommended books based on the number of clusters to obtain the number of recommendations for a single cluster;

[0131] According to the number of single cluster recommendations, traverse the multiple groups of sub-hotspot recommended books and sort them according to the multiple groups of third recommendation values ​​from large to small to obtain multiple clusters of selected recommended books, wherein the multiple clusters of selected recommended books have multiple clusters of selected third recommendation values;

[0132] According to the multi-cluster selected third recommendation values, the multi-cluster selected recommended books are serially adjusted from large to small to obtain the content-balanced recommended books.

[0133] Furthermore, the algorithm weight analysis module 4000 collects multi-algorithm user feedback information according to the user portrait tag, performs algorithm weight analysis, and obtains user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight and content balance recommendation weight, including:

[0134] Obtaining first user feedback information of a user short-term interest recommendation algorithm, wherein the first user feedback information includes a plurality of first user borrowing status information, wherein any one of the first user borrowing status information is a borrowing status or a non-borrowing status;

[0135] Counting the borrowing status proportions of the borrowing status information of the plurality of first users, and setting the proportions as weight factors of the user short-term interest recommendation algorithm;

[0136] Traverse the user long-term interest recommendation algorithm, hot spot recommendation algorithm and content balanced recommendation algorithm to obtain the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balanced recommendation algorithm;

[0137] Adding the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balance recommendation algorithm to obtain a sum of the weight factors;

[0138] Calculate the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balance recommendation algorithm respectively, and ratio them to the sum of the weight factors to obtain the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balance recommendation weight.

[0139] Further, the book ranking module 6000: performs weight frequency evaluation on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books according to the user short-term interest recommendation weight, the user long-term interest recommended weight, the hot spot recommended books and the content balanced recommended books to obtain a book fusion recommendation value, including:

[0140] Obtain a first recommended book according to the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books, wherein the first recommended book has a user short-term interest recommended serial number tag and / or a user long-term interest recommended serial number tag and / or a hot spot recommended serial number tag and / or a content balanced recommended serial number tag;

[0141] The user short-term interest recommendation weight and the inverse of the user short-term interest recommendation sequence number label are multiplied and fused to obtain a first fusion parameter;

[0142] The user's long-term interest recommendation weight and the inverse of the user's long-term interest recommendation sequence number label are multiplied and fused to obtain a second fusion parameter;

[0143] The hotspot recommendation weight is multiplied by the inverse of the hotspot recommendation sequence number label to obtain a third fusion parameter;

[0144] The content balanced recommendation weight and the inverse of the content balanced recommendation sequence number label are multiplied and fused to obtain a fourth fusion parameter;

[0145] The first fusion parameter, the second fusion parameter, the third fusion parameter and the fourth fusion parameter are added to obtain a first recommended book fusion recommendation value, which is added to the book fusion recommendation value.

[0146] Embodiment three:

[0147] See also Figure 4 , Figure 4 A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 4 As shown, this embodiment provides a computer-readable storage medium 600, on which a computer program 611 is stored. When the computer program 611 is executed by a processor, the following steps are implemented:

[0148] Obtain a collection of book titles through the library database;

[0149] Receive user portrait tags through the user end;

[0150] Based on the book title set and the user portrait label, multiple algorithms are executed in parallel to recommend books to obtain books recommended by users’ short-term interests, books recommended by users’ long-term interests, books recommended by hot topics, and books recommended by balanced content;

[0151] According to the user portrait tag, multi-algorithm user feedback information is collected, and algorithm weight analysis is performed to obtain user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights;

[0152] According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain a book fusion recommendation value;

[0153] According to the book fusion recommendation value, the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books are serialized and adjusted from large to small to obtain a book ranking result;

[0154] The book title set is sent to a visualization interface of a user terminal for display according to the book sorting result.

[0155] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0156] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0158] These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0161] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0162] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. A book recommendation method based on a hybrid recommendation algorithm, characterized in that: include: Obtain a collection of book titles through the library database; Receive user portrait tags through the user end; Based on the book title set and the user portrait label, multiple algorithms are executed in parallel to recommend books to obtain books recommended by users’ short-term interests, books recommended by users’ long-term interests, books recommended by hot topics, and books recommended by balanced content; According to the user portrait tag, multi-algorithm user feedback information is collected, and algorithm weight analysis is performed to obtain user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights; According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain a book fusion recommendation value; According to the book fusion recommendation value, the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books are serialized and adjusted from large to small to obtain a book ranking result; Sending the book title set to a visual interface of a user terminal for displaying the book ranking result; The step of recommending books based on the user's short-term interests or books based on the user's long-term interests includes: Obtaining a sample collection time zone, wherein when recommending a user's short-term interests, the sample collection time zone is the past six months, and when recommending a user's long-term interests, the sample collection time zone is the past two years; According to the user portrait label and the sample collection time zone, a first book title borrowing frequency set and a first book title borrowing duration set of the book title set by users with similar portraits are collected; Normalizing the first book title borrowing frequency set to obtain a first book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a first weight; Normalizing the first book title borrowing time set to obtain a first book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a second weight; According to the first weight and the second weight, the first book title borrowing frequency normalization parameter set and the first book title borrowing duration normalization parameter set are added to obtain a first book recommendation value of the book title set; Sort the first book recommendation values ​​from large to small, select a preset number of books with head numbers, and obtain the user's short-term interest recommended books or the user's long-term interest recommended books; According to the user portrait tag, a book title borrowing frequency set and a book title borrowing duration set of the book title set by users with similar portraits are collected, including: The user portrait tag includes age, education, gender, and occupation, wherein the age has an age matching weight, the education has an education matching weight, the gender has a gender matching weight, and the occupation has an occupation matching weight, and the sum of the age matching weight, the education matching weight, the gender matching weight, and the occupation matching weight is equal to 1; According to the age, the education level, the gender, and the occupation, a portrait similarity evaluation model is constructed in combination with the age matching weight, the education level matching weight, the gender matching weight, and the occupation matching weight, wherein the portrait similarity evaluation model is constructed based on a portrait similarity evaluation function; According to the portrait similarity evaluation model, extracting users whose portrait similarity coefficient is greater than or equal to a portrait similarity coefficient threshold as the similar portrait users; Among them, the portrait similarity evaluation function is: in, Characterizes the image similarity coefficient, Characterize the first user portrait label and weight parameter, Characterize the second user portrait label and weight parameter, Representing age matching weights, Represents the academic qualification matching weight, Represents the gender matching weight, Characterizes the occupation matching weight, Characterizes the age of the first user, Characterizes the first user's education level, Characterizes the gender of the first user, Characterize the first user occupation, Characterize the second user's age, Represents the second user's academic qualifications, Characterizes the second user's gender, Characterize the second user occupation; According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, the weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain the book fusion recommendation value, including: Obtain a first recommended book according to the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books, wherein the first recommended book has a user short-term interest recommended serial number tag and / or a user long-term interest recommended serial number tag and / or a hot spot recommended serial number tag and / or a content balanced recommended serial number tag; The user short-term interest recommendation weight and the inverse of the user short-term interest recommendation sequence number label are multiplied and fused to obtain a first fusion parameter; The user's long-term interest recommendation weight and the inverse of the user's long-term interest recommendation sequence number label are multiplied and fused to obtain a second fusion parameter; The hotspot recommendation weight is multiplied by the inverse of the hotspot recommendation sequence number label to obtain a third fusion parameter; The content balanced recommendation weight and the inverse of the content balanced recommendation sequence number label are multiplied and fused to obtain a fourth fusion parameter; The first fusion parameter, the second fusion parameter, the third fusion parameter and the fourth fusion parameter are added to obtain a first recommended book fusion recommendation value, which is added to the book fusion recommendation value.

2. The method according to claim 1, characterized in that The steps of recommending hot recommended books include: Collecting borrowing record data of the set of book titles in a preset time zone; According to the borrowing record data of the preset time zone, the book title set is traversed and counted to obtain a second book title borrowing frequency set and a second book title borrowing duration set; Normalizing the second book title borrowing frequency set to obtain a second book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a third weight; Normalizing the second book title borrowing time set to obtain a second book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a fourth weight; According to the third weight and the fourth weight, the second book title borrowing frequency normalization parameter set and the second book title borrowing duration normalization parameter set are added to obtain a second book recommendation value of the book title set; According to the second book recommendation value, the books are sorted from large to small, and a preset number of books with head serial numbers are selected to obtain the hot recommended books.

3. The method according to claim 1, characterized in that The steps of recommending books with balanced content include: According to the book categories, cluster analysis is performed on the book title set to obtain multiple clusters of book titles; Traversing the multiple clusters of book titles and performing hot spot recommendation analysis respectively to obtain multiple groups of sub-hot spot recommended books, wherein the multiple groups of sub-hot spot recommended books have multiple groups of third recommendation values; Divide the preset number of recommended books based on the number of clusters to obtain the number of recommendations for a single cluster; According to the number of single cluster recommendations, traverse the multiple groups of sub-hotspot recommended books and sort them according to the multiple groups of third recommendation values ​​from large to small to obtain multiple clusters of selected recommended books, wherein the multiple clusters of selected recommended books have multiple clusters of selected third recommendation values; According to the multi-cluster selected third recommendation values, the multi-cluster selected recommended books are serially adjusted from large to small to obtain the content-balanced recommended books.

4. The method according to claim 1, characterized in that According to the user portrait tag, multi-algorithm user feedback information is collected, and algorithm weight analysis is performed to obtain user short-term interest recommendation weights, user long-term interest recommendation weights, hot spot recommendation weights, and content balance recommendation weights, including: Obtaining first user feedback information of a user short-term interest recommendation algorithm, wherein the first user feedback information includes a plurality of first user borrowing status information, wherein any one of the first user borrowing status information is a borrowing status or a non-borrowing status; Counting the borrowing status proportions of the borrowing status information of the plurality of first users, and setting the proportions as weight factors of the user short-term interest recommendation algorithm; Traverse the user long-term interest recommendation algorithm, hot spot recommendation algorithm and content balanced recommendation algorithm to obtain the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balanced recommendation algorithm; Adding the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balance recommendation algorithm to obtain a sum of the weight factors; Calculate the weight factor of the user short-term interest recommendation algorithm, the weight factor of the user long-term interest recommendation algorithm, the weight factor of the hot spot recommendation algorithm and the weight factor of the content balance recommendation algorithm respectively, and ratio them to the sum of the weight factors to obtain the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balance recommendation weight.

5. A book recommendation system based on a hybrid recommendation algorithm, characterized in that: include: The book title acquisition module is used to obtain a collection of book titles through the library database; A user portrait acquisition module is used to receive user portrait tags through the user end; A multi-algorithm recommendation module, for executing multi-algorithm parallel recommendation based on the book title set and the user portrait tag, to obtain books recommended by the user's short-term interests, books recommended by the user's long-term interests, hot-spot recommended books and content-balanced recommended books; An algorithm weight analysis module is used to collect multi-algorithm user feedback information according to the user portrait label, perform algorithm weight analysis, and obtain user short-term interest recommendation weight, user long-term interest recommendation weight, hot spot recommendation weight and content balance recommendation weight; A weight frequency evaluation module is used to perform weight frequency evaluation on the books recommended by the user's short-term interests, the books recommended by the user's long-term interests, the books recommended by the hot spots, and the books recommended by the content balance according to the weight of the recommendation, the weight of the recommendation, and the weight of the content balance recommendation, so as to obtain a book fusion recommendation value; A book sorting module is used to perform serial adjustment on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books from large to small according to the book fusion recommendation value to obtain a book sorting result; A book ranking recommendation module, used for sending the book title set to a user terminal visual interface for displaying the book ranking result; The step of recommending books based on the user's short-term interests or books based on the user's long-term interests includes: Obtaining a sample collection time zone, wherein when recommending a user's short-term interests, the sample collection time zone is the past six months, and when recommending a user's long-term interests, the sample collection time zone is the past two years; According to the user portrait label and the sample collection time zone, a first book title borrowing frequency set and a first book title borrowing duration set of the book title set by users with similar portraits are collected; Normalizing the first book title borrowing frequency set to obtain a first book title borrowing frequency normalization parameter set, wherein the book title borrowing frequency normalization parameter has a first weight; Normalizing the first book title borrowing time set to obtain a first book title borrowing time normalization parameter set, wherein the book title borrowing time normalization parameter has a second weight; According to the first weight and the second weight, the first book title borrowing frequency normalization parameter set and the first book title borrowing duration normalization parameter set are added to obtain a first book recommendation value of the book title set; Sort the first book recommendation values ​​from large to small, select a preset number of books with head numbers, and obtain the user's short-term interest recommended books or the user's long-term interest recommended books; According to the user portrait tag, a book title borrowing frequency set and a book title borrowing duration set of the book title set by users with similar portraits are collected, including: The user portrait tag includes age, education, gender, and occupation, wherein the age has an age matching weight, the education has an education matching weight, the gender has a gender matching weight, and the occupation has an occupation matching weight, and the sum of the age matching weight, the education matching weight, the gender matching weight, and the occupation matching weight is equal to 1; According to the age, the education level, the gender, and the occupation, a portrait similarity evaluation model is constructed in combination with the age matching weight, the education level matching weight, the gender matching weight, and the occupation matching weight, wherein the portrait similarity evaluation model is constructed based on a portrait similarity evaluation function; According to the portrait similarity evaluation model, extracting users whose portrait similarity coefficient is greater than or equal to a portrait similarity coefficient threshold as the similar portrait users; Among them, the portrait similarity evaluation function is: in, Characterizes the image similarity coefficient, Characterize the first user portrait label and weight parameter, Characterize the second user portrait label and weight parameter, Representing age matching weights, Represents the academic qualification matching weight, Represents the gender matching weight, Characterizes the occupation matching weight, Characterizes the age of the first user, Characterizes the first user's education level, Characterizes the gender of the first user, Characterize the first user occupation, Characterize the second user's age, Represents the second user's academic qualifications, Characterizes the second user's gender, Characterize the second user occupation; According to the user short-term interest recommendation weight, the user long-term interest recommendation weight, the hot spot recommendation weight and the content balanced recommendation weight, the weight frequency evaluation is performed on the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books to obtain the book fusion recommendation value, including: Obtain a first recommended book according to the user short-term interest recommended books, the user long-term interest recommended books, the hot spot recommended books and the content balanced recommended books, wherein the first recommended book has a user short-term interest recommended serial number tag and / or a user long-term interest recommended serial number tag and / or a hot spot recommended serial number tag and / or a content balanced recommended serial number tag; The user short-term interest recommendation weight and the inverse of the user short-term interest recommendation sequence number label are multiplied and fused to obtain a first fusion parameter; The user's long-term interest recommendation weight and the inverse of the user's long-term interest recommendation sequence number label are multiplied and fused to obtain a second fusion parameter; The hotspot recommendation weight is multiplied by the inverse of the hotspot recommendation sequence number label to obtain a third fusion parameter; The content balanced recommendation weight and the inverse of the content balanced recommendation sequence number label are multiplied and fused to obtain a fourth fusion parameter; The first fusion parameter, the second fusion parameter, the third fusion parameter and the fourth fusion parameter are added to obtain a first recommended book fusion recommendation value, which is added to the book fusion recommendation value.

6. A non-transitory computer-readable storage medium, characterized in that: The storage medium stores a computer software program, which, when executed by a processor, implements a book recommendation method based on a hybrid recommendation algorithm as described in any one of claims 1 to 4.

Images