Dynamic library management method and system based on artificial intelligence

Abstract

The invention discloses a dynamic library management method and system based on artificial intelligence, and relates to the technical field of library management, and the method comprises the following five steps: firstly, constructing a reader portrait through collecting basic information, borrowing history and other data of a reader by using TF-IDF analysis and a machine learning algorithm; secondly, a positioning tag and a sensor are deployed to be combined with a book RFID tag, a path is optimized according to reader behavior data, and the book retrieval efficiency is improved; monitoring and analyzing borrowing behavior data and reader portraits, establishing a borrowing risk assessment model, assessing risks by adopting a logistic regression algorithm, providing decision suggestions for workers, quantitatively adjusting the borrowing quantity, and reducing the risk of overdue or damaged books; through application of the artificial intelligence technology, automation, intellectualization and individuation of library management are realized, the management efficiency and the service quality are remarkably improved, and the operation effect of the library is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of library management, and specifically to a dynamic library management method and system based on artificial intelligence. Background Art

[0002] Traditional library management methods mainly rely on manual operations and simple automation systems. In the processes of book borrowing, returning, and retrieval, staff need to manually record and manage a large amount of book information and reader data. This management method has many deficiencies. First, the work of classifying, organizing, and shelving books is time-consuming and laborious. Especially in large libraries, with a wide variety of books and a large number of them, staff need to spend a lot of time classifying and locating books, resulting in low efficiency of book management. For example, after a book is returned, staff need to put it back in the correct position according to its classification number, and this process is prone to errors, making it difficult for readers to quickly locate the required books when searching for them. Second, traditional libraries also have limitations in reader services. Regarding readers' borrowing needs and reading preferences, libraries often lack effective means of understanding and analysis, and it is difficult to provide personalized book recommendations and precise services for readers. For example, when readers are looking for interesting books, they may need to spend a lot of time browsing bookshelves or querying catalogs, and the library cannot actively recommend relevant books based on readers' reading history and interests, reducing the borrowing experience and satisfaction of readers.

[0003] With the development of information technology, some libraries have begun to introduce electronic management systems and simple automated equipment in order to improve management efficiency and service quality. However, there are still many deficiencies in these existing technologies. First of all, the existing electronic management systems have relatively single functions, mainly focusing on aspects such as book borrowing registration, return records, and inventory management, lacking in-depth analysis of readers and intelligent book recommendation functions. For example, the system cannot automatically analyze readers' interest preferences based on their borrowing history and reading habits and recommend relevant books accordingly, resulting in readers having difficulty quickly finding the books they are interested in among a vast number of books. Secondly, existing automated equipment such as self-service borrowing and returning machines, although simplifying the borrowing process to a certain extent, still has limitations in book positioning and navigation. For example, when readers use self-service borrowing and returning machines, they can quickly complete the borrowing procedures, but still need to rely on manual or simple catalog guidance when looking for books, unable to achieve precise book positioning and navigation, making it still difficult for readers to find books in the library. In addition, existing technologies also have deficiencies in borrowing risk management and overdue management. The risk assessment of readers' borrowing behaviors is not accurate enough, unable to provide effective decision-making support for library staff, resulting in difficulties in making reasonable adjustments to borrowing strategies when facing readers with different risk levels. At the same time, in terms of overdue management, there is a lack of effective behavior guidance and incentive mechanisms. For readers who fail to return books on time, mainly relying on traditional means such as fines, it is difficult to effectively encourage readers to return books on time, affecting the circulation efficiency of books and the normal operation of the library.

[0004] To solve the above defects, a technical solution is provided now. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems of low management efficiency, lack of personalization in services, and deficiencies in borrowing risk and overdue management in traditional libraries, and to propose a dynamic library management method and system based on artificial intelligence.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A dynamic library management method based on artificial intelligence, comprising the following steps:

[0008] S1: Construction of reader portraits and demand analysis, collecting multi-faceted data of readers, using TF-IDF analysis to extract interest preferences and reading habit characteristics, and then using machine learning algorithms to construct portraits, segment groups, and analyze book demands;

[0009] S2: Intelligent book positioning and navigation guidance, deploying positioning tags and sensors on bookshelves to track positions in combination with book RFID tags. When readers query, indoor positioning technology is used to generate and present navigation paths based on position information, and the paths are optimized using readers' behavior data;

[0010] S3: Borrowing risk assessment and decision-making. Monitor and analyze borrowing behavior data, build a model with the reader profile, evaluate risks using the logistic regression algorithm, provide decision-making suggestions for staff based on the results, and quantitatively adjust the borrowing quantity.

[0011] S4: Intelligent recommendation and reservation borrowing. Generate a recommendation list using the collaborative filtering algorithm based on the reader profile and book-related data. Readers can make reservation borrowings. Arrange reserved positions according to the reservations and in-stock situations and notify the borrowings to achieve seamless connection.

[0012] S5: Overdue management and behavior guidance. Push reminders when the overdue preset critical point is reached, analyze the reasons for those who have not returned on time and take corresponding strategies. At the same time, establish a social incentive and credit publicity mechanism to promote timely return.

[0013] Furthermore, the process of S1 is as follows:

[0014] Collect the basic information of readers, borrowing history, book browsing records, in-library stay time, and location data.

[0015] Use natural language processing technology to analyze the search terms and book review text data of readers on the library online platform, and extract the interest preference characteristics and reading habit characteristics of readers.

[0016] Based on the collected data, use machine learning algorithms to build a reader profile, divide readers into different interest groups, and analyze the book needs of each group and individual readers to provide a basis for book recommendation and borrowing service optimization.

[0017] Furthermore, the specific operation steps for analyzing the interest preference characteristics and reading habit characteristics of readers in S1 are as follows:

[0018] Use the term frequency-inverse document frequency formula to measure the reader's interest in different topic words. Let the text data set of the reader be D, the term frequency of the word t in the document d be tf(t, d), the number of documents containing the word t be df(t), and the total number of documents be N. Then the TF-IDF value of the word for the reader is By calculating the TF-IDF values of each topic-related word, obtain an interest preference vector for subsequent clustering analysis and profile construction, reflecting the interest weights of readers in different topics.

[0019] When analyzing reading habit characteristics, calculate the average borrowing cycle C a and the standard deviation σ of the borrowing frequency f ; The average borrowing cycle where n is the number of borrowings, F q is the statistical browsing frequency, D p is the browsing depth, and obtain the reading habit characteristic vector Fh = {C a , F q , D p …}; Standard deviation of borrowing frequency where f i represents the reciprocal of the time interval between the i-th borrowing and the previous borrowing, is the average borrowing frequency; The standard deviation is used to reflect the stability of the readers' borrowing frequency, and together with the average borrowing cycle, it describes the reading habits and identifies different types of readers, namely regular borrowers and random borrowers.

[0020] Furthermore, the specific operation steps of S2 are as follows:

[0021] Deploy ultra-wideband positioning tags and sensors on the library shelves, and combine with the RFID tag information of books to track the location of books;

[0022] When a reader queries a certain book through the library mobile application, based on the reader's current location and the real-time location of the book, use indoor positioning technology to generate the optimal navigation path from the reader's location to the target book location, and present it to the reader in the form of a visual map on the mobile application to guide the reader to quickly find the required book; When generating the navigation path, use a personalized recommendation mechanism to optimize the navigation path.

[0023] Furthermore, the specific operation steps of optimizing the navigation path by the personalized recommendation mechanism in S2 are as follows:

[0024] Continuously collect the reader's behavior data in the library, including borrowing history records, movement trajectories in the library, and the staying time and browsing behavior in front of the bookshelves; and store it in the library's big data storage system to form a reader behavior database; Preprocess the collected data, including data cleaning, data standardization, and data classification;

[0025] Use data analysis algorithms and machine learning techniques to deeply analyze the reader behavior data; Adopt the clustering analysis algorithm to divide readers with similar borrowing behaviors and in-library activity patterns into different groups; Use the association rule mining algorithm to find the association relationships between different book categories and the associations between reader behavior and book interests;

[0026] When a reader initiates a book query request, first search for their relevant behavior characteristics and interest preferences in the reader behavior database and interest model according to the reader's identity information; Combine the library's bookshelf layout map and real-time information on the density of people flow and the occupancy of passages, and design a dynamic path planning algorithm; This algorithm is improved on the basis of the traditional path planning algorithm, taking the reader's interest points and the areas of books of interest as reference factors for path planning;

[0027] Calculate a comprehensive evaluation score for each candidate path, with evaluation factors including path length, the number and weights of areas of interest passed through, and the degree of crowd congestion; finally, select the path with the highest comprehensive evaluation score as the navigation path recommended to the reader.

[0028] During the process of the reader walking along the navigation path, continuously monitor the changes in the reader's position and behavior; when the reader deviates from the predetermined path, immediately re-evaluate the path between the current position and the target book position, and generate a new navigation path based on the latest library environment information and the reader's new points of interest, and push it to the reader.

[0029] Furthermore, the specific operation steps of S3 are as follows:

[0030] Continuously monitor and analyze the reader's borrowing behavior data, including borrowing frequency, number of overdue times, and timeliness of return, and combine the reader portrait information to establish a borrowing risk assessment model.

[0031] The model uses machine learning algorithms to conduct a risk assessment of the reader's next borrowing behavior, predict the possibility of overdue or damaged books, and provide decision-making suggestions for library staff based on the assessment results.

[0032] When establishing the borrowing risk assessment model, use the logistic regression algorithm. Let the reader's risk assessment value be P(y = 1|x), where y = 1 indicates the risk of overdue or damaged books, and x is the feature vector. Then the logistic regression formula is where w is the weight vector and b is the bias term; the feature vector x contains the values of borrowing frequency, number of overdue times, average borrowing period, and interest preference features after standardization; optimize w and b through training data to enable the model to accurately predict the borrowing risk probability of readers and provide a quantitative basis for decision-making.

[0033] For the decision to increase the borrowing quantity for low-risk readers, determine the new borrowing quantity L according to the risk assessment value and the current borrowing limit L 0 to determine the new borrowing quantity L; let the increase ratio coefficient be α. When P < P 0 where P 0 is the low-risk threshold, L = L 0 (1 + α(1 - P)), dynamically adjust the borrowing quantity according to the risk level; for high-risk readers, require guarantee measures or shorten the borrowing period, and set different guarantee amounts or shortening ratios according to the risk value.

[0034] Furthermore, the specific operation steps of S4 are as follows:

[0035] According to the reader portrait and the classification, popularity, and borrowing history data of books, use collaborative filtering algorithms or content-based recommendation algorithms to generate a personalized book recommendation list for readers.

[0036] In the collaborative filtering algorithm, the Pearson correlation coefficient formula is used to measure the similarity of interests between readers. Suppose the ratings of reader u and reader v for n books are r ui and r vi , and their average ratings are and Then the Pearson correlation coefficient between readers and is By calculating the similarity between readers, the neighbor readers most similar to the target reader are found, and a recommendation list is generated based on the borrowing history and ratings of the neighbor readers to improve the accuracy and pertinence of the recommendation;

[0037] Readers browse the recommended books on the mobile application and directly perform the reservation borrowing operation; according to the reservation situation and the in - stock status of the books, the reserved in - stock positions of the books are automatically arranged, and after the books are returned, the readers are notified to come and borrow, realizing the seamless connection of book borrowing, improving the circulation efficiency of books and the borrowing satisfaction of readers.

[0038] Further, the specific operation steps of S5 are as follows:

[0039] When the books are approaching the due date, reminders are sent to readers by means of text messages and mobile application push. The reminder content includes the consequences of overdue, as well as convenient return methods and location guides;

[0040] For readers who have not returned the books overdue, data analysis is used to mine the reasons for their overdue, including overdue due to forgetting, business trips, and book loss; for different reasons, different coping strategies are adopted. For readers who forget to return, the reminder frequency and method are optimized; for readers who are unable to return on time due to unexpected situations, an online application function for extending the return is provided and approved according to the reader's credit record and past performance; for the situation of book loss, the reader is guided to go through the compensation process, and a channel for purchasing the same book to return to the library is provided, and it is recorded in the reader's credit file for subsequent reference in borrowing management;

[0041] Establish a social incentive and credit publicity mechanism to promote timely return.

[0042] Further, the specific operation steps of establishing a social incentive and credit publicity mechanism in S5 to promote timely return are as follows:

[0043] Introduce social functions in the library mobile application. Readers who return books on time are given preset points or virtual badge rewards. These points are used to exchange relevant rights and interests of the library, including longer borrowing periods and priority borrowing of popular books;

[0044] Meanwhile, a borrowing credit ranking list is set up. On the premise of the reader's authorization, their credit rating and borrowing performance are displayed to friends or readers in the same interest group, using social pressure and incentive mechanisms to encourage readers to return books on time;

[0045] For readers who have not returned books overdue, after their credit rating drops to a preset level, the use of their preset social functions is restricted until they return the books and improve their credit status, thereby enhancing the reader's attention to overdue behavior.

[0046] An artificial intelligence-based dynamic library management system, comprising:

[0047] A data collection module, used to collect readers' basic information, various borrowing data, and book RFID and location data, providing basic data for subsequent use;

[0048] A data analysis and processing module, used to analyze readers' online data and borrowing and browsing records, extract features to construct portraits, segment groups, and analyze requirements; also used to collect and process readers' in-library behavior data, analyze and mine associations, and provide support for other modules;

[0049] A book positioning and navigation module, used to combine book tags and sensors to track positions, generate and optimize navigation paths according to readers' queries, and update in real time to guide the search for books;

[0050] A borrowing risk assessment module, used to monitor and analyze borrowing data and portraits, build a model with algorithms to predict risks, and provide decision-making suggestions for staff;

[0051] An intelligent recommendation and reservation borrowing module, used to generate a recommendation list with algorithms based on portraits and book data. Readers can reserve books for borrowing, and the system arranges reservations and notifications to achieve seamless connection;

[0052] An overdue management module, used to push reminders when approaching overdue, analyze the reasons for overdue and take corresponding strategies, and at the same time establish a social incentive and credit publicity mechanism to promote timely return.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] (1) In the present invention, through artificial intelligence technology, the automation and intelligence of library management are realized, significantly improving management efficiency; in terms of book positioning and navigation, using advanced positioning technology and personalized recommendation mechanisms, the system can quickly generate the optimal navigation path, reduce manual intervention, and improve the speed and accuracy of book retrieval and management; when readers query books, the navigation path is generated immediately to guide them to quickly find the required books, avoiding the time-consuming and laborious traditional manual search; in terms of borrowing risk assessment, through machine learning algorithms to analyze readers' behaviors, a precise risk assessment model is established to provide decision-making support for the library, optimize borrowing strategies, and reduce management costs and risks;

[0055] (2) Through constructing detailed reader portraits and providing personalized services, the present invention has greatly improved the borrowing experience of readers; by using machine learning algorithms to analyze reader data, accurately extracting interest preferences and reading habits, providing personalized book recommendations for readers, helping them quickly find books they are interested in, and improving borrowing satisfaction; at the same time, the intelligent recommendation and reservation borrowing functions simplify the borrowing process. Readers can directly make reservation borrowings on the mobile application, automatically arrange the reserved positions of books and notify readers in a timely manner, reducing waiting time and inconvenience, enabling readers to more conveniently enjoy library resources and services;

[0056] (3) Through effective borrowing risk management and overdue management mechanisms, the present invention has enhanced the operational effectiveness of the library; in terms of borrowing risk management, an accurate risk assessment model enables the library to reasonably adjust borrowing strategies, increase the borrowing quantity for low-risk readers, and improve the book circulation efficiency; taking measures such as requiring guarantees or shortening the borrowing period for high-risk readers to reduce the risk of overdue or damaged books, reasonably allocate resources, and improve resource utilization rate. In terms of overdue management, by analyzing and mining the reasons for overdue through data, adopting targeted strategies, and establishing social incentives and credit publicity mechanisms, encouraging readers to return books on time, reducing the number of overdue books, improving the book turnover rate and circulation efficiency, reducing operating costs, and enhancing the overall operational level of the library. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings;

[0058] Figure 1 is the method flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] It should be understood that the terms "including" and "comprising" used in the specification and claims of this disclosure indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0061] It should also be understood that the terms used in this disclosure statement are for the purpose of describing specific embodiments only and are not intended to limit this disclosure. As used in this disclosure statement and the claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms. It should be further understood that the term "and / or" used in this disclosure statement and the claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0062] As Figure 1 shown, an artificial intelligence-based dynamic library management method includes the following steps:

[0063] Reader portrait construction and demand analysis, collecting readers' multi-faceted data, using analysis such as TF-IDF to extract interest preferences and reading habit characteristics, and then using machine learning algorithms to construct portraits, segment groups and analyze book demands;

[0064] Collecting readers' basic information, borrowing history, book browsing records, in-library stay time and location data; using natural language processing technology to analyze the search keywords and book review text data of readers on the library online platform to extract readers' interest preference characteristics and reading habit characteristics;

[0065] Specifically, the term frequency-inverse document frequency (TF-IDF) formula is used to measure the degree of readers' interest in different topic words. Let the text data set of readers be D, the term frequency of word t in document d be tf(t, d), the number of documents containing word t be df(t), and the total number of documents be N. Then the TF-IDF value of the word for readers is By calculating the TF-IDF values of each topic-related word, an interest preference vector is obtained for subsequent clustering analysis and portrait construction, which can more accurately reflect the interest weights of readers in different topics; when analyzing reading habit characteristics, the average borrowing cycle C a and the standard deviation σ f of the borrowing frequency are calculated; the average borrowing cycle (n is the number of borrowings), the browsing frequency (F q ) and the browsing depth (D p ) are statistically analyzed, and a reading habit feature vector F h ={C a , F q , Dp...}; the standard deviation of the reading frequency where f i represents the reciprocal of the time interval between the i-th borrowing and the previous borrowing (i.e., the borrowing frequency), is the average borrowing frequency; the standard deviation is used to reflect the stability of readers' borrowing frequencies, and together with the average borrowing cycle, it describes reading habits and helps identify different types of readers such as regular borrowers and random borrowers;

[0066] Based on the collected data, machine learning algorithms are used to construct reader portraits, readers are segmented into different interest groups, and the book needs of each group and individual readers are analyzed to provide a basis for book recommendations and the optimization of borrowing services.

[0067] Intelligent book positioning and navigation guidance. Location tags and sensors are deployed on the bookshelves to track the location of books in combination with book RFID tags. When readers query, indoor positioning technology is used to generate and present a navigation path based on the location information, and the path is optimized using readers' behavior data;

[0068] Ultra-wideband (UWB) location tags and sensors are deployed on the library bookshelves. Combining with the RFID tag information of books, the location of books is tracked. When a reader queries a certain book through the library mobile application, based on the reader's current location and the real-time location of the book, indoor positioning technology is used to generate the optimal navigation path from the reader's location to the target book location and present it to the reader in the form of a visual map on the mobile application to guide the reader to quickly find the required book. When generating the navigation path, a personalized recommendation mechanism is used to optimize the navigation path. The process of the personalized recommendation mechanism is as follows:

[0069] Continuously collect readers' behavior data in the library, including borrowing history records (detailed records of the book categories, borrowing times, return times, etc. for each borrowing), movement trajectories in the library (recording the location coordinates of readers at different time points through the library's positioning system, such as Wi-Fi positioning or Bluetooth positioning technology), and the staying time and browsing behavior in front of the bookshelves (using sensors near the bookshelves to determine when readers approach the bookshelves, the staying duration, and whether there are actions of taking and putting back books). These data are stored in the library's big data storage system to form a reader behavior database; preprocess the collected data, including data cleaning (removing outliers and incorrect data, such as unreasonable staying times or incorrect location coordinates), data standardization (converting data in different formats and units into a unified standard for subsequent analysis), and data classification (classifying and organizing according to dimensions such as reader ID and time series);

[0070] Use data analysis algorithms and machine learning techniques to conduct in-depth analysis of reader behavior data. Use cluster analysis algorithms to divide readers with similar borrowing behaviors and in-library activity patterns into different groups. For example, readers who frequently borrow literature books and stay in the literature bookshelf area for a long time are classified as literature lovers; readers who frequently borrow science and technology books and attend related lectures or seminars are classified as science and technology explorers. Construct interest models for each reader group and individual. Use association rule mining algorithms to find out the associations between different book categories and the associations between reader behavior and book interests. For example, it is found that readers who have borrowed an introductory computer programming book have a higher probability of subsequently borrowing more advanced programming books or related algorithm books; for readers who stay in the history and culture bookshelf area for a long time and frequently borrow books from a specific historical period, their interest in in-depth research on that period is inferred, and these association information and interest weights are recorded in the interest model;

[0071] When a reader initiates a book query request, the path planning system first searches for the reader's relevant behavioral characteristics and interest preferences in the reader behavior database and interest model based on the reader's identity information. A dynamic path planning algorithm is designed based on the library's shelf layout diagram and real-time crowd density and channel occupancy information. This algorithm improves on traditional path planning algorithms (such as the A* algorithm or the Di jkstra algorithm) and takes readers' points of interest and areas of possible interest as important reference factors for path planning. A comprehensive evaluation score is calculated for each candidate path, and the evaluation factors include path length, the number and weight of interest areas passed through, and the degree of crowd congestion; finally, the path with the highest comprehensive evaluation score is selected as the navigation path recommended to the reader.

[0072] As readers walk along the navigation path, the system continuously monitors the readers' positions and behavior changes. If readers deviate from the predetermined path, the system immediately re-evaluates the path between the current location and the target book location, and quickly generates a new navigation path and pushes it to the reader based on the latest library environment information (such as changes in traffic flow, temporary adjustments to bookshelves, etc.) and the reader's possible new points of interest (such as approaching a new bookshelf area and staying for a period of time while deviating from the path).

[0073] Borrowing risk assessment and decision-making: monitor and analyze borrowing behavior data and reader portraits to build models, use logistic regression algorithms to assess risks, provide decision-making suggestions to staff based on the results, and quantitatively adjust the borrowing quantity;

[0074] Continuously monitor and analyze the borrowing behavior data of readers, including borrowing frequency, number of overdue times, and timeliness of return. Combine the reader profile information to establish a borrowing risk assessment model. The model uses machine learning algorithms to conduct a risk assessment on the next borrowing behavior of readers, predict the possibility of overdue or damaged books, and provide decision-making suggestions for library staff based on the assessment results. When establishing the borrowing risk assessment model, if the logistic regression algorithm is used, assuming that the risk assessment value of the reader is P(y = 1|x) (y = 1 indicates the risk of overdue or damaged books, and x is the feature vector), the logistic regression formula is where w is the weight vector and b is the bias term. The feature vector x contains the values of borrowing frequency, number of overdue times, average borrowing period, and interest preference features after standardization processing. Optimize w and b through training data to enable the model to accurately predict the borrowing risk probability of readers and provide a quantitative basis for decision-making;

[0075] For the decision to increase the borrowing quantity for low-risk readers, the new borrowing quantity L can be determined according to the risk assessment value and the current borrowing limit L 0 Let the increase ratio coefficient be α (0 < α < 1). When P < P 0 (low-risk threshold), L = L 0 (1 + α(1 - P)). Dynamically adjust the borrowing quantity according to the risk level to achieve more reasonable resource allocation. For high-risk readers, require guarantee measures or shorten the borrowing period. Similarly, different guarantee amounts or shortening ratios can be set according to the risk value to make the decision-making process more scientific and standardized.

[0076] Intelligent recommendation and reservation borrowing. Generate a recommendation list using the collaborative filtering algorithm based on the reader profile and book-related data. For readers' reservation borrowing, the system arranges reserved positions according to the reservation and in-stock situation and notifies the borrowing to achieve seamless connection;

[0077] Generate a personalized book recommendation list for readers according to the reader profile and the classification, popularity, and borrowing history data of books using the collaborative filtering algorithm or content-based recommendation algorithm;

[0078] In the collaborative filtering algorithm, the Pearson correlation coefficient formula is used to measure the interest similarity between readers. Let the ratings of reader u and reader v for n books be r ui and r vi , and their average ratings be and Then the Pearson correlation coefficient between readers and is By calculating the similarity between readers, find the neighbor readers most similar to the target reader, and generate a recommendation list based on the borrowing history and ratings of the neighbor readers to improve the accuracy and pertinence of the recommendation;

[0079] Readers can browse the recommended books on the mobile application and directly make reservation borrowing operations; according to the reservation situation and the in - shelf status of the books, automatically arrange the reserved in - shelf positions of the books, and after the books are returned, notify the readers to come and borrow, realizing seamless connection of book borrowing, improving the circulation efficiency of books and the borrowing satisfaction of readers.

[0080] Overdue management and behavior guidance, push reminders when approaching the due date, analyze the reasons for those who have not returned on time and take corresponding strategies, and at the same time establish a social incentive and credit publicity mechanism to promote timely return;

[0081] When the books are approaching the due date, send reminders to readers by means of text messages and mobile application push. The reminder content includes the consequences of overdue, as well as convenient return methods and location guides; for readers who have not returned on time, use data analysis to dig out the reasons for their overdue, such as whether it is due to forgetting, business trips, or book loss. For different reasons, adopt different coping strategies. For readers who forget to return, optimize the reminder frequency and method; for readers who are unable to return on time due to special circumstances, provide an online application function for extension of return, and approve according to the reader's credit record and past performance; for the situation of book loss, guide the reader through the compensation process, provide channels to purchase the same book and return it to the library, and record it in the reader's credit file for subsequent reference in borrowing management.

[0082] Establish social incentives and credit publicity: Introduce social functions in the library mobile application. Give preset points or virtual badges as rewards to readers who return books on time. These points are used to exchange relevant rights and interests of the library, including longer borrowing periods and priority borrowing of popular books. At the same time, set up a borrowing credit ranking list. On the premise of reader authorization, display their credit levels and borrowing performances to friends or readers in the same interest group, and use social pressure and incentive mechanisms to encourage readers to return books on time; for readers who have not returned on time, after their credit levels drop to a certain extent, restrict the use of some of their social functions until they return the books and improve their credit status, so as to enhance readers' attention to overdue behaviors.

[0083] An artificial - intelligence - based dynamic library management system, including:

[0084] Data collection module: Collect various data such as readers' basic information, borrowing, as well as book RFID and location data, providing basic data for follow - up;

[0085] Data analysis and processing module: Analyze readers' online data and borrowing and browsing records, extract features to construct portraits, segment groups and analyze needs; collect and process readers' in - library behavior data, analyze and mine associations, and provide support for other modules;

[0086] Book positioning and navigation module: Combines book tags and sensors to track the location, generates and optimizes the navigation path according to the reader's query, and updates the guidance for finding books in real time;

[0087] Borrowing risk assessment module: Monitors and analyzes borrowing data and portraits, uses algorithms to build models to predict risks, and provides decision-making suggestions for staff;

[0088] Intelligent recommendation and reservation borrowing module: Generates a recommendation list using algorithms based on portraits and book data. Readers can reserve books for borrowing, and the system arranges reservations and notifications to achieve seamless connection;

[0089] Overdue management module: Pushes reminders when approaching the due date; Analyzes the reasons for overdue and adopts corresponding strategies; Establishes a social incentive and credit publicity mechanism to promote timely return.

[0090] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation manners. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A dynamic library management method based on artificial intelligence, characterized in that: The following steps are involved: S1: Reader portrait construction and demand analysis: collect various data of readers, use TF-IDF analysis to extract interest preferences and reading habit characteristics, and then use machine learning algorithms to construct portraits, segment groups and analyze book needs; S2: Intelligent book positioning and navigation guidance. Positioning tags and sensors are deployed on bookshelves to track the location of books in combination with RFID tags. When readers query, they use indoor positioning technology to generate and present navigation paths based on location information, and use reader behavior data to optimize the path. S3: Borrowing risk assessment and decision-making, monitoring and analyzing borrowing behavior data and reader portraits to build models, using logistic regression algorithms to assess risks, providing decision-making suggestions to staff based on the results and quantitatively adjusting the borrowing quantity; S4: Intelligent recommendation and reservation borrowing: Based on the reader portrait and book-related data, a collaborative filtering algorithm is used to generate a recommendation list. Readers make reservations for borrowing, and reserved spaces are arranged and borrowing is notified based on reservations and shelf status, achieving seamless connection; S5: Overdue management and behavior guidance: push reminders when the preset overdue critical point is reached, analyze the reasons for overdue payments and take corresponding strategies, and establish social incentives and credit publicity mechanisms to promote on-time repayment.

2. The method for dynamic library management based on artificial intelligence according to claim 1, characterized in that: The process of S1 is as follows: Collect readers' basic information, borrowing history, book browsing history, time spent in the library and location data; Use natural language processing technology to analyze readers’ search terms and book review text data on the library’s online platform to extract readers’ interest preference characteristics and reading habit characteristics; Based on the collected data, machine learning algorithms are used to build reader portraits, segment readers into different interest groups, and analyze the book needs of each group and individual readers, providing a basis for optimizing book recommendations and borrowing services.

3. The method for dynamic library management based on artificial intelligence according to claim 2, characterized in that: The specific steps of analyzing the reader's interest preference characteristics and reading habit characteristics in S1 are as follows: The term frequency-inverse document frequency formula is used to measure the reader's interest in different topic words. Suppose the reader's text data set is D, the term frequency of term t in document d is tf(t, d), the number of documents containing term t is df(t), and the total number of documents is N. Then the TF-IDF value of the term for the reader is By calculating the TF-IDF value of each topic-related word, an interest preference vector is obtained for subsequent clustering analysis and portrait construction to reflect the readers' interest weights on different topics. When analyzing reading habits, the average borrowing period C is calculated based on borrowing and browsing records. a and the standard deviation of borrowing frequency σ f Average loan period Where n is the number of borrowings, F q To count browsing frequency, D p is the browsing depth, and the reading habit feature vector F is obtained h ={C a , F q , D p …}; Standard deviation of reading frequency where f i It represents the reciprocal of the time interval between the i-th borrowing and the last borrowing. is the average borrowing frequency; the standard deviation is used to reflect the stability of readers' borrowing frequency, and together with the average borrowing period, it describes reading habits and identifies different types of readers: regular borrowers and random borrowers.

4. The method for dynamic library management based on artificial intelligence according to claim 1, characterized in that: The specific operation steps of S2 are as follows: Deploy ultra-wideband positioning tags and sensors on library shelves, and combine them with the book’s RFID tag information to track the book’s location; When a reader searches for a book through the library's mobile application, the optimal navigation path from the reader's location to the target book's location is generated using indoor positioning technology based on the reader's current location and the book's real-time location. This path is then presented to the reader in the form of a visual map on the mobile application, guiding the reader to quickly find the desired book. When generating the navigation path, a personalized recommendation mechanism is used to optimize the navigation path.

5. The method for dynamic library management based on artificial intelligence according to claim 4, characterized in that: The specific operation steps of the personalized recommendation mechanism in S2 to optimize the navigation path are as follows: Continuously collect readers' behavior data in the library, including borrowing history, movement trajectory in the library, and dwell time in front of bookshelves and browsing behavior; store them in the library's big data storage system to form a reader behavior database; pre-process the collected data, including data cleaning, data standardization and data classification; Use data analysis algorithms and machine learning techniques to conduct in-depth analysis of reader behavior data; Using cluster analysis algorithms, readers with similar borrowing behaviors and in-library activity patterns are divided into different groups; Using association rule mining algorithms, find the association relationships between different book categories and the associations between reader behavior and book interests; When a reader initiates a book query request, first search for their relevant behavioral characteristics and interest preferences in the reader behavior database and interest model based on the reader's identity information; combining the library's bookshelf layout map and real-time information on the density of people flow and the occupancy of passages, design a dynamic path planning algorithm; this algorithm is improved based on traditional path planning algorithms, taking the reader's points of interest and the areas of books they are interested in as reference factors for path planning; Calculate a comprehensive evaluation score for each candidate path, and the evaluation factors include path length, the number and weights of the interest areas passed through, and the degree of congestion of the people flow; finally, select the path with the highest comprehensive evaluation score as the navigation path recommended to the reader; During the process of the reader walking along the navigation path, continuously monitor the reader's position and behavioral changes; when the reader deviates from the predetermined path, immediately re-evaluate the path between the current position and the target book position, and generate a new navigation path based on the latest library environment information and the reader's new points of interest and push it to the reader.

6. The method of dynamic library management based on artificial intelligence according to claim 1, characterized in that: The specific operation steps of S3 are as follows: Continuously monitor and analyze the reader's borrowing behavior data, including borrowing frequency, number of overdue times, and timeliness of return, and combine the reader portrait information to establish a borrowing risk assessment model; The model uses machine learning algorithms to conduct a risk assessment of the reader's next borrowing behavior, predict the possibility of overdue or damaged books, and provide decision-making suggestions for library staff based on the assessment results; When establishing the borrowing risk assessment model, the logistic regression algorithm is used. The risk assessment value of the reader is set to P(y=1|x), where y=1 indicates the risk of overdue or damaged books, and x is the feature vector. The logistic regression formula is: Where w is the weight vector and b is the bias term; the feature vector x contains the standardized values ​​of borrowing frequency, overdue times, average borrowing period, and interest preference features; by optimizing w and b through training data, the model can accurately predict the reader's borrowing risk probability and provide a quantitative basis for decision-making; For the decision to increase the borrowing quantity for low-risk readers, determine the new borrowing quantity L according to the risk assessment value and the current borrowing limit L0; let the increase ratio coefficient be α, when P < P0, P0 is the low-risk threshold, L = L0(1 + α(1 - P)), and dynamically adjust the borrowing quantity according to the risk level; For high-risk readers, require guarantee measures or shorten the borrowing period, and set different guarantee amounts or shortening ratios according to the risk value.

7. The method of dynamic library management based on artificial intelligence according to claim 1, characterized in that: The specific operation steps of S4 are as follows: Based on the reader portrait and the classification, popularity, and borrowing history data of books, use collaborative filtering algorithms or content-based recommendation algorithms to generate a personalized book recommendation list for the reader; In the collaborative filtering algorithm, the Pearson correlation coefficient formula is used to measure the similarity of interests between readers; suppose that the ratings of readers u and v for n books are r and ui and r vi , and their average scores are and Then the Pearson correlation coefficient of readers and is By calculating the similarity between readers, we can find the neighbor readers who are most similar to the target reader, and generate a recommendation list based on the borrowing history and ratings of the neighbor readers, thus improving the accuracy and pertinence of the recommendation. The reader browses the recommended books on the mobile application and directly conducts reservation borrowing operations; according to the reservation situation and the in-shelf status of the books, automatically arrange the reserved in-shelf positions of the books, and after the books are returned, notify the reader to come and borrow, realizing seamless connection of book borrowing, improving the circulation efficiency of books and the borrowing satisfaction of readers.

8. The method of dynamic library management based on artificial intelligence according to claim 1, characterized in that: The specific operation steps of S5 are as follows: When the book is approaching the due date, send a reminder to the reader by means of text message or push on the mobile application, and the reminder content includes the consequences of overdue, as well as convenient return methods and location guides; For readers who have overdue books, we use data analysis to find out the reasons for their overdue payments, including forgetfulness, business trips, and book loss. We adopt different response strategies for different reasons. For readers who forget to return books, we improve the reminder frequency and optimize the method. For readers who cannot return books on time due to unexpected circumstances, we provide online application for deferred return, and approve the application based on the reader's credit record and past performance. For lost books, we guide readers through the compensation process, provide channels for purchasing the same books and returning them to the library, and record the information in the reader's credit file for reference in subsequent borrowing management. Establish social incentives and credit disclosure mechanisms to promote on-time returns.

9. The method for dynamic library management based on artificial intelligence according to claim 8, characterized in that: The specific steps for establishing a social incentive and credit publicity mechanism in S5 to promote timely return are as follows: Introducing social features in the library's mobile app, giving readers preset points or virtual badges for returning books on time. These points can be used to redeem library benefits, including longer loan periods and priority borrowing of popular books; At the same time, a borrowing credit ranking list is set up. With the reader's authorization, their credit rating and borrowing performance are displayed to their friends or readers in the same interest group, using social pressure and incentive mechanisms to encourage readers to return books on time. For readers who have overdue payments, once their credit rating drops to a preset level, their use of preset social functions will be restricted until they return the books and improve their credit status, thereby increasing readers' attention to overdue behavior.

10. The artificial intelligence-based dynamic library management system applied to the artificial intelligence-based dynamic library management method according to any one of claims 1 to 9, characterized in that: include: Data collection module, used to collect basic information of readers, borrowing data, and book RFID and location data, to provide basic data for subsequent use; The data analysis and processing module is used to analyze readers' online data and borrowing and browsing records, extract features to build portraits, segment groups and analyze needs; it is also used to collect and process readers' in-library behavior data, analyze and mine associations, and provide support for other modules; The book positioning and navigation module is used to combine book tags and sensors to track locations, generate and optimize navigation paths based on reader queries, and update guides for book finding in real time; The borrowing risk assessment module is used to monitor and analyze borrowing data and profiles, use algorithms to build models to predict risks, and provide decision-making suggestions for staff; The intelligent recommendation and reservation borrowing module is used to generate a recommendation list based on the portrait and book data using an algorithm. Readers can make reservations for borrowing, and the system arranges reservations and notifications to achieve seamless connection; The overdue management module is used to push reminders when overdue payments are approaching, analyze the reasons for the overdue payments and take corresponding strategies. At the same time, it establishes social incentives and credit publicity mechanisms to promote on-time repayments.

Images