Library self-service borrowing service data processing method

Abstract

The invention relates to a data processing method for self-service borrowing service of a library. Constructing a personal behavior portrait of the user by using the historical borrowing record, borrowing duration, borrowing category and behavior characteristics including borrowing frequency and preference type of the user; historical data of user behaviors are analyzed through a time series data analysis technology, a user borrowing preference mode is mined, and periodic and seasonal changes are captured; fusing the historical borrowing data with spatio-temporal information including borrowing peak time periods and borrowing demand differences of different areas in a library, and constructing a multi-dimensional spatio-temporal data model; a multi-objective optimization algorithm is adopted, and dynamic resource allocation is carried out on the premise that the borrowing requirement is met; according to the real-time data of the borrowing system, the storage positions of the books are adjusted, the books are dynamically scheduled through the self-service borrowing terminal, and the load balance of each borrowing point of the library or the self-service borrowing machine is controlled; and modes in the borrowing history of the user are deeply analyzed.

Description

Technical Field

[0001] The invention relates to a data processing method, in particular to a data processing method for a library self-service borrowing service. Background Art

[0002] At present, the data processing method of the library's self-service borrowing service, although it has many innovative functions and intelligent operations, still has certain shortcomings and disadvantages, especially compared with the traditional library self-service borrowing system, the existing technology still has some limitations. Taking the Chinese invention patent "2021107079090 A self-service management of book borrowing smart library system and operation method" as an example, it can be compared and analyzed from multiple aspects to find the problems and room for improvement in data processing and user experience of the technical solution. First of all, in the smart library system, the borrowing process and the book management system rely on the interaction of smart bookshelves, smart terminals and client APPs, and use magnetic suction mechanisms, sensors and positioning chips to locate and borrow and return books. The advantage of this system is that it can realize the rapid positioning and borrowing of books, but at the level of data processing, there are still problems with the instability of information transmission and coordination between systems. Specifically, in the process of data transmission and real-time monitoring, the system needs to continuously transmit information through wireless or wired connections. However, if there is network instability or signal loss, it may cause delays or errors in data processing. For example, the borrowing records of books may not be synchronized to the cloud system in time, which will affect the accuracy of inventory and cause some book inventory information to lag behind. In addition, since it relies on the accuracy of positioning chips and magnetic suction mechanisms, if these hardware devices fail or signal interference occurs, it may affect the user's borrowing experience, especially during peak hours or when the equipment is heavily loaded, which may easily cause problems or delays in the borrowing process.

[0003] Secondly, although the patent solution realizes the management and operation of borrowing through client APP and smart terminals, it still lacks sufficient depth and flexibility in personalized services and intelligent data processing. For example, the prediction of borrowing needs and the analysis of user behavior are still in the basic stage, and it is impossible to effectively explore the real needs and borrowing habits of users. In this system, the borrowing of books mainly depends on the manual operation of users and real-time resource query, and there are not enough intelligent functions to recommend books in advance or optimize the borrowing order. Specifically, although the smart bookshelf can sense the status of books through sensors, it lacks the analysis of users' historical borrowing records and cannot provide personalized recommendations based on historical data and user preferences. Therefore, users may still need to manually browse a large number of books to find books that meet their personal needs. This method not only affects the borrowing efficiency of users, but also increases the burden of book management and data processing. Furthermore, the existing system has obvious deficiencies in abnormal behavior monitoring and risk warning. For example, although the smart terminal provides verification functions for borrowing and returning books, the monitoring ability of users' abnormal behavior is weak. The system currently relies on manual verification and simple sensor detection, and cannot realize automatic abnormal behavior identification and real-time intervention. For example, the system cannot promptly identify certain books that have been left unreturned for a long time, or cannot detect abnormal behavior of frequently borrowing popular books but not returning them on time through data analysis. In addition, the system lacks sufficient emergency mechanisms when dealing with abnormal behavior. Once data anomalies or behavioral fluctuations occur, it cannot automatically generate early warning reports or take corresponding preventive measures. Therefore, library administrators still need to spend a lot of time and energy on manual verification, which increases management costs and risks.

[0004] In addition, although the patent scheme has achieved a certain degree of resource optimization through dynamic borrowing scheduling and intelligent navigation systems, in actual operation, how to efficiently schedule books and optimize spatiotemporal resources still faces great challenges. Especially during high-demand periods, how to accurately predict demand through spatiotemporal demand functions and achieve dynamic pricing and resource scheduling remains a difficult problem. The current system design focuses more on location-based book borrowing, but lacks a comprehensive analysis of time, space and demand fluctuations. The demand during the peak borrowing period is difficult to reflect in real time in resource scheduling, which may lead to excessive book borrowing in some areas, resulting in an unbalanced load on the borrowing terminal. At the same time, there are certain limitations in the dynamic adjustment of borrowing fees and spatiotemporal demand prediction. The changes in borrowing fees are more adjusted according to the demand fluctuations during peak periods, but the system fails to fully consider multiple factors such as user behavior patterns and book categories, making it difficult to achieve a more refined pricing mechanism. Summary of the invention

[0005] The purpose of the present invention is to provide a data processing method for library self-service lending service, so as to solve some of the drawbacks and shortcomings pointed out in the background technology.

[0006] The present invention solves the above-mentioned technical problems by adopting the following technical solution, which includes the following steps:

[0007] S1. Personalized borrowing demand prediction based on intelligent user behavior modeling:

[0008] S1.1. Use the user's historical borrowing records, borrowing duration, borrowing category, and behavioral characteristics including borrowing frequency and preference type to build a user's personal behavior profile;

[0009] S1.2. Analyze historical data of user behavior through time series data analysis technology, dig out user borrowing preference patterns, and capture periodic and seasonal changes;

[0010] S2. Dynamic scheduling of book inventory based on multidimensional space-time model:

[0011] S2.1. Integrate historical borrowing data with spatiotemporal information including peak borrowing time periods and borrowing demand differences in different areas of the library to construct a multidimensional spatiotemporal data model;

[0012] S2.2, adopt multi-objective optimization algorithm to dynamically allocate resources on the premise of meeting borrowing needs;

[0013] S2.3, according to the real-time data of the borrowing system, adjust the storage location of books, dynamically dispatch books through the self-service borrowing terminal, and control the load balancing of various borrowing points or self-service borrowing machines in the library;

[0014] S3. Abnormal borrowing pattern recognition based on behavior change detection:

[0015] S3.1. In-depth analysis of patterns in user borrowing history, including borrowing time, frequency, book type, and borrowing cycle, to establish a behavioral normality model;

[0016] S3.2, using a change point detection algorithm to detect abnormal fluctuations or deviations from normal behavior patterns in user borrowing behavior, and identify abnormal behaviors, including long-term retention without return, frequent borrowing of popular books but not returning them;

[0017] S3.3. After detecting abnormal behavior, generate a risk warning report and provide targeted response measures through intelligent reasoning analysis, including freezing accounts and reminding administrators to handle;

[0018] S4. Dynamic pricing mechanism during borrowing peak period based on adaptive spatiotemporal optimization:

[0019] S4.1. Identify peak borrowing behavior through data analysis, identify the demand for various books in different time periods, and build a spatiotemporal demand model;

[0020] S4.2. Adopt dynamic pricing strategies for books with high demand or high borrowing frequency, including increasing borrowing fees to regulate demand, or reducing fees for low-demand books to encourage borrowing;

[0021] S5. User behavior verification based on intelligent trust model:

[0022] S5.1. Build a behavioral trust model based on user borrowing behavior, including borrowing history, return timeliness, and book damage;

[0023] S5.2. Based on the reliability and consistency of user behavior, a scoring mechanism is used to conduct dynamic credit assessment of users and update the trust value in real time.

[0024] Furthermore, the abnormal borrowing pattern recognition method based on behavior change detection includes:

[0025] The model is built through a weighted scoring function. The goal is to comprehensively measure the user's borrowing behavior through multiple features including borrowing frequency, book category, and borrowing period. The total score B(u) is used to represent the behavioral characteristics of user u.

[0026]

[0027] Where: B(u) is the total borrowing behavior score of user u, which is calculated by weighting the features of each dimension; f i is the value of the i-th feature, indicating the value of user u in a certain behavior dimension, including borrowing frequency; c i is the value of the i-th feature, indicating the user's borrowing preference in a certain category, which is the borrowing volume of a specific book category including novels and science books; p i is the value of the i-th feature, indicating the user's borrowing period; r i is the value of the i-th feature, indicating the deviation of the return time, reflecting whether the user returns the book on time; w1, w2, w3, w4 are the weight coefficients of the corresponding features; n is the number of features, indicating the number of different dimensions included in the behavior model.

[0028] Furthermore, the abnormal borrowing pattern recognition method based on behavior change detection includes:

[0029] Change point detection is introduced to monitor changes in user behavior. When user borrowing behavior fluctuates drastically, abnormal behavior is discovered in a timely manner, triggering warnings or further intervention. By calculating the change in behavior score C(u,t), behavioral deviations are identified.

[0030] C(u,t)=|B(u)-B(u,t)| 2

[0031] Where: C(u,t) is the change point detection value, which is used to measure the difference between user u's borrowing behavior at time t and past behavior; if the value is greater than the preset threshold, it means that the behavior is abnormal and intervention is required; B(u) is the normal behavior score of user u, that is, the score of his borrowing behavior under normal circumstances; B(u,t) is the borrowing behavior score of user u at time point t; |·| 2 It is the square of the Euclidean distance, which indicates the change in the behavior score. If the behavior score B(u,t) is significantly different from the normal behavior score B(u), it indicates that the behavior is abnormal.

[0032] Furthermore, the abnormal borrowing pattern recognition method based on behavior change detection includes:

[0033] Introducing dynamic risk assessment to quantitatively analyze user behavior through a behavior anomaly scoring mechanism. The higher the score, the more abnormal the behavior. The management strategy for users is adjusted based on this score. The score is based on multi-dimensional behavioral characteristics, and the deviation between each characteristic and the expected value is integrated.

[0034]

[0035] in:

[0036] R(u) is the abnormality score of user u, which is used to evaluate the abnormality of user behavior. The higher the score, the more abnormal the behavior and the greater the risk. i is the value of the i-th feature, including the user's borrowing frequency; p i is the value of the i-th feature, including the user's borrowing period; μ i is the expected value of the i-th feature, indicating the expected value of the feature under normal circumstances; μ p is the expected value of the borrowing cycle feature, which represents the normal borrowing cycle of the user; α i ,β i is the weighting coefficient of each feature, which controls the contribution of each feature to the score; n is the number of features, indicating the feature dimension used to calculate the score.

[0037] Furthermore, the abnormal borrowing pattern recognition method based on behavior change detection includes:

[0038] Use spatiotemporal analysis to optimize resource scheduling and borrowing processes based on the temporal and spatial distribution of user borrowing; dynamically adjust library resource allocation during peak demand periods and popular areas by combining spatiotemporal utility functions;

[0039]

[0040] Among them, E(u,t,s) is the spatiotemporal utility function, which represents the borrowing demand utility of user u at a specific time t and space s; B(u) is the borrowing behavior score of user u, which represents the user's behavioral feature score and depends on the borrowing frequency, cycle, and category; is the indicator function, at time t i and the spatial region s j If the user borrows a book, the function is 1, otherwise it is 0. It is used to indicate whether the user behavior occurs under specific time and space conditions. time ∫ space is a double integral, representing the combined summation over the time and space dimensions.

[0041] Furthermore, the method of the dynamic pricing mechanism during the borrowing peak period based on adaptive spatiotemporal optimization is as follows:

[0042] A model based on spatiotemporal demand analysis is used to dynamically identify and predict borrowing demand in different time periods and spatial regions. Through real-time data analysis, the spatiotemporal variation of borrowing behavior is captured and applied to resource scheduling. The spatiotemporal demand function is used:

[0043]

[0044] Among them, D(t,s) is the demand for books in the time interval [t0,t1] and the space interval [s0,s1]; here D(t,s) represents the total borrowing demand in a specific time period and spatial location; f(t,s) is a weight factor, which reflects the impact of changes in time and space on borrowing demand; the weight factor is set according to different time characteristics including weekdays, weekends and holidays, and different spatial characteristics including the popularity of different areas of the library; B(t,s) is the borrowing volume at a specific time and space point, indicating the number of borrowing behaviors at a specific time and place.

[0045] Furthermore, the method of the dynamic pricing mechanism during the borrowing peak period based on adaptive spatiotemporal optimization is as follows:

[0046] Introduce a dynamic pricing strategy to adjust borrowing fees in real time according to fluctuations in borrowing demand, increase borrowing fees during high-demand periods or for popular books, and reduce borrowing fees during low-demand periods or for unpopular books; and the dynamic pricing adjustment formula is:

[0047] P(t,s)=α·D(t,s)+β

[0048] Among them, P(t,s) is the borrowing fee at time t and space s; the borrowing fee is dynamically calculated based on the spatiotemporal demand; α is the adjustment factor, which controls the relationship between the borrowing fee and the demand; specifically, when the demand D(t,s) increases, P(t,s) will also increase accordingly; the value of the adjustment factor α determines the sensitivity of the borrowing fee to demand fluctuations; β is the benchmark fee, that is, the basic fee under normal demand conditions; this fee is the standard borrowing fee when there is no special demand fluctuation; D(t,s) is the output of the spatiotemporal demand function, which represents the borrowing demand at a certain point in time and space.

[0049] Furthermore, the method of the dynamic pricing mechanism during the borrowing peak period based on adaptive spatiotemporal optimization is as follows:

[0050] By introducing intelligent demand forecasting and combining time series forecasting algorithms to predict future demand based on historical borrowing data, time trends and spatial distribution, the demand forecasting and resource scheduling model:

[0051]

[0052] in, is the predicted demand, which represents the borrowing demand in the future time and space interval; i is the prediction factor weight, which reflects the contribution of each feature to demand forecasting; including γ i Weighted based on historical borrowing frequency, seasonal changes, and holidays; i (t) is the i-th time-related characteristic function, which represents the demand changes in the time dimension including holidays, working days, and the beginning of the semester; g i (s) is the i-th spatial characteristic function, which represents the demand change of the hotspot location including the library area in the spatial dimension; t0, t1 are time intervals, representing the prediction period; s0, s1 are spatial intervals, representing the predicted spatial range.

[0053] The data processing method for the library self-service borrowing service proposed in the present invention combines intelligent user behavior modeling, abnormal borrowing pattern recognition, spatiotemporal optimization scheduling and dynamic pricing mechanism, which can significantly improve the efficiency, user experience and resource management capabilities of the library self-service borrowing system, and has many beneficial effects:

[0054] The dynamic scheduling of book inventory based on the multi-dimensional space-time model can achieve the reasonable allocation and optimal scheduling of library resources. Combined with the demand of peak borrowing periods and popular areas, book resources can be reasonably allocated during high-demand periods to avoid excessive concentration or waste of resources in a certain space-time area. By real-time monitoring of borrowing data, the balance of books between different areas and borrowing points can be ensured, and the resource utilization efficiency of the library can be improved.

[0055] Through personalized borrowing demand prediction based on intelligent user behavior modeling, the system can accurately analyze and predict each user's borrowing preferences and needs, provide customized book recommendations and borrowing tips, and enhance the user experience. Through time series data analysis, it can capture the cyclical and seasonal changes in user behavior, making library services more in line with user needs, improving borrowing rates and user satisfaction.

[0056] The system introduces an abnormal borrowing pattern recognition method based on behavior change detection. Through the weighted scoring function and change point detection algorithm, it can accurately monitor and identify abnormal fluctuations in user behavior. For abnormal behaviors such as long-term failure to return books, frequent borrowing of popular books but not returning them, the system can promptly identify and generate risk warning reports, and take measures such as freezing accounts and reminding administrators to effectively reduce the loss and risk of book resources and protect the interests of the library.

[0057] Through the dynamic pricing mechanism during borrowing peak periods based on adaptive spatiotemporal optimization, the system can flexibly adjust borrowing fees during high-demand periods or for popular books, regulate demand, and avoid excessive concentration of resources. At the same time, borrowing fees can be reduced during periods of low demand or for unpopular books to stimulate users to borrow, thereby better balancing borrowing demand and library resources and achieving dual optimization of resource scheduling and fee collection. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 The present invention is a flow chart of a data processing method for a library self-service borrowing service.

[0059] Figure 2 This is a flow chart of the abnormal borrowing pattern recognition method based on behavior change detection of the present invention.

[0060] Figure 3 The present invention is a method flow chart of a dynamic pricing mechanism during borrowing peak period based on adaptive spatiotemporal optimization. DETAILED DESCRIPTION

[0061] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0062] Combined with Figure 1Flowchart, a data processing method for library self-service borrowing service, S1 stage involves personalized borrowing demand prediction based on intelligent user behavior modeling. The core of this method is to build a personalized behavior profile of each user by analyzing the user's historical borrowing records, borrowing duration, borrowing category and other information, combined with behavioral characteristics such as borrowing frequency and preference type. The first step in this process is to collect detailed borrowing data of users, including borrowing time, type of books, borrowing frequency and borrowing duration, etc. Through these data, the user's behavior pattern and interest preferences can be fully portrayed. For example, some users may tend to borrow a certain type of books on holidays, while other users may have the habit of borrowing specific textbooks in a specific time period (such as before an exam). The key to building a behavioral profile is to convert this data into a structured model and analyze this data through an algorithm to identify the user's borrowing pattern. Secondly, time series data analysis technology is applied to the in-depth mining of user historical data. Using time series analysis methods, the system can reveal the periodic and seasonal changes in user behavior. Cyclic changes may refer to the borrowing behavior of users in a specific time period, such as an increase in the number of books borrowed at the beginning of each year, while seasonal changes may be related to external factors such as holidays and the start of semesters. Through these analyses, the system can identify users' borrowing preferences and generate a dynamic model that adapts to changes in user behavior, thereby providing accurate data support for personalized borrowing demand predictions. The innovation of this method is that through intelligent user behavior modeling, changes in user behavior can be captured in real time, and the model has high adaptability and can be continuously updated and optimized over time to ensure that the prediction results are consistent with the latest changes in user behavior.

[0063] The dynamic scheduling of book inventory based on the multidimensional spatiotemporal model aims to achieve efficient scheduling of library resources through in-depth analysis of borrowing history data and spatiotemporal information. In this stage, the borrowing history data is first combined with spatiotemporal information, which includes the peak borrowing time period and the differences in borrowing demand in different areas of the library. The borrowing history data provides the library with past borrowing behavior patterns, including the borrowing volume, borrowing category, and borrowing frequency of books in a specific time and space; the spatiotemporal information includes the time dimension of borrowing (such as weekdays and holidays, the beginning and end of the semester, specific seasons, etc.) and the spatial dimension (such as the differences in borrowing demand in different areas or bookshelves).

[0064] By integrating these data, a multidimensional spatiotemporal data model is constructed, which can accurately capture the borrowing demand of various types of books in the library at different times and in different regions. The multidimensional spatiotemporal model can not only analyze the borrowing volume at a certain point in time or in a region, but also predict the borrowing peak and resource scarcity that may occur in the future by comparing data in different time and space dimensions. Based on this model, the system can use a multi-objective optimization algorithm to dynamically schedule the book inventory. The application of the multi-objective optimization algorithm means that while meeting the borrowing needs of users, it is also necessary to consider the balance of resources and the load balancing of borrowing points. Specifically, this algorithm will comprehensively consider multiple goals, such as maximizing the satisfaction of borrowing needs, minimizing the waste of resources in the library, and balancing the load of each self-service borrowing machine or borrowing point to achieve efficient and fair resource allocation. The algorithm dynamically adjusts the storage location of books based on real-time borrowing data and system feedback to optimize the distribution of books in the library. During peak borrowing hours, the system may move popular books to a location close to the self-service borrowing machine or borrowing point to shorten the waiting time for borrowing;

[0065] During low-demand periods, the system can move low-demand books to a more distant area to optimize space usage. The self-service borrowing terminal plays a key role in this process. It is not only responsible for providing a borrowing operation interface, but also can receive inventory and location adjustment instructions in real time, and perform dynamic scheduling tasks to ensure that the distribution of books matches the borrowing demand. In addition, the load balance of the borrowing points or self-service borrowing machines in the library is also controlled. By monitoring the load of each borrowing point in real time, the system can flexibly adjust the resource allocation of the borrowing point to avoid overloading some borrowing points while other borrowing points are idle.

[0066] Abnormal borrowing pattern recognition based on behavioral change detection aims to identify and respond to possible abnormal borrowing behaviors by deeply analyzing the user's borrowing history data. At this stage, it is necessary to first conduct a detailed analysis of the user's borrowing behavior, especially multi-dimensional data such as borrowing time, borrowing frequency, book type, and borrowing cycle. These data help build a behavioral normality model, which can define a typical user's "normal" borrowing pattern, that is, regular borrowing behavior in different time periods, different book categories, and borrowing cycles. For example, some users may borrow books on weekends and holidays, while others may be accustomed to borrowing textbooks at the beginning of the semester. Through in-depth analysis of these behavioral characteristics, the system can establish a personalized normal behavior pattern for each user.

[0067] Next, a change point detection algorithm (such as CUSUM or PELT algorithm) is used to analyze the time series data of user borrowing behavior to identify deviations from normal behavior patterns. The core of the change point detection algorithm is to monitor changes in user behavior in real time and promptly detect borrowing behaviors that do not conform to normal patterns. For example, when a user's borrowing frequency suddenly increases, or the borrowing time is extended, or a book is frequently borrowed but not returned on time, the change point detection algorithm can detect these abnormal fluctuations and mark them as potential abnormal behaviors. The purpose of this process is to identify behaviors such as books that have been left unreturned for a long time or popular books that have been frequently borrowed but not returned, which are common high-risk behaviors in library management. After detecting these abnormal behaviors, the system will automatically generate a risk warning report, summarizing detailed information on the abnormal behavior, such as the degree of deviation from the behavior pattern, the time window when the abnormality occurred, and the related books.

[0068] At this point, the system will not only provide data-based warnings, but will also conduct intelligent reasoning analysis to determine the potential risks that abnormal behavior may bring, and provide corresponding countermeasures based on the risk level. For example, for users who frequently borrow but do not return books, the system may suggest freezing the user's account and restricting their further borrowing rights; or the system will remind the administrator to manually intervene in the user to further verify their borrowing behavior and take appropriate management measures. In addition, intelligent reasoning analysis can also propose personalized response plans based on the type of abnormal behavior. For example, in the case of long-term detention and non-return, the system may suggest urging users to return books by sending reminder notifications, setting warning mechanisms, or imposing borrowing restrictions.

[0069] The dynamic pricing mechanism for borrowing peak period based on adaptive spatiotemporal optimization aims to optimize resource allocation and user borrowing behavior by adopting dynamic pricing strategies through intelligent analysis of library borrowing data and spatiotemporal demand patterns. At this stage, it is first necessary to identify the changing patterns of borrowing peak period and borrowing behavior through data analysis. This process involves the collection and processing of a large amount of borrowing data. It mainly analyzes the borrowing demand in different time periods (such as weekdays, weekends, holidays, etc.) and different book categories to identify which books are in high demand in a specific time period and which books are in low demand in certain time periods. The spatiotemporal demand model can effectively reflect the changing trends and demand fluctuations of borrowing behavior by combining data in time and space dimensions. For example, some books may have a surge in demand on the eve of an exam, but less demand in other time periods. The library can identify this phenomenon through the spatiotemporal demand model and then predict future borrowing peaks and resource-constrained periods.

[0070] The key to building this model is to accurately identify the peak borrowing time periods and demand curves of different books through comprehensive analysis of historical borrowing data, time series analysis, and spatial data, so as to predict which time periods and which books will face high demand pressure. Based on this demand model, the library can implement dynamic pricing strategies during high-demand periods and regulate borrowing demand by adjusting borrowing fees. During periods of high demand or for books with high frequency of borrowing, the system will automatically increase borrowing fees, and suppress excessive borrowing and excessive concentration of resources through price mechanisms. For example, on the eve of an exam, the borrowing demand for certain reference books will increase sharply. At this time, by increasing the borrowing fees of these books, the borrowing pressure can be effectively reduced and resource tension and retention can be avoided. For books with low demand, the library can reduce borrowing fees or offer discounts to encourage users to borrow these books, thereby reducing the idleness and backlog of these books.

[0071] The core purpose of the dynamic pricing strategy is to balance the supply and demand of library resources through a flexible pricing mechanism, improve the efficiency of resource utilization, guide users' borrowing behavior through price leverage, and optimize the library's resource allocation. This strategy can not only cope with resource pressure during high-demand periods, but also stimulate users' borrowing interest during low-demand periods, ensuring efficient use of library resources.

[0072] User behavior verification based on intelligent trust model aims to establish a dynamic behavior trust model through in-depth analysis of user borrowing behavior, and then effectively verify and evaluate user borrowing behavior. At this stage, it is necessary to first build a behavior trust model based on user borrowing behavior. The construction of the model depends on multiple key indicators, including user borrowing history, timeliness of returning books, damage to books, etc. Borrowing history reflects information such as the types of books borrowed by users in the past, borrowing frequency, and borrowing duration. These data provide a basis for evaluating user borrowing behavior. Timeliness of return is another important factor, indicating whether the user returns the book on time. Users who return on time are considered to have a high level of behavioral credibility, while users who return late will have their trust value reduced. The damage of books is also a key indicator for trust evaluation. If the books borrowed by the user are obviously damaged or lost when they are returned, the system will automatically record the user's improper behavior and reduce his trust value. By combining these multi-dimensional data, the system can perform dynamic credit evaluation based on the reliability and consistency of user behavior.

[0073] The behavioral trust model uses a scoring mechanism to quantify the trust value of users. The scoring mechanism scores users based on the performance of each behavior (such as timely return, high frequency of borrowing, intact books, etc.). For example, each time a user returns a book on time, a certain trust score will be increased, while each overdue return will be deducted. The damage of the book will also affect the score. Severe damage will result in a large deduction of points, while minor damage will have a smaller impact. The scoring mechanism combines data from multiple dimensions, so that the user's trust value can reflect the quality of his borrowing behavior in real time. Over time, the trust value will be dynamically updated according to the user's behavior, and the system will continuously adjust the user's credit assessment based on the new borrowing behavior. Such a real-time update mechanism enables the library to judge the user's credit status based on the latest borrowing behavior and take corresponding management measures in a timely manner. For example, when the user's trust value is lower than a certain threshold, the system may restrict his borrowing rights, or add additional verification measures when the user borrows, such as requiring identity proof or collecting a certain deposit; for users with higher trust values, the system can provide more relaxed borrowing conditions, such as extending the borrowing period and exempting overdue fines. The introduction of this intelligent trust model makes the library's borrowing management more personalized and intelligent, and can provide customized services based on the user's credit status. It also improves the system's ability to identify and manage abnormal borrowing behavior. Different from traditional borrowing management methods, behavior verification based on intelligent trust models can dynamically adjust the trust assessment of users, avoid overly rigid management methods, and improve the management efficiency of library resources and user experience.

[0074] Embodiment 1:

[0075] Combined with Figure 2 Flowchart, in this embodiment, the abnormal borrowing pattern recognition method based on behavior change detection uses a weighted scoring function to comprehensively measure the user's borrowing behavior characteristics to achieve the identification of abnormal borrowing behavior. The implementation steps of the method will be illustrated by specific examples, and actual data will be introduced to calculate and verify the feasibility of each step. Suppose we are analyzing the borrowing behavior of user "Zhang San" and trying to identify whether he has an abnormal borrowing pattern. Suppose that the borrowing behavior data of "Zhang San" has been collected, and the specific behavioral characteristics include borrowing frequency, book category preference, borrowing cycle, and return time deviation. In order to perform scoring modeling, we first set the weight coefficient and characteristic value of each feature, and use the weighted scoring formula to calculate the total score B(u) to measure the normality of its borrowing behavior.

[0076] The borrowing behavior data of user "Zhang San" is extracted by the system and quantified into feature data of multiple dimensions, including:

[0077] Borrowing frequency (f i): reflects the frequency of borrowing books by a user in a certain period of time. Assume that "Zhang San" has borrowed 12 books in the past 30 days, and the borrowing frequency is 0.4 (i.e., one book is borrowed every 3 days).

[0078] Book category preference (c i ): reflects the book categories preferred by users, and counts the borrowing volume of a specific category. Assume that "Zhang San" likes to borrow novels, and has borrowed 8 novels, 2 science books, and 2 history books in the past 30 days. Therefore, the borrowing volume of the novel category is 8, and the preference for science books and history books is lower.

[0079] Loan period (p i ): reflects the time span of a user borrowing books. Assume that "Zhang San" is used to borrowing 4 books per month, and usually borrows a new book around the 10th day of each month.

[0080] Return time deviation (r i ): reflects whether the user returns the books on time. Assume that in the past 30 days, "Zhang San" borrowed 4 books, 1 of which was not returned on time and was delayed by 3 days, and the other 3 books were returned on time.

[0081] We assign corresponding values ​​to each feature based on the user's borrowing behavior data:

[0082] Borrowing frequency f1 = 0.4

[0083] Book category preference (novels) c1 = 8 (the number of borrowed novels is 8)

[0084] Borrowing period p1 = 4 (borrowing 4 books per month)

[0085] Return time deviation r1 = 0.25 (because one book was returned 3 days late, and the other books were returned on time)

[0086] In this method, weight coefficients w1, w2, w3, w4 are used to indicate the importance of each feature in the final score. To calculate the user's total behavior score, we assign a weight coefficient to each feature. Suppose we use the following weight distribution:

[0087] The weight of borrowing frequency w1 = 0.3

[0088] The weight of book category preference w2 = 0.2

[0089] The weight of the borrowing period w3 = 0.25

[0090] The weight of the return time deviation w4 = 0.25

[0091] The selection of these weight coefficients is based on the analysis of the impact of different features on user borrowing behavior. For example, borrowing frequency and return time deviation are usually the most critical indicators, so these two features have higher weights.

[0092] According to the formula:

[0093]

[0094] Where B(u) is the total borrowing behavior score of user u, f i ,c i ,p i , and r i are the values ​​of borrowing frequency, book category preference, borrowing period and return time deviation respectively, w1, w2, w3, w4 are the weight coefficients of the corresponding features, and n is the number of features.

[0095] Substituting the data of "Zhang San", we can calculate his behavior score:

[0096] B(u)=(0.3·0.4)+(0.2·8)+(0.25·4)+(0.25·0.25)

[0097] B(u)=0.12+1.6+1.0+0.0625=2.7825

[0098] Therefore, the total behavior score of "Zhang San" is B(u)=2.7825.

[0099] Through the calculated score, we can evaluate the user's borrowing behavior. If the score is lower than a certain threshold (for example, 3.0), the user's behavior may be abnormal, and the system will issue a warning or mark the user as a high-risk user. Assume that when the threshold we set is 3.0, "Zhang San" scores 2.7825, which is lower than the threshold, indicating that he may have abnormal borrowing behavior, such as leaving books unreturned, frequently borrowing popular books and not returning them, etc.

[0100] At this point, the system will further analyze the changes in the user's borrowing pattern through a change point detection algorithm to identify whether there is a risk of deviation from normal behavior. For example, if the borrowing frequency of "Zhang San" has increased sharply in recent period of time, or the borrowing volume of a certain category of books (such as textbooks) has increased significantly, it may trigger the early warning system, further manual intervention or increased monitoring of the user.

[0101] Based on the changes in user scores and warnings, the library's self-service borrowing system will take appropriate measures. For example, for users with low scores, the system can require them to pay a deposit or restrict borrowing of books in specific categories; for users with high scores, the system may provide more borrowing privileges, such as extending the borrowing period or exempting some overdue fines. This dynamic adjustment strategy can effectively manage borrowing behavior and reduce the impact of abnormal borrowing behavior on library resources.

[0102] Suppose that in a certain data analysis, the system calculates in real time that the borrowing behavior score of "Zhang San" is 2.78, which is lower than the set threshold of 3.0. Therefore, the system triggers the abnormal behavior detection mechanism, and combines the user's borrowing history and return time deviation to determine that he may not return the books for a long time. Further analysis shows that "Zhang San" has failed to return borrowed books on time twice in the past three months, and the borrowing frequency is increasing month by month. Therefore, the system decides to restrict his borrowing rights until he returns the borrowed books and improves his borrowing behavior.

[0103] Assume that in the past borrowing history of "Zhang San", the system has calculated his normal behavior score B(u) and continuously tracked his borrowing behavior changes during the monitoring process. Change point detection helps us identify which behavior changes exceed the normal range by calculating the change in behavior score C(u,t), thereby triggering early warning or intervention measures.

[0104] Step 1: Calculate the user's normal behavior score B(u)

[0105] Before change point detection, we first need to calculate the normal behavior score B(u) for each user. This score combines multiple behavioral characteristics such as borrowing frequency, book category preference, borrowing cycle, and return time deviation, reflecting the user's typical borrowing pattern. As mentioned above, the borrowing behavior score of "Zhang San" B(u) = 2.7825, which is calculated based on his borrowing history data.

[0106] Step 2: Monitor user behavior in real time and calculate the behavior score B(u,t) at time point t

[0107] Next, we monitor the borrowing behavior of user "Zhang San" in real time and calculate his borrowing behavior score B(u,t) at a specific time point t. Assume that at time point t (for example, in the borrowing cycle one month later), "Zhang San" borrows more popular books and the borrowing frequency increases to 2 books every 3 days. His book category preference has changed, borrowing more science books instead of fiction books, and the return time deviation has increased, and the number of books not returned on time has reached 2.

[0108] At this time, the calculated behavior score B(u,t) is:

[0109] B(u,t)=(0.3·0.67)+(0.2·12)+(0.25·6)+(0.25·0.5)=0.201+2.4+1.5+0.125

[0110] =4.226

[0111] Therefore, the user's behavior score at time point t, B(u,t)=4.226, shows a significant fluctuation compared to the normal behavior score, B(u)=2.7825.

[0112] Step 3: Calculate the behavior change C(u,t)

[0113] Through the change point detection formula, we calculate the behavior change C(u,t) of user "Zhang San" at time point t:

[0114] Cu,t=|Bu-Bu,t| 2 =|2.7825-4.226| 2 =|-1.4435| 2 =2.083

[0115] According to the formula, the behavior change C(u,t)=2.083, which means that the user's behavior score has changed significantly at time point t.

[0116] Step 4: Determine behavioral abnormalities and trigger intervention

[0117] In order to identify whether there is abnormal behavior, we need to set a threshold δ. If the calculated behavior change C(u,t) exceeds the threshold, it is considered that the user's borrowing behavior has deviated significantly and intervention is required. Assuming that we set the threshold δ=1.5, since C(u,t)=2.083 is greater than the threshold 1.5, the system will determine that the user's borrowing behavior has fluctuated abnormally and intervention is required.

[0118] Step 5: Implementation of interventions

[0119] When abnormal behavior is detected, the system can take different intervention measures. For example, for the abnormal fluctuation of "Zhang San", the system can do the following:

[0120] Since Zhang San's borrowing behavior is significantly different from his historical pattern, the system can temporarily freeze his borrowing account and require him to return the books that were not returned on time before he can continue to borrow. If the system detects abnormal behavior of the user, the administrator will receive a warning and can manually review the user's borrowing record to determine whether there is malicious behavior (such as frequently borrowing popular books without returning them). The system can also send notifications to users to remind them of the inconsistency of their recent borrowing behavior, require users to improve their return habits, and set restrictions on future borrowing.

[0121] Assume that in the past month, Zhang San frequently borrowed popular books but did not return them on time, which led to a significant increase in his borrowing frequency and a change in the regularity of his borrowing cycle. Through change point detection, the system detects the deviation of Zhang San's borrowing behavior from the normal mode in real time, and calculates the behavior change to be 2.083, which exceeds the preset threshold of 1.5. The system immediately initiates intervention measures: freezing Zhang San's borrowing rights and notifying the administrator. The administrator found through manual review that Zhang San had borrowed 10 popular books in the past two months, 3 of which were not returned on time, and the types of books he borrowed tended to be academic rather than novels. This information helped the administrator understand the changes in Zhang San's borrowing behavior, and finally decided to set a new borrowing rule for him: he can only borrow 3 books per month and is required to pay a certain deposit when returning the books.

[0122] Assuming that the borrowing behavior of user "Zhang San" has fluctuated in the past few months, we will use a dynamic risk assessment mechanism based on the user's multi-dimensional behavioral characteristics to calculate his abnormal score R(u) and take appropriate management measures based on the score.

[0123] First, we need to define the scoring function and each behavior feature. In this scheme, the scoring function comprehensively considers the user's borrowing frequency f i and loan period p i , and compare it with the expected value μ under normal circumstances i and μ p Compare and calculate the deviation of each feature. The formula is as follows:

[0124]

[0125] Where R(u) is the abnormal score of user u, indicating the abnormal degree of his borrowing behavior; f i and p i are the borrowing frequency and borrowing period of the user respectively; μ i and μ p are the expected values ​​of these features, namely the borrowing frequency and period under normal circumstances; α i and β iis the weighting coefficient of the feature, which is used to control the impact of each feature on the score; n is the number of features, indicating how many features are used to calculate the score.

[0126] Assume that the borrowing behavior of "Zhang San" in the past few months has the following data:

[0127] Borrowing frequency f i :The user's borrowing frequency in the past three months is 8 books per month, and under normal circumstances, the user's borrowing frequency expectation μ i Borrow 6 books per month.

[0128] Loan period p i :The average borrowing period for users in the past three months was once every 4 days, while under normal circumstances, the expected borrowing period μ p The loan period is every 5 days.

[0129] Based on these data, we can deduce the deviation of user "Zhang San":

[0130] Borrowing frequency deviation: |f i μ i |=|86|=2

[0131] Borrowing cycle deviation: |p i μ p |=|45|=1

[0132] In this scoring mechanism, we assume that the weighting coefficients we set are as follows:

[0133] α i =0.7: Borrowing frequency contributes more to the score, so a higher weight is set.

[0134] β i =0.3: The loan period contributes less to the score, so a lower weight is set.

[0135] Therefore, the calculation formula of the anomaly score R(u) is:

[0136] R(u)=0.7·|86|+0.3·|45|=0.7·2+0.3·1=1.4+0.3=1.7

[0137] Through calculation, the abnormal score R(u) of user "Zhang San" is 1.7, which indicates that his borrowing behavior is relatively large compared with normal borrowing behavior. In order to determine whether the user's behavior is abnormal, the system needs to set a threshold T. Suppose we set the threshold T to 1.5. If the user's abnormal score R(u) is greater than the threshold, it indicates that his borrowing behavior is abnormal and needs corresponding risk assessment and intervention.

[0138] In this example, the score of user "Zhang San" is 1.7, which is greater than the threshold of 1.5, so the system judges his behavior as abnormal.

[0139] Next, the system initiates a risk assessment process and takes the following actions:

[0140] 1. Risk assessment: The system will assess the severity of abnormal behavior. If the score is between 1.5 and 2, it means that the user's behavior is highly abnormal and may require further investigation and intervention; if the score is greater than 2, it means that the user may have malicious behavior or abuse the system and needs to be closely monitored.

[0141] 2. Management measures: For users with a rating of 1.7, the library can take the following management measures:

[0142] Warning: Send warning notifications to users to remind them that their borrowing frequency is abnormal and ask them to improve their borrowing behavior.

[0143] Borrowing limit: The system can limit the number of books a user can borrow, for example, limiting the number of books a user can borrow to no more than 6 per month.

[0144] Manual intervention: Administrators can further review users' borrowing records to determine whether there is any abuse of the system, such as frequently borrowing popular books but not returning them.

[0145] As the borrowing behavior of users changes, the system will dynamically adjust the management strategy. For example, if Zhang San's borrowing behavior improves in the next month and the score drops to 1.2, the system will automatically reduce the management restrictions on him and restore his normal borrowing rights. If the behavior does not improve and the score continues to remain at a high level, the system will strengthen monitoring and consider taking more stringent measures.

[0146] Assume that in the library's self-service borrowing system, user "Li Si" has a series of borrowing behavior data, and the library needs to optimize resource allocation based on this data and spatiotemporal conditions. For example, user "Li Si" usually borrows books on Tuesday and Friday afternoons, and the library's borrowing demand during these periods is often high. Therefore, the library system will use spatiotemporal utility functions to dynamically adjust resource allocation and borrowing processes.

[0147] First, we need to define the spatiotemporal utility function E(u,t,s), which is used to measure the borrowing demand utility of user u within a specific time t and spatial region s. The formula is as follows:

[0148]

[0149] Among them, E(u,t,s) represents the borrowing demand utility of user u in time t and space s; B(u) is the borrowing behavior score of user ·, which represents the comprehensive characteristics of his borrowing behavior, which depends on the borrowing frequency, cycle, category, etc. is the indicator function, at time t i and space j If the user borrows the book, the function is 1, otherwise it is 0; the double integral ∫ time ∫ space It represents the comprehensive sum of time and space dimensions, measuring the borrowing demand in different time periods and spatial areas.

[0150] Assume that user "Li Si" has the following borrowing behavior data:

[0151] Borrowing frequency f i :The user borrows 4 books per week, with a high borrowing frequency, and is a high-demand user.

[0152] Loan period p i : The interval between each book borrowing by a user is 5 days.

[0153] Borrowing category preference: Users tend to borrow literature and science and technology books, with literature accounting for 60% and science and technology accounting for 40%.

[0154] Through these data, we can calculate the borrowing behavior score B(u) of user "Li Si". Assuming the borrowing frequency f i The weight w1 = 0.5, the borrowing period p i The weight w2 = 0.3, the borrowing category c i The weight w3 = 0.2, then the total borrowing behavior score B(u) of user "Li Si" can be expressed as:

[0155] B(u)=w1·f i +w2·p i +w3·c i =0.5·4+0.3·5+0.2·0.6=2+1.5+0.12=3.62

[0156] Assume that the library's spatial area is divided into two areas: Area 1 (near the self-service lending machine) and Area 2 (near the bookshelf), and the peak borrowing hours are every Tuesday and Friday afternoon. Now, we need to calculate the demand utility of user "Li Si" in these time and space areas based on his borrowing behavior.

[0157] On Tuesday and Friday afternoons, user Li Si usually borrows books, so the indicator function It is 1 in these time periods and regions. In other time periods and regions, it is 0. We can calculate its spatiotemporal utility by double integration:

[0158]

[0159] Substituting the actual time and space data into the formula, we get the following calculation process:

[0160] For area 1 (the area where the self-service lending machine is located), on Tuesday and Friday afternoons, the spatiotemporal utility of user "Li Si" is:

[0161]

[0162] For area 2 (the area where the bookshelf is located), on Tuesday and Friday afternoons, the spatiotemporal utility of user "Li Si" is:

[0163]

[0164] Through the spatiotemporal utility function, we get the borrowing demand utility of user "Li Si" in different time and space areas. In this process, we can observe that on Tuesday and Friday afternoons, the demand utility of user "Li Si" in these two areas is higher. Therefore, the library system can dynamically adjust resource allocation according to these spatiotemporal utility values.

[0165] Embodiment 2:

[0166] Combined with Figure 3 Flowchart,In the process of self-service borrowing services in the library, combined with real-time data analysis,,it is possible to accurately identify borrowing needs in different time periods and spatial areas, and make dynamic pricing adjustments based on this,,thereby achieving reasonable resource scheduling and optimizing library resource allocation.,Next, we will explain the application of this method in detail through specific examples,,explain each step and demonstrate the calculation process.

[0167] First, the library needs to dynamically identify and predict borrowing demand in different time periods and spatial regions through a spatiotemporal demand analysis model. In order to capture the spatiotemporal variation of borrowing behavior, we need to construct a spatiotemporal demand function:

[0168]

[0169] Here, D(t,s) represents the demand for books in the time interval [t0,t1] and the space interval [s0,s1]. This function helps libraries predict which time periods and areas have more concentrated demand by calculating the total amount of borrowing demand in a specific time period and space location.

[0170] f(t,s): is a weight factor that reflects the impact of changes in time and space on borrowing demand. The weight factor depends on specific time characteristics (such as weekdays, weekends, holidays) and spatial characteristics (such as the popularity of different areas in the library).

[0171] B(t,s): is the borrowing volume, which represents the number of borrowing behaviors at a specific time and space point (i.e., a specific time t and spatial location s).

[0172] Suppose that during a given week at the library, we collected the following data:

[0173] Borrowing volume B(t,s): The borrowing volume in the self-service borrowing area (area A) and the study area (area B) of the library is:

[0174] Area A (self-service borrowing area) has a borrowing capacity of 15 books per hour;

[0175] Area B (study area) has a loan capacity of 8 books per hour;

[0176] Time and space weight factors f(t,s): Considering that the needs of libraries vary in time and space, we set the weight factors as follows:

[0177] The weight factor f of the self-service borrowing area (area A) on weekdays (Monday to Friday) A =1.2, while the weight factor f in the learning area (area B) B =1.0; weight factor f in the self-service borrowing area (area A) on weekends (Saturday and Sunday) A =1.5, weight factor f in the learning area (area B) B =1.2; weight factor f in the self-service borrowing area during holidays (such as Spring Festival, National Day, etc.) A =2.0, weight factor f in the learning area B =1.5.

[0178] We use the spatiotemporal demand function to calculate the borrowing demand in a specific time period and spatial area. Take a hypothetical weekday as an example, assuming that the time interval of the day is from 10 am to 12 pm (i.e. [t0, t1] = [10, 12]), and calculate the borrowing demand in the self-service borrowing area (area A) and the study area (area B).

[0179] 1. Calculate the demand for area A (self-service borrowing area):

[0180]

[0181] Since on weekdays, the borrowing volume in area A is 15 books per hour, and the weight factor f A =1.2, so:

[0182]

[0183] 2. Calculate the demand for area B (learning area):

[0184]

[0185] The borrowing volume of area B is 8 books per hour, and the weight factor f B =1.0, so:

[0186]

[0187] By calculating the time-space demand D(t,s), the library can determine the borrowing demand in different time periods and spatial areas. For books with high demand or high frequency of borrowing, the library can adopt a dynamic pricing strategy to regulate demand. Assume that the library's demand increases significantly on weekends (such as Saturdays), especially in the self-service borrowing area (area A). In this case, the library can adopt the following dynamic pricing mechanism:

[0188] 1. Self-service borrowing area (area A): Due to the high demand in area A, the system increases the borrowing fee based on demand forecast. Assuming the normal borrowing fee is 1 yuan per book, during high demand periods (weekends), the library will increase the borrowing fee to 1.5 yuan per book.

[0189] 2. Study area (area B): Although the demand in area B is not that high, in order to encourage users to borrow and balance the demand, the library can choose to reduce the borrowing fee. Assuming the normal borrowing fee is 1 yuan per book, during the period of low demand, the library will reduce the borrowing fee to 0.8 yuan per book.

[0190] Combined with the calculation results of the spatiotemporal demand function, the library can further optimize resource scheduling. For example, the system recognizes that during the weekend, the demand in area A increases significantly, while the demand in area B is relatively low. Therefore, the library can dispatch more books to the self-service lending area and increase the number of self-service lending machines. At the same time, the system can also dynamically adjust the display location of books based on user behavior. For example, during the peak borrowing period, the library can place popular books near the self-service lending machine to reduce the waiting time for borrowing and improve the user experience.

[0191] This embodiment introduces a dynamic pricing mechanism with adaptive spatiotemporal optimization, which aims to adjust the borrowing fee in real time according to the fluctuation of borrowing demand, so as to achieve optimal allocation of resources and balance of demand. This method calculates the borrowing fee at each spatiotemporal point based on a dynamic pricing formula by analyzing the spatiotemporal demand. The specific formula is:

[0192] P(t,s)=α·D(t,s)+β

[0193] Among them, P(t,s) is the borrowing fee at time t and space s; D(t,s) is the spatiotemporal demand function, which represents the borrowing demand at a specific point in time and space; α is the adjustment factor, which controls the relationship between the borrowing fee and the demand; β is the benchmark fee, that is, the standard borrowing fee under normal demand conditions.

[0194] First, the library needs to analyze the borrowing demand in each time period and spatial area and calculate the time and space demand D(t,s). Through real-time data collection, we can obtain the borrowing volume in different areas (such as self-service borrowing area and study area) on a certain day. For example:

[0195] Self-service borrowing area (area A): during the morning peak period (7:00-9:00) on weekdays, the borrowing volume is 20 books per hour; during the evening peak period (17:00-19:00), the borrowing volume is 25 books per hour.

[0196] Study area (area B): during the morning peak on weekdays, the number of books borrowed per hour is 10; during the evening peak, the number of books borrowed per hour is 15. We can express these demands through the spatiotemporal demand function D(t,s) and use it as the basis for subsequent pricing.

[0197] Next, the library needs to set a baseline fee β and an adjustment factor α so that borrowing fees can be adjusted appropriately when demand fluctuates.

[0198] Baseline fee β: Assume that the borrowing fee for each book is 1 yuan in the absence of demand fluctuations.

[0199] Adjustment factor α: The value of this factor determines the sensitivity of borrowing fees to demand fluctuations. Assume that the adjustment factor α ranges from 0.05 to 0.2, and the specific value depends on the library's resource management strategy and the degree of demand fluctuations.

[0200] Here, we assume that α = 0.1, that is, for every 1 unit increase in demand for borrowing fees, the borrowing fee increases by 0.1 yuan.

[0201] Based on the aforementioned time and space demand D(t,s) and the set base fee and adjustment factor, we can calculate the borrowing fee in different time periods and spatial areas. For example:

[0202] Peak period of self-service borrowing area (area A): Assume that during the morning peak period (7:00-9:00) on weekdays, the borrowing demand in area A is D A = 20 books per hour. Using dynamic pricing formula:

[0203] P A =α·D A +β=0.1·20+1=2 yuan

[0204] Therefore, during this period, the borrowing fee in the self-service borrowing area is 2 yuan.

[0205] Low period of self-service borrowing area (area A): Assume that during the noon period (12:00-14:00) on weekdays, the borrowing demand in area A is D A = 10 books per hour. Using dynamic pricing formula:

[0206] P A =α·D A +β=0.1·10+1=2 yuan

[0207] Even if the demand decreases, the borrowing fee will not drop too much due to the influence of the set adjustment factor α=0.1. This strategy helps to maintain the basic income level.

[0208] Peak period of the learning area (area B): Assume that during the evening peak period (17:00-19:00) on weekdays, the borrowing demand in area B is D B = 15 books per hour. Using dynamic pricing formula:

[0209] P B =α·D B +β=0.1·15+1=2.5 yuan

[0210] At this time, the borrowing fee in the study area rose to 2.5 yuan to adapt to the increase in demand.

[0211] Low period in the learning area (area B): Assume that during the weekday morning (7:00-9:00), the borrowing demand in area B is D B = 10 books per hour. Using dynamic pricing formula:

[0212] P B =α·D B +β=0.1·10+1=2 yuan

[0213] This pricing strategy allows the library to keep borrowing costs reasonable even during periods of lower demand.

[0214] In addition to dynamically adjusting borrowing fees, libraries can also optimize resource allocation based on calculated demand and borrowing fees. For example: 1. Increase the number of self-service borrowing machines in high-demand areas: During the morning peak period, if the self-service borrowing area (area A) shows a high borrowing demand, the library can increase the number of self-service borrowing machines to improve borrowing efficiency and reduce waiting time in line. 2. Adjust the display location of books: If certain books are in high demand during a specific period of time (such as textbooks or reference books during exam week), the library can place these books in the most easily borrowed area to facilitate users to find them quickly. In actual operations, libraries can adjust pricing and resource scheduling through real-time data analysis. For example, suppose that in a certain week, the library finds that the demand for the self-service borrowing area continues to rise, and the effect of the borrowing fee adjustment is not satisfactory. It may be necessary to appropriately increase the adjustment factor α, or set temporary peak period charging standards based on actual conditions.

[0215] Combining historical borrowing data, time trends and spatial distribution, the time series prediction algorithm is used to predict future borrowing demand, thereby rationally scheduling resources and dynamically adjusting borrowing fees. Specifically, the core formula of the demand prediction model is:

[0216]

[0217] in, is the predicted demand, which represents the borrowing demand in the future time and space interval; i is the prediction factor weight, reflecting the contribution of each feature to demand forecast; i (t) is the time-related characteristic function, which indicates the demand change in the time dimension; g i (s) is the space-related characteristic function, which indicates the demand change in the spatial dimension; t0, t1 are time intervals, which indicate the predicted time range; s0, s1 are spatial intervals, which indicate the predicted spatial range.

[0218] First, we need to define the predictor weights γ i These factors are weighted based on historical borrowing frequency, seasonal changes, holidays, etc., reflecting the contribution of different characteristics to borrowing demand. The following weight ranges are set:

[0219] γ1: Historical borrowing frequency (value range: 0.3 to 0.5)

[0220] γ2: Seasonal variation (range: 0.2 to 0.3)

[0221] γ3: Holiday factor (value range: 0.2 to 0.4)

[0222] In addition, f i (t) and g i(s) are the characteristic functions of time and space dimensions respectively. In the time dimension, f i (t) is mainly weighted according to factors such as working days, weekends, holidays, and the beginning of semesters; in the spatial dimension, g i (s) is set according to the different borrowing demands in different areas of the library, especially hot spots such as the self-service borrowing area and the study area.

[0223] To better illustrate the demand forecasting process, we take the forecast of a certain week as an example and assume that we want to forecast the borrowing demand for next week. Set t0 = 2024 / 12 / 29 (Sunday), t1 = 2024 / 12 / 30 (Monday), that is, the forecast time interval is the first day of next week. Assume that the borrowing demand in this time period in the past two years has shown regular fluctuations, and the holiday factor has a greater impact on the demand. The value of f1(t) is higher on holidays and lower on weekdays.

[0224] Assume that f1(t) reflects holiday factors and its value range is 0.5, 1.0.

[0225] Assume that f2(t) reflects the change in demand between weekdays and the beginning of the semester, and its value range is 0.3, 0.5.

[0226] Based on historical borrowing data and time series analysis, we conclude that the demand function for this time interval is:

[0227] (Holiday factors)

[0228] In the spatial dimension, we focus on the popular areas of the library, such as the self-service lending area (area A) and the study area (area B). Assumptions:

[0229] g1(s): Demand change in the self-service lending area, ranging from 0.4 to 0.6.

[0230] g2(s): Demand change in the learning area, ranging from 0.3 to 0.5.

[0231] We can calculate the space requirements for these two areas based on historical data and the popularity of each area:

[0232] (Self-service borrowing area)

[0233] (Study Area)

[0234] After obtaining the demand functions in time and space, we can use the demand forecasting formula to calculate the future borrowing demand. Assume that in this forecasting model, the weight factor of historical borrowing frequency is γ1 = 0.4, seasonal variation is γ2 = 0.3, and holiday factor is γ3 = 0.3.

[0235] We substitute each factor into the formula for calculation:

[0236]

[0237] Substitute the known data:

[0238]

[0239] Therefore, the predicted borrowing demand is 0.208. This predicted value can reflect the library's borrowing demand in this time and space interval and help the library optimize resource scheduling and pricing strategies.

[0240] Based on demand forecasts, libraries can implement dynamic pricing in different time and space areas. For example, during periods of predicted high demand (such as holidays), you can consider increasing borrowing fees to reduce demand pressure, and conversely, during periods of low demand (such as weekdays), you can moderately reduce borrowing fees. In this way, the library can respond to changing borrowing needs in real time, improve resource utilization efficiency and user satisfaction.

Claims

1. A data processing method for library self-service borrowing service, characterized in that The following steps are involved: S1. Personalized borrowing demand prediction based on intelligent user behavior modeling: S1.

1. Use the user's historical borrowing records, borrowing duration, borrowing category, and behavioral characteristics including borrowing frequency and preference type to build a user's personal behavior profile; S1.

2. Analyze historical data of user behavior through time series data analysis technology, dig out user borrowing preference patterns, and capture periodic and seasonal changes; S2. Dynamic scheduling of book inventory based on multidimensional space-time model: S2.

1. Integrate historical borrowing data with spatiotemporal information including peak borrowing time periods and borrowing demand differences in different areas of the library to construct a multidimensional spatiotemporal data model; S2.2, adopt multi-objective optimization algorithm to dynamically allocate resources on the premise of meeting borrowing needs; S2.3, according to the real-time data of the borrowing system, adjust the storage location of books, dynamically dispatch books through the self-service borrowing terminal, and control the load balancing of various borrowing points or self-service borrowing machines in the library; S3. Abnormal borrowing pattern recognition based on behavior change detection: S3.

1. In-depth analysis of patterns in user borrowing history, including borrowing time, frequency, book type, and borrowing cycle, to establish a behavioral normality model; S3.2, using a change point detection algorithm to detect abnormal fluctuations or deviations from normal behavior patterns in user borrowing behavior, and identify abnormal behaviors, including long-term retention without return, frequent borrowing of popular books but not returning them; S3.

3. After detecting abnormal behavior, generate a risk warning report and provide targeted response measures through intelligent reasoning analysis, including freezing accounts and reminding administrators to handle; S4. Dynamic pricing mechanism during borrowing peak period based on adaptive spatiotemporal optimization: S4.

1. Identify peak borrowing behavior through data analysis, identify the demand for various books in different time periods, and build a spatiotemporal demand model; S4.

2. Adopt dynamic pricing strategies for books with high demand or high borrowing frequency, including increasing borrowing fees to regulate demand, or reducing fees for low-demand books to encourage borrowing; S5. User behavior verification based on intelligent trust model: S5.

1. Build a behavioral trust model based on user borrowing behavior, including borrowing history, return timeliness, and book damage; S5.

2. Based on the reliability and consistency of user behavior, a scoring mechanism is used to conduct dynamic credit assessment of users and update the trust value in real time.

2. A data processing method for library self-service borrowing service according to claim 1, characterized in that The abnormal borrowing pattern recognition method based on behavior change detection includes: The model is built through a weighted scoring function. The goal is to comprehensively measure the user's borrowing behavior through multiple features including borrowing frequency, book category, and borrowing period. The total score B(u) is used to represent the behavioral characteristics of user u. Where: B(u) is the total borrowing behavior score of user u, which is calculated by weighting the features of each dimension; f i is the value of the i-th feature, indicating the value of user u in a certain behavior dimension, including borrowing frequency; c i is the value of the i-th feature, indicating the user's borrowing preference in a certain category, which is the borrowing volume of a specific book category including novels and science books; p i is the value of the i-th feature, indicating the user's borrowing period; r i is the value of the i-th feature, indicating the deviation of the return time, reflecting whether the user returns the books on time; w1, w2, w3, w4 are the weight coefficients of the corresponding features; n is the number of features, indicating the number of different dimensions included in the behavior model.

3. A data processing method for library self-service lending service according to claim 2, characterized in that The abnormal borrowing pattern recognition method based on behavior change detection includes: Change point detection is introduced to monitor changes in user behavior. When user borrowing behavior fluctuates drastically, abnormal behavior is discovered in time, triggering warnings or further intervention. By calculating the change in behavior score C(u, t), behavioral deviations are identified. C(u,t)=|B(u)-B(u,t)| 2 Where: C(u, t) is the change point detection value, which is used to measure the difference between user u's borrowing behavior at time t and past behavior; if the value is greater than the preset threshold, it means that the behavior is abnormal and intervention is required; B(u) is the normal behavior score of user u, that is, the score of his borrowing behavior under normal circumstances; B(u, t) is the borrowing behavior score of user u at time point t; |·| 2 It is the square of the Euclidean distance, which indicates the change in the behavior score. If the behavior score B(u, t) is significantly different from the normal behavior score B(u), it indicates that the behavior is abnormal.

4. A data processing method for library self-service borrowing service according to claim 3, characterized in that The abnormal borrowing pattern recognition method based on behavior change detection includes: introducing dynamic risk assessment to quantitatively analyze user behavior through a behavior anomaly scoring mechanism, where users with higher scores have more abnormal behavior, and the management strategy for users is adjusted according to the score; the scoring is based on multi-dimensional behavior characteristics, and the deviation between each characteristic and the expected value is integrated.

5. A data processing method for library self-service borrowing service according to claim 4, characterized in that The abnormal borrowing pattern recognition method based on behavior change detection includes: Use spatiotemporal analysis to optimize resource scheduling and borrowing processes based on the temporal and spatial distribution of user borrowing; dynamically adjust library resource allocation during peak demand periods and popular areas by combining spatiotemporal utility functions; Among them, E(u, t, s) is the spatiotemporal utility function, which represents the borrowing demand utility of user u at a specific time t and space s; B(u) is the borrowing behavior score of user u, which represents the user's behavioral feature score and depends on the borrowing frequency, cycle, and category; is the indicator function, at time t i and spatial region s j If the user borrows a book, the function is 1, otherwise it is 0. It is used to indicate whether the user behavior occurs under specific time and space conditions. time ∫ space is a double integral, representing the integral over time and space The combined sum of dimensions.

6. A data processing method for library self-service borrowing service according to claim 1, characterized in that The method of the dynamic pricing mechanism during borrowing peak period based on adaptive spatiotemporal optimization: adopts a model based on spatiotemporal demand analysis to dynamically identify and predict borrowing demand in different time periods and spatial areas; through real-time data analysis, captures the spatiotemporal variation pattern of borrowing behavior, and applies this pattern to resource scheduling.

7. A data processing method for library self-service borrowing service according to claim 6, characterized in that The method of the dynamic pricing mechanism during the borrowing peak period based on adaptive spatiotemporal optimization: Introduce a dynamic pricing strategy to adjust borrowing fees in real time according to fluctuations in borrowing demand, increase borrowing fees during high-demand periods or for popular books, and reduce borrowing fees during low-demand periods or for unpopular books; and the dynamic pricing adjustment formula is: P(t,s)=α·D(t,s)+β Among them, P(t, s) is the borrowing fee at time t and space s; the borrowing fee is dynamically calculated based on the time-space demand; α is the adjustment factor, which controls the relationship between the borrowing fee and the demand; specifically, when the demand D(t, s) increases, P(t, s) will also increase accordingly; the value of the adjustment factor α determines the sensitivity of the borrowing fee to demand fluctuations; β is the benchmark fee, that is, the basic fee under normal demand conditions; this fee is the standard borrowing fee when there is no special demand fluctuation; D(t, s) is the output of the time-space demand function, which represents the borrowing demand at a certain time-space point.

8. A data processing method for library self-service borrowing service according to claim 7, characterized in that The method of the dynamic pricing mechanism during the borrowing peak period based on adaptive spatiotemporal optimization: By introducing intelligent demand forecasting and combining time series forecasting algorithms to predict future demand based on historical borrowing data, time trends and spatial distribution, the demand forecasting and resource scheduling model: in, is the predicted demand, which represents the borrowing demand in the future time and space interval; i is the prediction factor weight, which reflects the contribution of each feature to demand forecasting; including γ i Weighting is performed based on historical borrowing frequency, seasonal changes, and holidays; i (t) is the i-th time-related characteristic function, which represents the demand changes in the time dimension including holidays, working days, and the beginning of the semester; g i (s) is the i-th spatial characteristic function, which represents the demand change of the hotspot location including the library area in the spatial dimension; t0, t1 are time intervals, representing the prediction period; s0, s1 are spatial intervals, representing the predicted spatial range.

Images