Personalized book recommendation method and system, electronic equipment and storage medium

Abstract

The invention discloses a personalized book recommendation method and system, electronic equipment and a storage medium. The method comprises the following steps: calculating a first user preference degree of each book in historically read books; classifying the historical reading books according to the keyword similarity; performing initial classification on all books according to a preset category, and selecting a target book from each initial category as an initial cluster head book; constructing a fitness function based on the similarity between the keyword corresponding to the initial cluster head book and the keyword corresponding to each of other books; determining an optimal cluster head book based on the fitness function, and performing secondary classification on all books according to the optimal cluster head book; calculating a second user preference degree of each type of books in the first classification result according to the first user preference degree of each book; sorting each class of books in the second classification result based on the second user preference degree; and according to a sorting result, determining a recommended book. The book recommendation accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of personalized recommendation, and particularly to a personalized book recommendation method, system, electronic device and storage medium. Background Art

[0002] The Internet platform provides a large number of e-books for users, but not all books meet the preferences of each user. Therefore, book recommendation has gradually become a popular research direction in the field of book services. Existing recommendation algorithms include collaborative filtering algorithms, various deep learning algorithms, and content-based recommendation algorithms, etc.

[0003] These above-mentioned recommendation algorithms still have certain defects. For example, the collaborative filtering algorithm discovers users' preferences by mining users' historical behavior data, divides users into groups based on different preferences, and recommends products with similar tastes. However, the collaborative filtering algorithm requires a large number of user and item sample quantities. However, for new users or users who read fewer books, there is data sparsity, which will lead to poor recommendation effects. And the above-mentioned recommendation algorithms train a model by using a large number of sample data, and then predict the recommendation results through the trained model, there is a problem that the recommendation results cannot be explained.

[0004] Therefore, there is an urgent need for a technical solution that can solve the problems of poor recommendation effects and unexplainability existing in the existing recommendation algorithms. Summary of the Invention

[0005] The present application aims to propose a personalized book recommendation method, system, electronic device and storage medium, which can improve the accuracy of book recommendation and make the recommendation results explainable.

[0006] In a first aspect, an embodiment of the present application provides a personalized book recommendation method, and the method includes: Obtain all books and historical reading books in the book database; Calculate the first user preference degree of each book in the historical reading books; Based on the keywords extracted from each book in the historical reading books, classify the historical reading books according to the keyword similarity to obtain a first classification result; Perform an initial classification on all books according to a preset category to obtain a plurality of initial categories, and select a target book from each initial category as an initial cluster head book; Extract the keywords of each book in all books. For each initial category, construct a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determine the optimal cluster head book based on the fitness function, and perform a secondary classification on all books according to the optimal cluster head book to obtain a second classification result; Calculate the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book. Based on the second user preference degree, sort each category of books in the second classification result to obtain a sorting result. Determine the recommended books according to the sorting result.

[0007] Compared with the prior art, the first aspect of the present application has the following beneficial effects: This method obtains all books and historical reading books in the book database; calculates the first user preference degree of each book in the historical reading books; classifies the historical reading books according to the keyword similarity based on the keywords of each book in the extracted historical reading books to obtain a first classification result; initially classifies all books according to preset categories to obtain multiple initial categories, and selects a target book as an initial cluster head book from each initial category; extracts the keywords of each book in all books, and for each initial category, constructs a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determines the optimal cluster head book based on the fitness function, and performs secondary classification on all books according to the optimal cluster head book to obtain a second classification result; calculates the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book; sorts each category of books in the second classification result based on the second user preference degree to obtain a sorting result; determines the recommended books according to the sorting result. In this way, book recommendation based on user preference can better meet personalized needs and improve the accuracy of book recommendation. By determining the optimal cluster head book based on the fitness function and performing secondary classification on all books according to the optimal cluster head book to obtain a more accurate book classification. Then, based on the second user preference degree, sort each category of books in the second classification result, and then determine the recommended books according to the sorting result, which can improve the book recommendation effect and make the recommendation result interpretable.

[0008] In some embodiments, the calculating the first user preference degree of each book in the historical reading books includes: Divide the historical reading books into books that have been read but not added to the bookshelf, books that are not purchased or downloaded, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded according to user behavior; Calculate the first user preference degree of each book in the historical reading books according to the books that have been read but not added to the bookshelf, the books that are not purchased or downloaded, the books that are not purchased or downloaded but added to the bookshelf, and the books that are purchased or purchased and downloaded.

[0009] In some embodiments, based on the keywords extracted from each of the historical reading books, classifying the historical reading books according to keyword similarity to obtain a first classification result, including: Obtain the first title and the first introduction of each historical reading book, and extract the keywords in the first title and the first introduction; Convert the keywords in the first title into a first vector, and convert the keywords in the first introduction into a second vector; Calculate a first keyword similarity based on the first vector, and calculate a second keyword similarity based on the second vector; Calculate the overall similarity of the first keyword similarity and the second keyword similarity, and classify the historical reading books based on the overall similarity to obtain a first classification result.

[0010] In some embodiments, for each book among all the books, extract the keywords of each book. For each initial category, construct a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determine the optimal cluster head book based on the fitness function, and perform secondary classification on all the books according to the optimal cluster head book to obtain a second classification result, including: Obtain the second title and the second introduction of each book among all the books, and extract the keywords in the second title and the second introduction; Convert the keywords in the second title into a third vector, and convert the keywords in the second introduction into a fourth vector; For each initial category, calculate a third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each other book, and calculate a fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each other book; Construct a fitness function based on the third keyword similarity and the fourth keyword similarity; Based on the fitness function and the multiple initial categories, determine the optimal cluster head book corresponding to each initial category; Calculate the total similarity between the optimal cluster head book corresponding to each initial category and each book among all the books, and perform secondary classification on all the books according to the total similarity to obtain a second classification result.

[0011] In some embodiments, calculating the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each of the other books, and calculating the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each of the other books includes: Calculating the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each of the other books is: ; Calculating the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each of the other books is: ; Wherein, represents the third keyword similarity, represents the fourth keyword similarity, represents the third vector corresponding to the initial cluster head book the dimensional vector in represents the third vector corresponding to the a-th book the dimensional vector in represents the cluster space, represents the total number of dimensions in the third vector, represents the fourth vector corresponding to the initial cluster head book the dimensional vector in represents the fourth vector corresponding to the a-th book the dimensional vector in represents the total number of dimensions in the fourth vector.

[0012] In some embodiments, the classifying each class of books in the second classification result based on the second user preference degree to obtain a sorting result includes: Calculating the average preference evaluation score of each user for each book among all books; Calculating the total preference evaluation score of all users for each class of books in the second classification result according to the average preference evaluation score; Sorting each class of books in the second classification result according to the second user preference degree and the total preference evaluation score to obtain a sorting result.

[0013] In some embodiments, determining the recommended books according to the sorting result includes: Eliminating the books that the users have already read in each class of books in the sorting result to obtain the remaining sorted books; Sort all the books in each category of the remaining sorted books to obtain the sorted books for each category. Based on the remaining sorted books and the sorted books for each category, select multiple top-ranked book categories and multiple top-ranked books in each of the multiple top-ranked book categories to determine the recommended books.

[0014] In a second aspect, an embodiment of the present application further provides a personalized book recommendation system, which includes: A book acquisition unit for acquiring all the books and the historically read books in the book database. A first calculation unit for calculating the first user preference degree for each book in the historically read books. A book classification unit for classifying the historically read books according to the keyword similarity based on the keywords extracted from each book in the historically read books to obtain a first classification result. An initial classification unit for initially classifying all the books according to a preset category to obtain multiple initial categories, and selecting a target book from each initial category as an initial cluster head book. A secondary classification unit for extracting the keywords of each book in all the books. For each initial category, construct a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determine the optimal cluster head book based on the fitness function, and perform secondary classification on all the books according to the optimal cluster head book to obtain a second classification result. A second calculation unit for calculating the second user preference degree for each category of books in the first classification result according to the first user preference degree of each book. A category sorting unit for sorting each category of books in the second classification result based on the second user preference degree to obtain a sorting result. A book recommendation unit for determining the recommended books according to the sorting result.

[0015] In a third aspect, an embodiment of the present application further provides an electronic device, including at least one control processor and a memory for communicatively connecting with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute a personalized book recommendation method as described above.

[0016] Fourthly, an embodiment of the present application further provides a computer-readable storage medium storing computer-executable instructions for causing a computer to execute a personalized book recommendation method as described above.

[0017] It can be understood that the beneficial effects of the above second to fourth aspects compared with the related art are the same as those of the first aspect compared with the related art. For the relevant descriptions, reference can be made to the relevant descriptions in the first aspect, and details will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the accompanying drawings, where: Figure 1 is a schematic flowchart of an embodiment of the personalized book recommendation method provided by the present application; Figure 2 is a schematic structural diagram of an embodiment of the personalized book recommendation system provided by the present application; Figure 3 is a schematic structural diagram of an embodiment of the electronic device provided by the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and should not be construed as a limitation of the present application.

[0020] In the description of the present application, if the first, second, etc. are described only for the purpose of distinguishing technical features, they should not be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features or implicitly indicating the sequence of the indicated technical features.

[0021] In the description of the present application, it should be understood that the orientation or positional relationship indicated by terms such as up and down is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present application.

[0022] In the description of the present application, it should be noted that unless otherwise clearly defined, terms such as setting, installing, and connecting should be understood in a broad sense. Those skilled in the art can reasonably determine the specific meanings of the above terms in the present application in combination with the specific content of the technical solution.

[0023] First, parse several nouns involved in this application: word2vec: It is a group of related models used to generate word vectors. These models are shallow and two-layer neural networks, which are trained to reconstruct linguistic word texts. The network is represented by words and needs to guess the input words at adjacent positions. Under the assumption of the bag-of-words model in word2vec, the order of words is not important. After training, the word2vec model can be used to map each word to a vector, which can be used to represent the relationship between words. This vector is the hidden layer of the neural network.

[0024] TF-IDF: It is a commonly used weighting technique for information retrieval and data mining. TF is the term frequency, and IDF is the inverse document frequency.

[0025] The Internet platform provides users with a large number of e-books, but not all books meet the preferences of each user. Therefore, book recommendation has gradually become a popular research direction in the field of book services. Existing recommendation algorithms include collaborative filtering algorithms, various deep learning algorithms, and content-based recommendation algorithms, etc.

[0026] The above-mentioned recommendation algorithms still have certain defects. For example, the collaborative filtering algorithm discovers users' preferences by mining users' historical behavior data, divides users into groups based on different preferences, and recommends products with similar tastes. However, the collaborative filtering algorithm requires a large number of user and item sample quantities. However, for new users or users who read fewer books, there is data sparsity, which will lead to poor recommendation effects. And the above-mentioned recommendation algorithms train models by using a large number of sample data, and then predict the recommendation results through the trained models, resulting in the problem that the recommendation results cannot be explained.

[0027] Therefore, there is an urgent need for a technical solution that can solve the problems of poor recommendation effects and unexplainability existing in the existing recommendation algorithms.

[0028] To solve the problems of poor recommendation effects and unexplainability existing in the existing recommendation algorithms, this application proposes a personalized book recommendation method, system, electronic device, and storage medium.

[0029] Refer to Figure 1 , an embodiment of this application provides a personalized book recommendation method, and this method includes the following steps: Step S100, obtain all books and historically read books in the book database; Step S200, calculate the first user preference degree of each book in the historically read books; Step S300: Classify the historical reading books according to the keyword similarity based on the keywords of each book in the extracted historical reading books, and obtain the first classification result; Step S400: Initially classify all books according to preset categories to obtain multiple initial categories, and select a target book from each initial category as the initial cluster head book; Step S500: Extract the keywords of each book in all books. For each initial category, construct a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determine the optimal cluster head book based on the fitness function, and perform secondary classification on all books according to the optimal cluster head book to obtain the second classification result; Step S600: Calculate the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book; Step S700: Sort the categories of books in the second classification result based on the second user preference degree to obtain a sorting result; Step S800: Determine the recommended books according to the sorting result.

[0030] In this embodiment, by obtaining all books and historical reading books in the book database; calculating the first user preference degree of each book in the historical reading books; classifying the historical reading books according to the keyword similarity based on the keywords of each book in the extracted historical reading books to obtain the first classification result; initially classifying all books according to preset categories to obtain multiple initial categories, and selecting a target book from each initial category as the initial cluster head book; extracting the keywords of each book in all books, and for each initial category, constructing a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; determining the optimal cluster head book based on the fitness function, and performing secondary classification on all books according to the optimal cluster head book to obtain the second classification result; calculating the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book; sorting the categories of books in the second classification result based on the second user preference degree to obtain a sorting result; and determining the recommended books according to the sorting result. In this way, recommending books according to the user preference degree can better meet the personalized needs and improve the accuracy of book recommendation. By determining the optimal cluster head book based on the fitness function and performing secondary classification on all books according to the optimal cluster head book to obtain a more accurate book classification. Then, based on the second user preference degree, sorting the categories of books in the second classification result, and then determining the recommended books according to the sorting result can improve the book recommendation effect and make the recommendation result interpretable.

[0031] The above calculation of the first user preference for each book in the historical reading books. Since different user behaviors can represent different levels of hobbies of users, the historical reading books of users can be divided into books that have only been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded. The user preference for each book in the historical reading books of users can be calculated based on the books that have been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded.

[0032] It should be noted that the historical reading books of users in this embodiment are obtained under the authorization of the users.

[0033] The above classification of historical reading books according to the keyword similarity to obtain the first classification result can be to classify the historical reading books according to the keyword similarity by using the traditional clustering method to obtain the first classification result. This embodiment does not make specific limitations.

[0034] The above preset categories can be classification categories set according to the actual situation and can be changed according to the actual situation. This embodiment does not make specific limitations.

[0035] In some embodiments, calculating the first user preference for each book in the historical reading books includes: Dividing the historical reading books into books that have been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded according to user behaviors; Calculating the first user preference for each book in the historical reading books based on the books that have been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded.

[0036] In this embodiment, calculating the user preference for each book in the historical reading books of users based on the books that have been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded, and comprehensively considering various factors to calculate the user preference can improve the accuracy of the user preference, so as to recommend more books that meet the user's preferences to the user in the later stage and improve the recommendation accuracy.

[0037] In some embodiments, based on the keywords extracted from each book in the historical reading books, classifying the historical reading books according to the keyword similarity to obtain the first classification result includes: Obtaining the first title and the first introduction of each historical reading book, and extracting the keywords in the first title and the first introduction; Convert the keywords in the first title into a first vector, and convert the keywords in the first introduction into a second vector; Calculate the first keyword similarity based on the first vector, and calculate the second keyword similarity based on the second vector; Calculate the overall similarity of the first keyword similarity and the second keyword similarity, and classify the historical reading books based on the overall similarity to obtain a first classification result.

[0038] In this embodiment, obtain the first title and the first introduction of each historical reading book, and extract the keywords in the first title and the first introduction; convert the keywords in the first title into a first vector, and convert the keywords in the first introduction into a second vector; calculate the first keyword similarity based on the first vector, and calculate the second keyword similarity based on the second vector; calculate the overall similarity of the first keyword similarity and the second keyword similarity, and classify the historical reading books based on the overall similarity to obtain a first classification result. In this way, by comprehensively considering the similarity of the title and the introduction to classify the user's historical reading books, the accuracy of classifying the user's historical reading books can be improved.

[0039] It should be noted that the introduction in this embodiment can be information such as an abstract or a summary for summarizing the content of the book, and this embodiment does not make specific limitations.

[0040] The above similarity calculation can use cosine similarity, or other existing technologies well-known to those skilled in the art to calculate similarity, and this embodiment does not make specific limitations.

[0041] In some embodiments, extract the keywords of each book among all books. For each initial category, construct a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determine the optimal cluster head book based on the fitness function, and perform secondary classification on all books according to the optimal cluster head book to obtain a second classification result, including: Obtain the second title and the second introduction of each book among all books, and extract the keywords in the second title and the second introduction; Convert the keywords in the second title into a third vector, and convert the keywords in the second introduction into a fourth vector; For each initial category, calculate the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each other book, and calculate the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each other book; Construct a fitness function according to the third keyword similarity and the fourth keyword similarity; Based on the fitness function and multiple initial categories, determine the optimal cluster head books corresponding to each initial category; Calculate the total similarity between the optimal cluster head books corresponding to each initial category and each book among all the books, and perform secondary classification on all the books according to the total similarity to obtain the second classification result.

[0042] In this embodiment, obtain the second title and second introduction of each book among all the books, and extract the keywords in the second title and second introduction; convert the keywords in the second title into a third vector, and convert the keywords in the second introduction into a fourth vector; for each initial category, calculate the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each other book, and calculate the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each other book; construct a fitness function based on the third keyword similarity and the fourth keyword similarity; based on the fitness function and multiple initial categories, determine the optimal cluster head books corresponding to each initial category; calculate the total similarity between the optimal cluster head books corresponding to each initial category and each book among all the books, and perform secondary classification on all the books according to the total similarity to obtain the second classification result. In this way, since the initial classification is first performed through the original preset categories (i.e., manual classification), this classification may not be very accurate, but it can provide a rough classification of all the books. Then, by comprehensively considering the similarity of the title and introduction to construct a fitness function and using the adaptive particle swarm algorithm to classify all the books, the accuracy of the classification of all the books can be improved.

[0043] In some embodiments, calculating the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each other book, and calculating the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each other book includes: The calculation of the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each other book is: ; The calculation of the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each other book is: ;

[0044] Wherein, represents the third keyword similarity, represents the fourth keyword similarity, represents the third vector corresponding to the initial cluster head book the dimensional vector in Denote the third vector corresponding to the ath book in the dimensional vector,[[]] denote the cluster space,[[]] denote the total number of dimensions in the third vector,[[]] denote the fourth vector corresponding to the initial cluster head book in the dimensional vector,[[]] denote the fourth vector corresponding to the ath book in the dimensional vector,[[]] denote the total number of dimensions in the fourth vector.[[]]

[0045] In some embodiments, based on the second user preference, perform category sorting on each category of books in the second classification result to obtain a sorting result, including: Calculate the average preference evaluation score of all users for each book among all books; According to the average preference evaluation score, calculate the total preference evaluation score of all users for each category of books in the second classification result; According to the second user preference and the total preference evaluation score, perform category sorting on each category of books in the second classification result to obtain a sorting result.

[0046] In this embodiment, calculate the average preference evaluation score of all users for each book among all books; according to the average preference evaluation score, calculate the total preference evaluation score of all users for each category of books in the second classification result; according to the second user preference and the total preference evaluation score, perform category sorting on each category of books in the second classification result to obtain a sorting result. Thus, the purpose of adding the total preference evaluation score of all users is to prevent some book categories from not belonging to the user's reading book categories, that is, when the user's reading category is relatively single, other books cannot be recommended to the user. Through the above operations, books that many users like can be recommended to the user, and books can be better recommended to the user.

[0047] In some embodiments, determine the recommended books according to the sorting result, including: Eliminate the books that the user has already read in each category of books in the sorting result to obtain the remaining sorted books; Sort all the books in each category of the remaining sorted books to obtain the sorted books for each category; According to the remaining sorted books and the sorted books for each category, select multiple top-ranked book categories and multiple top-ranked books in each of the multiple top-ranked book categories to determine the recommended books.

[0048] In this embodiment, the books that the user has already read in each category of the sorting result are excluded to obtain the remaining sorted books; all the books in each category of the remaining sorted books are sorted to obtain the sorted books for each category; based on the remaining sorted books and the sorted books for each category, multiple books categories with top rankings are selected and multiple books with top rankings in each of the multiple books categories with top rankings are selected to determine the recommended books. In this way, by excluding the books that the user has already read in each category of the sorting result to obtain the remaining sorted books and making recommendations based on the remaining sorted books, it is possible to prevent the customer from reading books they dislike or reading the same books repeatedly, improve the book recommendation effect, and make the recommendation result interpretable. And even if the user is a new user, it is possible to recommend the most popular and most liked books by other users to the new user based on the books with top rankings in each category, improving the book recommendation effect.

[0049] For the convenience of those skilled in the art to understand, a set of best embodiments are provided below: Step S1: Calculate the preference degree of each book.

[0050] Obtain the user's historical reading books on the platform (which can be a book database or a book reading platform) (classify books according to user behavior, including books that have only been read but not added to the bookshelf, books that are downloaded without purchase, books that are not purchased or downloaded but added to the bookshelf, books that are purchased or purchased and downloaded), count the number of times each book has been read and the time of each reading of each book; purchases and downloads will automatically be added to the bookshelf, only free books can be downloaded without purchase, and other books need to be purchased to download.

[0051] , x Belong to the books that have been read but not added to the bookshelf; , y belongs to the books that are downloaded without purchase; , z belongs to the books that are not purchased or downloaded but added to the bookshelf; , belong to the books that are purchased or purchased and downloaded; According to the weight, the number of readings, and the reading time, calculate the user preference degree (i.e., the first user preference degree) of each book as : ; Among them, , , and Represents a weight parameter, which can be changed according to actual situations and is not specifically limited in this embodiment. For example, if a purchased or purchased and downloaded book indicates that the user likes it very much, the weight of the purchased or purchased and downloaded book can be set higher. For a book that has only been read but not added to the bookshelf, it indicates that the user's interest is not very high and the weight is set lower; Represents the number of times a book that has been read but not added to the bookshelf has been read, Represents the reading time of a book that has been read but not added to the bookshelf, Represents the number of times a book that is not purchased and downloaded has been read, Represents the reading time of a book that is not purchased and downloaded, Represents the number of times a book that is not purchased, not downloaded but added to the bookshelf has been read, Represents the reading time of a book that is not purchased, not downloaded but added to the bookshelf, Represents the number of times a purchased or purchased and downloaded book has been read, Represents the reading time of a purchased or purchased and downloaded book, Represents the th reading time of the book, Represents the th reading time of the book, Represents the th reading time of the book, Represents the th reading time of the book.

[0052] Step S2: Classify the books read by the user.

[0053] Use clustering or other methods to classify the books read by the user historically to obtain a first classification result; use clustering or other methods to classify all books to obtain a second classification result.

[0054] The classification of the books read by the user historically specifically includes the following content: Obtain the title and introduction of each book read by the user historically, and extract the keywords in the title and introduction; Convert the keywords in the title into a first vector A. Convert the keywords in the introduction into a second vector B; The keywords can be converted into the first vector or the second vector by means of word2vec or TF-IDF, etc., and this embodiment is not specifically limited.

[0055] Calculate the first keyword similarity according to the first vector, and calculate the second keyword similarity according to the second vector; Classify the user's historical reading books according to the first keyword similarity and the second keyword similarity to obtain the first classification result. If the classification evaluation value is greater than the preset value, it is regarded as one category. This is the classification of the books read by a user.

[0056] Specifically: Calculate the first keyword similarity between the first vectors corresponding to every two books in the user's historical reading books according to the first vector of each book For: ; Calculate the second keyword similarity between the second vectors corresponding to every two books in the user's historical reading books according to the second vector of each book For: ; Among them, represents the th dimensional vector in the first vector corresponding to one of every two books represents the th dimensional vector in the first vector corresponding to the other of every two books represents the total number of dimensions in the first vector, represents the th dimensional vector in the second vector corresponding to one of every two books represents the th dimensional vector in the second vector corresponding to the other of every two books represents the total number of dimensions in the second vector.

[0057] Overall similarity calculation: ; Among them, represents the overall similarity.

[0058] Preset the similarity threshold, and classify the two books with the overall similarity greater than the similarity threshold into the same category.

[0059] Classify all the books in the book library in the following way: Since the number of books in the book library is too large, it is not accurate enough to use ordinary classification methods. Therefore, classify all the books in the following way: First, make a rough initial classification of all the books in the book library according to the original preset classification to obtain the initial categories; Obtain the title and introduction of each book, and extract the keywords in the title and introduction; Convert the keywords in the title into the third vector P, and convert the keywords in the abstract into the fourth vector Q. The keywords can be converted into the third vector or the fourth vector by means of word2vec or TF-IDF, etc. This embodiment does not make specific limitations.

[0060] Calculate the third keyword similarity based on the third vector, and calculate the fourth keyword similarity based on the second vector; Construct a fitness function based on the third keyword similarity and the fourth keyword similarity; Based on this fitness function and the initial categories, use the adaptive particle swarm algorithm to perform secondary classification on all books to obtain the second classification result. If the classification evaluation value is greater than the preset value, it is regarded as one category. This is the classification of all books in the book library.

[0061] Specifically: Adaptive particle swarm optimization (APSO) is a particle swarm optimization algorithm with better search efficiency than classical particle swarm optimization (PSO). It can perform global search on the entire search space at a faster convergence rate. Select the book with the longest reading time in each category of the initial categories as the initial cluster head of this category to obtain a group of initial cluster heads. Based on the adaptive function and the initial cluster heads, initialize the local optimal positions of the nodes (i.e., books) and the global optimal positions of all current nodes, that is, calculate the corresponding initial local optimal values and initial global optimal values for each category. Since the iteration weight affects the convergence of the adaptive particle swarm algorithm to a large extent, a larger weight helps the adaptive particle swarm algorithm break through the bondage of local optimality, while a relatively smaller weight is beneficial to accelerating the convergence speed of the adaptive particle swarm algorithm. The commonly used linear iteration weight is as follows: ; Where represents the iteration weight, represents the maximum weight value, represents the minimum weight value, represents the maximum number of iterations, represents the number of iterations. In this embodiment, the weights in the above formula are further adjusted with non-linear adaptive weights, and the adjustment function is as follows: ; ; Where represents the minimum value of the iteration weight, T represents the total number of iterations, It represents the ratio of the best position of particle i at the current iteration t to the best position of all particles during the global iteration T. It represents the best position of particle i at the t-th iteration. It represents the best position of all particles during all iteration processes. It can be inferred from the definition of the adaptive weight function that is greater than . Adaptive optimization of the adaptive particle swarm algorithm is achieved through adaptive weight adjustment.

[0062] In order to enable the adaptive particle swarm algorithm to reach the solution space of this problem, the calculation formula of the fitness function proposed is as follows: ; where and represent the weight coefficients, represents the fitness value.

[0063] According to the third vector of each book, calculate the similarity between each book in the cluster space and the book serving as the initial cluster head to obtain the third keyword similarity as: ; According to the fourth vector of each book, calculate the similarity between each book in the cluster space and the book serving as the initial cluster head to obtain the fourth keyword similarity as: ; where represents the vector of the -th dimension in the third vector corresponding to the initial cluster head book, represents the vector of the -th dimension in the third vector corresponding to the a-th book, represents the cluster space, represents the total number of dimensions in the third vector, represents the vector of the -th dimension in the fourth vector corresponding to the initial cluster head book, represents the -th vector of the dimension in the fourth vector corresponding to the -th book, and

[0064] Based on the above adaptive fitness function, initial local optimal value, and initial global optimal value, continuously perform iterative updates of the adaptive particle swarm algorithm. The iterative formula is as follows: ; ; where, represents the velocity of particle i at time t + 1, d represents the iteration weight, represents the velocity of particle i at time t, and represent the learning factors, and represent random numbers, represents the best position of particle i in the t-th iteration, represents the best position of all particles in all iteration processes, represents the position of particle i at time t + 1, represents the position of particle i at time t.

[0065] The above iteration formula is used to obtain the final global optimal value for each category; the books with the final global optimal value for each category are used as the optimal cluster heads for that category. All books are reclassified according to the optimal cluster heads for that category to obtain the second classification result. According to the overall similarity calculate the books that are relatively similar to each optimal cluster head, and the books that are closest to an optimal cluster head in similarity are used as the same category as that optimal cluster head. Threshold-based clustering can be used, or other methods can be used for clustering, and this embodiment does not make specific restrictions. The overall similarity is calculated as follows: ; where, represents the -dimensional vector in the third vector corresponding to the optimal cluster head, represents the -dimensional vector in the third vector corresponding to the i-th book, h represents the total number of dimensions in the third vector, represents the -dimensional vector in the fourth vector corresponding to the optimal cluster head, represents the -dimensional vector in the fourth vector corresponding to the i-th book, r represents the total number of dimensions in the fourth vector.

[0066] Each book is compared with each optimal cluster head to see which optimal cluster head it is closest to in similarity, and the book is assigned to the category corresponding to the optimal cluster head with the closest similarity.

[0067] Step S3: Sorting of book categories.

[0068] According to the user preference degree of each book and the first classification result, calculate the total user preference degree of each category of books (i.e., the second user preference degree); ; in, represents the user's preference for the e-th book, s represents the number of books in each category in the first classification result, Indicates the total user preference for each type of books.

[0069] Calculate the average preference evaluation score of all users for each book in all books; then calculate the total preference evaluation score of all users of each category of books in the second classification results corresponding to all books based on the average preference evaluation score; specifically: In the library, some users will rate the books they have read. Therefore, most books will have a certain score, and a small number of books will have a score of zero if they have no score. Based on these scores, the average preference evaluation score of all users for each book in all books can be calculated. The calculation process is to add up the scores and divide by the number of users who have rated. Then add up the average preference evaluation scores of each book in each category of all books to get the total preference evaluation score of all users in each category of books in the second classification results.

[0070] According to the total preference degree of the users and the total preference evaluation scores of all users, the books in each category in the second classification result are sorted to obtain the sorting result of each category of books.

[0071] The purpose of adding all users' preference evaluation scores is to prevent some book categories from not belonging to the user's reading book categories, that is, when the user's reading category is relatively single, other books cannot be given to the user. Through the above operation, many books that users like can be recommended to users.

[0072] Step S4: Books are rejected.

[0073] Eliminate the books you have read from the ranking results of each category of books to get the remaining ranked books. The purpose is to eliminate books you don’t like and books you have finished reading to improve the recommendation effect.

[0074] Step S5: book recommendation.

[0075] According to the calculated average preference evaluation score of all users of each book, all books in each category of the remaining sorted books are sorted; Based on the top-ranked books in each category among the remaining sorted books, books that the user may like in multiple categories are recommended to the user.

[0076] In the above manner, even if the user is a new user, the most popular books that are most liked by other users can be recommended to the new user based on the top-ranked books in each category.

[0077] Reference Figure 2, an embodiment of the present application further provides a personalized book recommendation system, which includes a book acquisition unit 100, a first calculation unit 200, a book classification unit 300, an initial classification unit 400, a secondary classification unit 500, a second calculation unit 600, a category sorting unit 700, and a book recommendation unit 800, where: The book acquisition unit 100 is used to acquire all the books in the book database and the historically read books; The first calculation unit 200 is used to calculate the first user preference degree of each book in the historically read books; The book classification unit 300 is used to classify the historically read books according to the keyword similarity based on the keywords extracted from each book in the historically read books, and obtain a first classification result; The initial classification unit 400 is used to initially classify all the books according to preset categories, obtain multiple initial categories, and select a target book from each initial category as an initial cluster head book; The secondary classification unit 500 is used to extract the keywords of each book in all the books. For each initial category, a fitness function is constructed based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and the optimal cluster head book is determined based on the fitness function, and all the books are secondarily classified according to the optimal cluster head book to obtain a second classification result; The second calculation unit 600 is used to calculate the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book; The category sorting unit 700 is used to sort each category of books in the second classification result based on the second user preference degree to obtain a sorting result; The book recommendation unit 800 is used to determine the recommended books according to the sorting result.

[0078] It should be noted that since a personalized book recommendation system in this embodiment and the above-mentioned personalized book recommendation method are based on the same inventive concept, the corresponding content in the method embodiment is equally applicable to the system embodiment of the present application, and will not be elaborated here.

[0079] Refer to Figure 3 , an embodiment of the present application further provides an electronic device, and this electronic device includes: At least one memory; At least one processor; At least one program; The program is stored in the memory, and the processor executes at least one program to implement the above-mentioned personalized book recommendation method of the present disclosure.

[0080] The electronic device can be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), an in-vehicle computer, etc.

[0081] The following provides a detailed introduction to the electronic device according to the embodiments of the present application.

[0082] The processor 1600 can be implemented by using a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present disclosure; The memory 1700 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 1700 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 1700, and are called by the processor 1600 to execute the personalized book recommendation method of the embodiments of the present disclosure.

[0083] The input / output interface 1800 is used to implement information input and output; The communication interface 1900 is used to implement communication interaction between this device and other devices, and can implement communication through a wired method (such as USB, network cable, etc.), or can also implement communication through a wireless method (such as a mobile network, WIFI, Bluetooth, etc.); The bus 2000 transmits information between various components of the device (such as the processor 1600, the memory 1700, the input / output interface 1800, and the communication interface 1900); Among them, the processor 1600, the memory 1700, the input / output interface 1800, and the communication interface 1900 are communicatively connected to each other inside the device through the bus 2000.

[0084] The embodiments of the present disclosure also provide a storage medium, which is a computer-readable storage medium, and the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute the above-mentioned personalized book recommendation method.

[0085] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes a memory remotely disposed relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0086] The embodiments described in the embodiments of the present disclosure are for more clearly illustrating the technical solutions of the embodiments of the present disclosure, and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. Those skilled in the art will know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present disclosure are equally applicable to similar technical problems.

[0087] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present disclosure, and may include more or fewer steps than shown in the figures, or combine certain steps, or different steps.

[0088] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0089] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices can be implemented as software, firmware, hardware, and appropriate combinations thereof.

[0090] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of this application and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0091] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" may mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (one)" or similar expressions thereof refer to any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c may mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0092] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0093] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0094] In addition, the functional units in each embodiment of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0095] When an integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes multiple instructions for causing an electronic device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs. The embodiments of the present application have been described in detail above with reference to the accompanying drawings. However, the present application is not limited to the above embodiments, and various changes can be made without departing from the gist of the present application within the knowledge scope of those of ordinary skill in the art.

[0096] The embodiments of the present application have been described in detail above with reference to the accompanying drawings. However, the present application is not limited to the above embodiments, and various changes can be made without departing from the gist of the present application within the knowledge scope of those of ordinary skill in the art.

Claims

1. A personalized book recommendation method, characterized in that, The method includes: Obtaining all books in the book database and the historically read books; Calculating the first user preference degree for each book in the historically read books; Based on the keywords extracted from each book in the historically read books, classifying the historically read books according to keyword similarity to obtain a first classification result; Performing an initial classification on all books according to preset categories to obtain multiple initial categories, and selecting a target book from each initial category as an initial cluster head book; Extracting the keywords of each book in all books, and for each initial category, constructing a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and determining an optimal cluster head book based on the fitness function, and performing a secondary classification on all books according to the optimal cluster head book to obtain a second classification result; Calculating the second user preference degree for each category of books in the first classification result according to the first user preference degree of each book; Based on the second user preference degree, sorting the categories of books in the second classification result to obtain a sorting result; Determining the recommended books according to the sorting result.

2. The personalized book recommendation method according to claim 1, wherein The calculating the first user preference degree for each book in the historically read books includes: Dividing the historically read books into books that have been read but not added to the bookshelf, books that are not purchased and downloaded, books that are not purchased and not downloaded but added to the bookshelf, and books that are purchased or purchased and downloaded according to user behavior; Calculating the first user preference degree for each book in the historically read books according to the books that have been read but not added to the bookshelf, the books that are not purchased and downloaded, the books that are not purchased and not downloaded but added to the bookshelf, and the books that are purchased or purchased and downloaded.

3. The personalized book recommendation method according to claim 1, wherein The classifying the historically read books according to keyword similarity based on the keywords extracted from each book in the historically read books to obtain a first classification result includes: Obtaining the first title and the first introduction of each historically read book, and extracting the keywords in the first title and the first introduction; Converting the keywords in the first title into a first vector, and converting the keywords in the first introduction into a second vector; Calculating a first keyword similarity according to the first vector, and calculating a second keyword similarity according to the second vector; Calculating the overall similarity of the first keyword similarity and the second keyword similarity, and classifying the historically read books based on the overall similarity to obtain a first classification result.

4. The personalized book recommendation method according to claim 1, characterized in that The extracting the keywords of each book in all books, and for each initial category, constructing a fitness function based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; And determining an optimal cluster head book based on the fitness function, and performing a secondary classification on all books according to the optimal cluster head book to obtain a second classification result includes: Obtaining the second title and the second introduction of each book in all books, and extracting the keywords in the second title and the second introduction; Convert the keywords in the second title into a third vector, and convert the keywords in the second introduction into a fourth vector; For each initial category, calculate the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each of the other books, and calculate the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each of the other books; Construct a fitness function based on the third keyword similarity and the fourth keyword similarity; Based on the fitness function and the multiple initial categories, determine the optimal cluster head book corresponding to each initial category; Calculate the total similarity between the optimal cluster head book corresponding to each initial category and each book among all the books, and perform secondary classification on all the books according to the total similarity to obtain a second classification result.

5. The personalized book recommendation method according to claim 4, characterized in that The calculating the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each of the other books, and calculating the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each of the other books includes: The calculation of the third keyword similarity between the third vector corresponding to the initial cluster head book and the third vectors corresponding to each of the other books is: ； The calculation of the fourth keyword similarity between the fourth vector corresponding to the initial cluster head book and the fourth vectors corresponding to each of the other books is: ； Among them, represents the third keyword similarity, represents the fourth keyword similarity, represents the third vector corresponding to the initial cluster head book in the dimensional vector, represents the third vector corresponding to the a-th book in the dimensional vector, represents the cluster space, represents the total number of dimensions in the third vector, represents the fourth vector corresponding to the initial cluster head book in the dimensional vector, represents the fourth vector corresponding to the a-th book in the dimensional vector, represents the total number of dimensions in the fourth vector.

6. The personalized book recommendation method according to claim 1, wherein The sorting the categories of each type of book in the second classification result based on the second user preference degree to obtain a sorting result includes: Calculate the average preference evaluation score of all users for each book among all the books; According to the average preference evaluation score, calculate the total preference evaluation score of all users for each type of book in the second classification result; According to the second user preference degree and the total preference evaluation score, sort the categories of each type of book in the second classification result to obtain a sorting result.

7. The personalized book recommendation method according to claim 1, characterized in that The determining the recommended books according to the sorting result includes: Eliminate the books that users have already read from each type of book in the sorting result to obtain the remaining sorted books; Sort all the books in each category of the remaining sorted books to obtain the sorted books for each category; According to the remaining sorted books and the sorted books for each category, select multiple book categories with higher rankings and select multiple books with higher rankings in each of the multiple book categories with higher rankings to determine the recommended books.

8. A personalized book recommendation system, characterized in that, The system includes: A book acquisition unit for acquiring all the books and historical reading books in the book database; A first calculation unit for calculating the first user preference degree of each book in the historical reading books; A book classification unit for classifying the historical reading books according to the keyword similarity based on the keywords extracted from each book in the historical reading books to obtain a first classification result; An initial classification unit for initially classifying all the books according to a preset category to obtain multiple initial categories, and selecting a target book from each initial category as an initial cluster head book; The secondary classification unit is used to extract keywords of each book among all books. For each initial category, a fitness function is constructed based on the similarity between the keywords corresponding to the initial cluster head book and the keywords corresponding to each other book; and the optimal cluster head book is determined based on the fitness function, and all books are secondarily classified according to the optimal cluster head book to obtain a second classification result; The second calculation unit is used to calculate the second user preference degree of each category of books in the first classification result according to the first user preference degree of each book; The category sorting unit is used to sort each category of books in the second classification result based on the second user preference degree to obtain a sorting result; The book recommendation unit is used to determine the recommended books according to the sorting result.

9. An electronic device, characterized in that, It includes at least one control processor and a memory for communicatively connecting with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the personalized book recommendation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to execute the personalized book recommendation method according to any one of claims 1 to 7.

Images