Library digital resource-oriented user information analysis system

Abstract

The invention provides a library digital resource-oriented user information analysis system, which relates to the field of intelligent library management and is characterized in that data preprocessing and semantic analysis are carried out on borrowing historical records, collection records and search records of a user in a digital library so as to obtain style features of themes which the user is interested in reading; meanwhile, semantic analysis is conducted on text data of a to-be-recommended book to obtain content semantic features of the to-be-recommended book, and then the relation between the style features of the user reading interested theme and the content semantic features of the to-be-recommended book is established in the same feature space; and determining whether to put the to-be-recommended book into the book recommendation list or not. Therefore, the accurate book recommendation service can be provided for the user, and the satisfaction degree of the user is improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent library management, and more specifically, to a user information analysis system for library digital resources. Background Art

[0002] Library digital resources refer to various information resources that are acquired, organized, stored and provided with services by the library through digital technology. It allows readers to access the library at any time without being restricted by the library's opening hours and geographical location as long as there is an Internet connection, greatly improving the convenience of obtaining resources.

[0003] In digital libraries, books recommended to readers are often based on their popularity, without fully considering whether the book meets the readers' reading needs. Or, the reading needs of readers are considered but often lack depth, resulting in the recommended books not being adequately matched to the readers' reading preferences. This, to a certain extent, limits the further improvement of the service quality of digital libraries.

[0004] Therefore, an optimized user information analysis solution for library digital resources is needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, this application provides a user information analysis system for library digital resources, which obtains the theme style features of the user's reading interest by performing data preprocessing and semantic analysis on the user's borrowing history records, collection records, and search records in the digital library, and obtains the content semantic features of the books to be recommended by performing semantic analysis on the text data of the recommended books, and then establishes a connection between the theme style features of the user's reading interest and the content semantic features of the books to be recommended in the same feature space, and then decides whether to put the books to be recommended into the book recommendation list. In this way, it can provide users with accurate book recommendation services and improve user satisfaction.

[0006] According to one aspect of the present application, a user information analysis system for library digital resources is provided, which includes: The user information and book information acquisition module is used to acquire the user's borrowing history, collection record and search record in the digital library and the text data of the books to be recommended; A user information and book information encoding module, for encoding the borrowing history, collection record and search record of the user in the digital library and the text data of the book to be recommended to obtain a theme style feature vector of the user's reading interest and a content semantic feature vector of the book to be recommended; A user information feature book information feature matching module is used to perform heterogeneous significant feature fusion based on attribute space adaptation on the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to obtain a user reading interest-book content semantic matching feature vector; The book recommendation result generating module is used to determine whether to add the to-be-recommended book to the user's book recommendation list based on the user's reading interest-book content semantic matching feature vector.

[0007] Furthermore, the user information book information encoding module includes: A user information feature extraction unit, used for extracting features from the borrowing history records, collection records and search records of the user in the digital library to obtain a feature vector of the user's reading interest subject style; The book information feature extraction unit is used to extract content features from the text data of the book to be recommended to obtain a content semantic feature vector of the book to be recommended.

[0008] Furthermore, the user information feature extraction unit includes: A user information preprocessing subunit, used for preprocessing the borrowing history records, collection records and search records of the user in the digital library to obtain preprocessed user reading portrait data; A user information related word embedding subunit, used to segment the pre-processed user reading portrait data and pass it through a word embedding layer to obtain a sequence of user reading portrait keyword embedding vectors; The user information related word semantic encoding subunit is used to pass the sequence of the user reading portrait keyword embedding vector through the user reading portrait data semantic encoder to obtain the user reading interest topic style feature vector.

[0009] Furthermore, the book information feature extraction unit is used to: segment the text data of the book to be recommended and then pass it through a book text data semantic encoder to obtain the content semantic feature vector of the book to be recommended.

[0010] Furthermore, the user reading portrait data semantic encoder is a bidirectional long short-term memory neural network model, and the book text data semantic encoder is a converter-based encoder including a word embedding layer.

[0011] Furthermore, the user information feature and book information feature matching module is used to: Constructing a user reading interest-book content distance topological matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; Constructing a user reading interest-book content feature value granularity association matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; Based on the user reading interest-book content distance topological matrix, the user reading interest-book content feature value granularity association matrix is ​​topologically modulated to obtain a user reading interest-book content topological modulation association matrix; Taking the user reading interest-book content topological modulation association matrix as the perspective co-projection space, projecting the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector into the perspective co-projection space to obtain the perspective modulation user reading interest theme style feature vector and the perspective modulation to-be-recommended book content semantic feature vector; The perspective-modulated user reading interest theme style feature vector and the perspective-modulated to-be-recommended book content semantic feature vector are fused to obtain the user reading interest-book content semantic matching feature vector.

[0012] Furthermore, the book recommendation result generation module is used to: pass the user's reading interest-book content semantic matching feature vector through a book addition list classifier to obtain a book addition classification result, and the book addition classification result is used to indicate whether the to-be-recommended book is added to the user's book recommendation list.

[0013] Furthermore, the book recommendation result generating module is used to: Use the fully connected layer of the book addition list classifier to perform fully connected encoding on the user reading interest-book content semantic matching feature vector to obtain the user reading interest-book content semantic matching encoding classification feature vector; The user reading interest-book content semantic matching encoded classification feature vector is input into the Softmax classification function of the book addition list classifier to obtain the book addition classification result.

[0014] This application has significant technical effects due to the adoption of the above technical solutions: The user information analysis system for library digital resources provided by the application obtains the user's reading interest theme style features by performing data preprocessing and semantic analysis on the user's borrowing history records, collection records, and search records in the digital library, and obtains the content semantic features of the books to be recommended by performing semantic analysis on the text data of the recommended books, and then establishes the connection between the user's reading interest theme style features and the content semantic features of the books to be recommended in the same feature space, and then decides whether to put the books to be recommended into the book recommendation list. In this way, it can provide users with accurate book recommendation services and improve user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 A system block diagram of a user information analysis system for library digital resources according to an embodiment of the present application.

[0017] Figure 2 Schematic diagram of data flow of a user information analysis system for library digital resources according to an embodiment of the present application.

[0018] Figure 3 The present invention is a block diagram of a user information and book information encoding module in a user information analysis system for library digital resources according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a user information feature extraction unit in a user information analysis system for library digital resources according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0021] Library digital resources refer to various information resources that are acquired, organized, stored, and provided with services through digital technology. With the advancement of technology, digital libraries have become an important part of the information society. Specifically, digital libraries use digital means to transform books, periodicals, newspapers, audio-visual materials, maps, etc. into digital resources, and build a huge digital resource library. It has successfully broken the time and space limitations of traditional libraries and greatly improved the convenience of obtaining resources.

[0022] However, in the actual operation of digital libraries today, it is not difficult to find that books recommended to readers are often selected based on the popularity of the recommended books. Although this method can attract readers' attention to a certain extent, it does not fully consider whether the book really meets the readers' reading needs. Even if some digital libraries consider readers' reading needs, they often lack in-depth exploration and only stay on the surface analysis. This leads to the unsatisfactory match between the recommended books and readers' reading preferences, making it difficult to meet readers' personalized needs. To a certain extent, this undoubtedly limits the further improvement of the service quality of digital libraries.

[0023] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and solutions for user information analysis of library digital resources.

[0024] Figure 1 A system block diagram of a user information analysis system for library digital resources according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of user information analysis system for library digital resources according to an embodiment of the present application. Figure 1 and Figure 2 As shown, in the user information analysis system 100 for library digital resources, it includes: a user information book information acquisition module 110, which is used to obtain the user's borrowing history record, collection record and search record in the digital library and the text data of the book to be recommended; a user information book information encoding module 120, which is used to respectively encode the user's borrowing history record, collection record and search record in the digital library and the text data of the book to be recommended to obtain the user's reading interest theme style feature vector and the content semantic feature vector of the book to be recommended; a user information feature book information feature matching module 130, which is used to perform heterogeneous significant feature fusion based on attribute space adaptation on the user's reading interest theme style feature vector and the content semantic feature vector of the book to be recommended to obtain the user reading interest-book content semantic matching feature vector; a book recommendation result generation module 140, which is used to determine whether to add the book to be recommended to the user's book recommendation list based on the user reading interest-book content semantic matching feature vector.

[0025] In the embodiment of the present application, the user information book information acquisition module 110 is used to obtain the user's borrowing history record, collection record and search record in the digital library and the text data of the books to be recommended. It should be understood that the user's borrowing history record and collection record in the digital library reflect the user's reading habits, preferences and areas of interest. For example, frequent borrowing and collection of a certain type of books may indicate that the user has a strong interest in the subject. The user's search record provides the user's current interests and needs, which can help identify the user's focus in a specific time period. The borrowing history, collection record and search record jointly construct the user's reading portrait. By analyzing the user's borrowing, collection and search records, the user's interests and preferences can be understood, thereby providing book recommendations that better meet their needs. The text data of the books to be recommended provides the basic information and content of the books, which can help the system understand the theme and style of the books. By analyzing the text data of the recommended books, it can be matched with the user's reading portrait, thereby realizing personalized recommendation of books, ensuring that the recommendation results are more in line with the user's actual interests, thereby improving the service quality and user experience of the digital library.

[0026] In the embodiment of the present application, the user information book information encoding module 120 is used to encode the user's borrowing history, collection record and search record in the digital library and the text data of the recommended book to obtain the user's reading interest theme style feature vector and the content semantic feature vector of the recommended book. Specifically, Figure 3 FIG. 1 is a block diagram of a user information and book information encoding module in a user information analysis system for library digital resources according to an embodiment of the present application. Figure 3 As shown, the user information book information encoding module 120 includes: a user information feature extraction unit 121, which is used to extract features from the user's borrowing history records, collection records and search records in the digital library to obtain the user's reading interest theme style feature vector; a book information feature extraction unit 122, which is used to extract content features from the text data of the book to be recommended to obtain the content semantic feature vector of the book to be recommended.

[0027] More specifically, Figure 4 FIG. 4 is a block diagram of a user information feature extraction unit in a user information analysis system for library digital resources according to an embodiment of the present application. Figure 4As shown, the user information feature extraction unit 121 includes: a user information preprocessing subunit 1211, which is used to perform data preprocessing on the user's borrowing history records, collection records and search records in the digital library to obtain preprocessed user reading portrait data; a user information related word embedding subunit 1212, which is used to segment the preprocessed user reading portrait data and then pass it through a word embedding layer to obtain a sequence of user reading portrait keyword embedding vectors; a user information related word semantic encoding subunit 1213, which is used to pass the sequence of user reading portrait keyword embedding vectors through a user reading portrait data semantic encoder to obtain the user's reading interest topic style feature vector.

[0028] Considering that the user's borrowing history records, collection records and search record data obtained from the digital library database may contain noise, redundant information, and inconsistent formats, the technical solution of the present application requires data preprocessing of the user's borrowing history records, collection records and search records in the digital library to obtain preprocessed user reading portrait data. Through preprocessing, problematic data in the original collected user data can be removed, thereby improving the accuracy and reliability of the data and providing a high-quality data foundation for subsequent user portrait analysis. It is worth noting here that the preprocessed user reading portrait data are all text data.

[0029] User reading portrait data usually contains a variety of information. In order to divide this information into smaller and meaningful units to make the user analysis more accurate, it is necessary to first perform word segmentation on the preprocessed user reading portrait data. Word segmentation can help the system better understand the semantic relationship between these data, thereby promoting in-depth analysis of user interests. Next, the preprocessed user portrait keywords obtained after word segmentation are passed through the word embedding layer to obtain a sequence of user reading portrait keyword embedding vectors. Those of ordinary skill in the art should know that word embedding is the conversion of discrete words into continuous vector representations, so that machine learning models can process text data. The vector of each word not only contains the information of the word itself, but also captures the semantic relationship between words. By performing word embedding on the preprocessed user portrait keywords after word segmentation, semantically similar words will be mapped to similar vector space positions, which helps the system understand the user's potential interests.

[0030] Taking into account that the word embedding layer converts keywords into vectors, but these vectors themselves cannot fully capture the user's interests and preferences, in order to further extract deep semantic features, it is necessary to pass the sequence of the user reading portrait keyword embedding vectors through the user reading portrait data semantic encoder to obtain the user's reading interest theme style feature vector. The user's reading interest is often closely related to the context. Through the semantic encoder, the user's reading habits and preference characteristics can be better captured. The user reading portrait data semantic encoder described in this application is a bidirectional long short-term memory neural network model. The bidirectional long short-term memory neural network model will consider the previous and next information of the sequence at the same time, which means that it can use the contextual information in front of and behind the user reading portrait keyword embedding vector when processing each user reading portrait keyword embedding vector, so as to more comprehensively understand the input sequence data, and the obtained user reading interest theme style feature vector can better characterize the user's reading style and theme preference.

[0031] More specifically, in the embodiment of the present application, the book information feature extraction unit is used to: segment the text data of the book to be recommended and then pass it through a book text data semantic encoder to obtain the content semantic feature vector of the book to be recommended.

[0032] Considering that book text data usually contains rich information, and vocabulary is the basic unit of expressing meaning in text sentences, in order to cut the text data of the books to be recommended into meaningful word units, so as to more accurately understand the text content of the books to be recommended, it is necessary to perform word segmentation on the text data of the books to be recommended. In order to capture the semantic information in the text data after word segmentation, so as to reflect the key information such as the content and theme of the books to be recommended, and then provide a basis for book recommendation, it is necessary to input it into the book text data semantic encoder to obtain the semantic feature vector of the content of the books to be recommended. The book text data semantic encoder described in this application is a converter-based encoder including a word embedding layer. It should be understood that the word embedding layer can convert the words of the book content after word segmentation into vector representations, and these vectors can capture the semantic relationship between words. The encoder based on the converter can further capture the long-distance dependency and global information in the word embedding vector sequence obtained after the embedding process, so as to generate a more accurate semantic feature vector, i.e., the semantic feature vector of the content of the book to be recommended.

[0033] In an embodiment of the present application, after the text data of the book to be recommended is segmented, a possible implementation method for obtaining the semantic feature vector of the content of the book to be recommended through a book text data semantic encoder can be: the text data of the book to be recommended is segmented to obtain a plurality of book information keywords; the plurality of book information keywords are passed through a word embedding layer to convert each of the plurality of book information keywords into a book information keyword embedding vector to obtain a plurality of book information keyword embedding vectors, wherein the word embedding layer uses a learnable embedding matrix to embed the each book information keyword; the plurality of book information keyword embedding vectors are arranged into a book information input vector; the book information input vector is respectively converted into a book information keyword embedding vector through the learnable embedding matrix into a book information query vector and a book information key vector; calculating the product between the transposed vector of the book information query vector and the book information key vector to obtain a book information self-attention association matrix; normalizing the book information self-attention association matrix to obtain a standardized book information self-attention association matrix; inputting the standardized book information self-attention association matrix into a Softmax activation function for activation to obtain a book information self-attention feature matrix; multiplying the book information self-attention feature matrix with each book information keyword embedding vector in the multiple book information keyword embedding vectors as a value vector to obtain the multiple book information feature vectors; cascading the multiple book information feature vectors to obtain the semantic feature vector of the content of the book to be recommended.

[0034] In the embodiment of the present application, the user information feature book information feature matching module 130 is used to: perform heterogeneous significant feature fusion based on attribute space adaptation on the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to obtain the user reading interest-book content semantic matching feature vector. It should be understood that the user's reading interest theme style feature vector reflects the user's reading preferences and interest points, while the to-be-recommended book content semantic feature vector expresses the theme and content of the book. In order to calculate the degree of matching between the user's interest and the book content in various dimensions, it is necessary to fuse the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to reflect the degree of matching between the user and the book.

[0035] In particular, considering that the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector have different data sources and processing methods, they have obvious feature volatility. And because these two vectors come from different feature spaces, their feature manifolds may not completely match. This mismatch will cause the fused feature vector to fail to maintain the smoothness of the feature manifold during feature fusion. A smooth feature manifold means that the change of the feature vector in space is continuous and reasonable, while the roughness caused by the mismatch will make the fused feature vector appear vague and uncertain when representing the matching relationship between the user's reading interest and the book content, which will affect the fused feature expression. Therefore, in the technical solution of the present application, the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector are subjected to heterogeneous significant feature fusion based on attribute space adaptation to obtain a user reading interest-book content semantic matching feature vector.

[0036] Specifically, in an embodiment of the present application, the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector are subjected to a heterogeneous significant feature fusion based on attribute space adaptation to obtain a user reading interest-book content semantic matching feature vector, including: constructing a user reading interest-book content distance topological matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; constructing a user reading interest-book content feature value granularity association matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; and matching the user reading interest-book content semantic matching feature vector based on the user reading interest-book content distance topological matrix. A topological correlation modulation is performed on a reading interest-book content feature value granularity association matrix to obtain a user reading interest-book content topological modulation association matrix; the user reading interest-book content topological modulation association matrix is ​​used as a perspective co-projection space, and the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector are projected into the perspective co-projection space to obtain a perspective-modulated user reading interest theme style feature vector and a perspective-modulated to-be-recommended book content semantic feature vector; the perspective-modulated user reading interest theme style feature vector and the perspective-modulated to-be-recommended book content semantic feature vector are fused to obtain the user reading interest-book content semantic matching feature vector.

[0037] In an embodiment of the present application, another implementation expression of performing heterogeneous significant feature fusion based on attribute space adaptation on the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to obtain a user reading interest-book content semantic matching feature vector may be: processing the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector using the following formula to obtain a user reading interest-book content semantic matching feature vector; wherein the formula is: ; ; ; ; ; in, represents the feature vector of the topic style that the user is interested in reading, represents the semantic feature vector of the content of the book to be recommended, The user reading interest-book content distance topological matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector is: , The distance topological matrix of the content of the books that the user is interested in reading The eigenvalues ​​of the positions, represents the calculation of the user's reading interest topic style feature vector The feature value of the position and the first The distance of the eigenvalue of the position, and and are column vectors, represents the distance topological matrix between the user's reading interests and the book contents, represents the transpose of a vector, represents matrix multiplication, represents the granularity association matrix of the user's reading interest and the feature value of the book content, represents the natural exponential function, represents the topological modulation association matrix of the book content that the user is interested in reading, represents the perspective modulation user's reading interest topic style feature vector, represents the semantic feature vector of the content of the book to be recommended modulated by the perspective, and represents the weighted hyperparameter, Indicates the semantic matching feature vector of the book content that the user is interested in reading.

[0038] That is, in the technical solution of the present application, the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector are subjected to heterogeneous significant feature fusion based on attribute space adaptation, which first constructs a user reading interest-book content distance topological matrix between the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector. This step utilizes the idea of ​​graph theory, taking the eigenvalues ​​of each position in the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector as nodes, and using a distance metric function to determine the low-dimensional embedding expression of the edges between nodes, so as to construct a feature expression that can quantify the topological correlation relationship between the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector. This step is crucial to understanding the spatial distribution of data points and their interconnections, and also provides basic structural information for subsequent steps.

[0039] Next, a user reading interest-book content feature value granularity association matrix is ​​constructed between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector. This step focuses on exploring the correlation between the internal attributes of the feature vector. In a specific example, the product between the transposed vector of the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector is calculated to obtain the user reading interest-book content feature value granularity association matrix. Furthermore, the user reading interest-book content feature value granularity association matrix is ​​topologically modulated based on the user reading interest-book content distance topological matrix to obtain the user reading interest-book content topological modulation association matrix. That is, the feature value granularity association information of the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector is driven to wander on the distance topological map to construct a low-dimensional modulated perspective co-projection space domain for mapping the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to a continuous high-dimensional regression space attribute.

[0040] Next, the user reading interest-book content topological modulation association matrix is ​​used as the perspective co-projection space, and the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector are projected into the perspective co-projection space to obtain the perspective modulation user reading interest theme style feature vector and the perspective modulation to-be-recommended book content semantic feature vector. That is, the key to this technical solution is to find an implicit third space with modulation capability, in which as much information of the original high-dimensional data as possible is retained, so that the converted new feature vector not only reduces the dimension, but also maintains the relative position relationship between the original features.

[0041] Finally, the perspective-modulated user reading interest theme style feature vector and the perspective-modulated recommended book content semantic feature vector are fused to obtain the user reading interest-book content semantic matching feature vector. Fusion can be achieved in a variety of forms, ranging from simple arithmetic average to complex ensemble learning algorithms, depending on the application scenario and the goals pursued. The final user reading interest-book content semantic matching feature vector combines the main features of the two sets of input data and provides a more comprehensive data representation, which is not only conducive to improving the performance of machine learning models, but also promotes the ability of cross-modal or multi-source data analysis, thereby significantly enhancing the understanding and utilization efficiency of complex data structures.

[0042] In an embodiment of the present application, the book recommendation result generation module 140 is used to determine whether to add the to-be-recommended book to the user's book recommendation list based on the user's reading interest-book content semantic matching feature vector. Specifically, in an embodiment of the present application, the book recommendation result generation module is used to: pass the user's reading interest-book content semantic matching feature vector through a book addition list classifier to obtain a book addition classification result, and the book addition classification result is used to indicate whether to add the to-be-recommended book to the user's book recommendation list. More specifically, in an embodiment of the present application, the book recommendation result generation module is used to: use the fully connected layer of the book addition list classifier to fully connect the user's reading interest-book content semantic matching feature vector to obtain a user's reading interest-book content semantic matching encoding classification feature vector; input the user's reading interest-book content semantic matching encoding classification feature vector into the Softmax classification function of the book addition list classifier to obtain the book addition classification result. It should be understood that by inputting the user's reading interest-book content semantic matching feature vector into the book addition list classifier, the degree of matching between each book and the user's interest can be accurately evaluated, thereby providing a personalized recommendation service for the user. Specifically, the book addition list classifier is a model based on machine learning algorithms that can parse and classify the user's interest in reading - book content semantic matching feature vectors, thereby determining whether each book should be added to the user's book recommendation list. Through the judgment of the classifier, the books that best match the user's interests can be screened out, thereby ensuring the accuracy and pertinence of the book recommendation results.

[0043] In summary, the user information analysis system 100 for library digital resources based on the embodiment of the present application is explained, which obtains the user's reading interest theme style features by performing data preprocessing and semantic analysis on the user's borrowing history records, collection records, and search records in the digital library, and obtains the content semantic features of the books to be recommended by performing semantic analysis on the text data of the recommended books, and then establishes the connection between the user's reading interest theme style features and the content semantic features of the books to be recommended in the same feature space, and then decides whether to put the books to be recommended into the book recommendation list. In this way, it can provide users with accurate book recommendation services and improve user satisfaction.

[0044] As described above, the user information analysis system 100 for library digital resources according to the embodiment of the present application can be implemented in various terminal devices, such as a server for user information analysis for library digital resources. In one example, the user information analysis system 100 for library digital resources according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the user information analysis system 100 for library digital resources can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the user information analysis system 100 for library digital resources can also be one of the many hardware modules of the terminal device.

[0045] Alternatively, in another example, the user information analysis system 100 for library digital resources and the terminal device may also be separate devices, and the user information analysis system 100 for library digital resources may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0046] In the embodiments provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0047] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0048] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0049] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application.

[0050] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present application is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present application. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0051] In addition, it is clear that the word "comprising" does not exclude other units, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit of the technical solution of the present application.

Claims

1. A user information analysis system for library digital resources, characterized in that: include: The user information and book information acquisition module is used to acquire the user's borrowing history, collection record and search record in the digital library and the text data of the books to be recommended; A user information and book information encoding module, for encoding the borrowing history, collection record and search record of the user in the digital library and the text data of the book to be recommended to obtain a theme style feature vector of the user's reading interest and a content semantic feature vector of the book to be recommended; A user information feature book information feature matching module is used to perform heterogeneous significant feature fusion based on attribute space adaptation on the user's reading interest theme style feature vector and the to-be-recommended book content semantic feature vector to obtain a user reading interest-book content semantic matching feature vector; The book recommendation result generating module is used to determine whether to add the to-be-recommended book to the user's book recommendation list based on the user's reading interest-book content semantic matching feature vector.

2. The user information analysis system for library digital resources according to claim 1 is characterized in that: The user information and book information encoding module includes: A user information feature extraction unit, used for extracting features from the borrowing history records, collection records and search records of the user in the digital library to obtain a feature vector of the user's reading interest subject style; The book information feature extraction unit is used to extract content features from the text data of the book to be recommended to obtain a content semantic feature vector of the book to be recommended.

3. The user information analysis system for library digital resources according to claim 2 is characterized in that: The user information feature extraction unit comprises: A user information preprocessing subunit, used for preprocessing the borrowing history records, collection records and search records of the user in the digital library to obtain preprocessed user reading portrait data; A user information related word embedding subunit, used to segment the pre-processed user reading portrait data and pass it through a word embedding layer to obtain a sequence of user reading portrait keyword embedding vectors; The user information related word semantic encoding subunit is used to pass the sequence of the user reading portrait keyword embedding vector through the user reading portrait data semantic encoder to obtain the user reading interest topic style feature vector.

4. The user information analysis system for library digital resources according to claim 3 is characterized in that: The book information feature extraction unit is used to: segment the text data of the book to be recommended and then pass it through a book text data semantic encoder to obtain the content semantic feature vector of the book to be recommended.

5. The user information analysis system for library digital resources according to claim 4 is characterized in that: The user reading portrait data semantic encoder is a bidirectional long short-term memory neural network model, and the book text data semantic encoder is a converter-based encoder including a word embedding layer.

6. The user information analysis system for library digital resources according to claim 5 is characterized in that: The user information feature book information feature matching module is used to: Constructing a user reading interest-book content distance topological matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; Constructing a user reading interest-book content feature value granularity association matrix between the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector; Based on the user reading interest-book content distance topological matrix, the user reading interest-book content feature value granularity association matrix is ​​topologically modulated to obtain a user reading interest-book content topological modulation association matrix; Taking the user reading interest-book content topological modulation association matrix as the perspective co-projection space, projecting the user reading interest theme style feature vector and the to-be-recommended book content semantic feature vector into the perspective co-projection space to obtain the perspective modulation user reading interest theme style feature vector and the perspective modulation to-be-recommended book content semantic feature vector; The perspective-modulated user reading interest theme style feature vector and the perspective-modulated to-be-recommended book content semantic feature vector are fused to obtain the user reading interest-book content semantic matching feature vector.

7. The user information analysis system for library digital resources according to claim 6 is characterized in that: The book recommendation result generation module is used to: pass the user's reading interest-book content semantic matching feature vector through a book addition list classifier to obtain a book addition classification result, and the book addition classification result is used to indicate whether to add the to-be-recommended book to the user's book recommendation list.

8. The user information analysis system for library digital resources according to claim 7 is characterized in that: The book recommendation result generating module is used to: Use the fully connected layer of the book addition list classifier to perform fully connected encoding on the user reading interest-book content semantic matching feature vector to obtain the user reading interest-book content semantic matching encoding classification feature vector; The user reading interest-book content semantic matching encoded classification feature vector is input into the Softmax classification function of the book addition list classifier to obtain the book addition classification result.

Images