A smart campus data management method and system

Abstract

The application relates to a kind of wisdom campus data management method and system. Including using the data behavior modeling of semantic consistency orientation;Abstract all system data as behavior event unit;The modeling result is uniformly converted into semantic vector, and is written into campus semantic behavior database;Establish event standardization rule library;Using the dynamic data flow scheduling engine of behavior link diagram is constructed;According to behavior event node, construct behavior path atlas, extract high-frequency behavior link;Introduce graph search algorithm to calculate data scheduling path;Using the polymorphic perception and difference identification mechanism of heterogeneous behavior state;Capture the deviation between actual behavior event and expected behavior link;Build behavior state comparison model, identify potential abnormal state or strategy failure risk;Using the adaptive optimization and feedback updating mechanism of strategy evolution atlas;Introduce reinforcement learning algorithm, strategy scoring and value return calculation on strategy atlas;Optimize link parameters, resource scheduling logic, notification rhythm.

Description

Technical Field

[0001] This invention relates to the field of smart campus technology, specifically to a smart campus data management method and system. Background Technology

[0002] Existing smart campus data management methods and systems have, to some extent, promoted the construction of information-based and intelligent management in universities, especially in terms of data integration, business collaboration, and information sharing, achieving initial results. However, from the perspectives of overall system architecture, data processing capabilities, behavioral modeling accuracy, response flexibility, and system evolution mechanisms, there are still many shortcomings and obvious drawbacks, hindering the development of smart campuses to a higher level. First, from the perspective of data management, most existing systems are still at the basic integration level of "system docking + database aggregation," lacking a unified data semantic standard and structural mapping mechanism. Different subsystems, such as academic affairs systems, access control systems, teaching evaluation systems, library systems, and OA platforms, have huge differences in data formats, interface protocols, and update cycles, making it difficult to achieve high-quality data integration. Even if physical connectivity is achieved, the lack of semantic consistency often makes it difficult to interact at the logical layer, ultimately resulting in high data redundancy, serious duplicate calls, weak contextual relationships, and system analysis results that are difficult to support accurate decision-making. Secondly, regarding data scheduling, existing systems generally adopt a static interface call strategy. This means that systems pull data through fixed API connections or data middleware relays. The data call process is not optimizable, the scheduling path is not selectable, and the call granularity is uncontrollable. Once the system scales up or the number of call requests surges, it easily leads to performance bottlenecks and data congestion. Furthermore, when dealing with highly sensitive data such as grades, teacher evaluations, and personal privacy, security access control often remains at the level of permission switches and role labels, failing to achieve dynamic, path-based, or scenario-based fine-grained evaluation. This results in both "over-authorization" and "insufficient permissions" issues, seriously affecting the system's security and adaptability. In terms of behavioral modeling, most existing smart campus systems use single-event records or transactional data structures for behavioral analysis, such as recording whether a student has checked in, evaluated a teacher, or logged in. This ignores the temporal, frequency, intensity, behavioral intent, and semantic relationships between behavioral objects behind the behavior. The system struggles to construct complete user behavior chains or paths, unable to understand the student's true state based on the "behavioral process," and can only judge the "behavioral result." Therefore, it cannot achieve higher-level behavioral intelligence tasks such as trend prediction, dynamic profiling, or path recommendation. Furthermore, regarding user anomaly detection, although some systems have introduced behavioral anomaly recognition algorithms, they often rely solely on single-variable threshold triggers, such as "no attendance for three consecutive days" or "abnormal decline in grades." These lack context and fail to consider the interactions between behaviors, making them ineffective in identifying potential risks and prone to false alarms or missed detections. In addition, most current smart campus systems lack dynamic evolution mechanisms for behavioral data, meaning they lack self-learning capabilities. They cannot accumulate experience, optimize paths, or improve strategy accuracy during operation, causing the originally intelligent system to degenerate into an information aggregation platform with "fixed processes and static rules," deviating from the core goal of truly behavior-driven data intelligence.There are also many technical challenges, such as weak data consistency guarantee mechanisms under high concurrency of multiple systems, imperfect control of long-term data link scheduling latency, separation of individual and group behavior modeling, and lack of causal modeling ability between behavioral data and management response. These problems are easily amplified in actual system operation. For example, during critical periods such as concentrated course selection at the beginning of the semester, peak grade inquiry, and end-of-semester evaluation collection, the system exhibits "high latency, low hit rate, and weak strategy" in scheduling, discrimination, and response, ultimately affecting the accuracy, timeliness, and stability of smart campus services. Summary of the Invention

[0003] The purpose of this invention is to provide a smart campus data management method and system, thereby addressing some of the shortcomings and deficiencies pointed out in the background art.

[0004] The technical solution adopted by this invention to solve the above-mentioned technical problems is as follows: A smart campus data management method and system, comprising: S1, adopting data behavior modeling oriented towards semantic consistency:

[0005] S1.1. Abstract all system data into behavioral event units; use a spatiotemporal behavioral graph model to express the dynamic relationships between entities;

[0006] S1.2. The modeling results are uniformly converted into semantic vectors and written into the campus semantic behavior database; and an event standardization rule base is established to enable event recognition, fusion, and alignment.

[0007] S2. Constructed using a dynamic data flow scheduling engine based on behavioral link graphs:

[0008] S2.1 Construct a behavior path graph based on behavior event nodes, and extract high-frequency behavior links including course selection → entering the classroom → after-class evaluation → grade assessment; each link is assigned flow cost and security weight;

[0009] S2.2 Introduce a graph search algorithm to calculate the data scheduling path; activate the link dataset on demand during runtime to enable cross-system data linkage and minimal redundancy flow;

[0010] S3. Employ a multi-modal perception and difference recognition mechanism for heterogeneous behavioral states:

[0011] S3.1, Capture the deviation between actual behavioral events and expected behavioral links; conduct multidimensional analysis of behavioral deviations, including frequency, intensity, and associated objects, and introduce a behavioral state difference function;

[0012] S3.2 Construct a behavioral state comparison model to identify potential abnormal states or policy failure risks; the output difference data is used for system adjustment, user prompts, and policy reconstruction.

[0013] S4. Adaptive optimization and feedback update mechanism based on strategy evolution graph:

[0014] S4.1 Each data scheduling and behavioral response is written into the strategy evolution graph as a strategy execution event; user behavior feedback, including changes in performance after intervention and increases in resource clicks, is collected to construct causal paths;

[0015] S4.2 Introduce reinforcement learning algorithms to perform policy scoring and value reward calculation on the policy graph;

[0016] S4.3, optimize link parameters, resource scheduling logic, and notification rhythm to form a closed loop of behavioral data → management response → effect evaluation → strategy evolution.

[0017] Furthermore, the method for constructing the dynamic data flow scheduling engine of the behavior link graph includes:

[0018] Raw data from various subsystems, including the academic affairs system, access control system, and teaching evaluation platform, are abstracted into event units with behavioral characteristics. Each behavioral event encapsulates the user's behavioral intent, occurrence time, behavioral location, and system interaction state. Through this abstraction, data from different sources and in different formats have a unified semantic expression foundation.

[0019] A multi-dimensional behavior graph is constructed from behavioral events, where nodes represent behavioral events and edges represent logical, temporal, or causal relationships between events. Based on the behavior graph, the scheduling overhead of each candidate path is evaluated. A link cost function is introduced to quantify the communication complexity, processing latency, and resource consumption between modules during data retrieval. The function is defined as follows:

[0020]

[0021] in:

[0022] Φ(l ij ) indicates from event node e i to e j The link scheduling cost function value; r ij The system resource consumption weight value represents the system resources required for data transmission and processing on this link; c ij Indicates the complexity level of the data structure in the link; t ij This represents the average latency value recorded for this link during historical accesses; log(t) ij +2) is used to compress the effect of delay. +2 is to avoid invalidating the logarithm of 0.

[0023] Furthermore, the method for constructing the dynamic data flow scheduling engine of the behavior link graph includes:

[0024] Further assess the security risks of data usage on each path, considering user data permissions, information sensitivity, and historical call behavior; introduce a security weight function, expressed as follows:

[0025]

[0026] in:

[0027] Λ(l ij This represents the security risk weight value on this link; a higher value indicates a higher security requirement. ij This represents the sensitivity level of the data involved in the link; a higher value indicates more sensitive data. ij This indicates the degree of permission matching for the current user under the data type, including whether they have view or modify permissions; d ij α represents the access frequency of the link data in history; β represents the risk adjustment coefficients set in the system, used to adjust the proportion of sensitivity and permission risk in the assessment as needed.

[0028] Furthermore, the method for constructing the dynamic data flow scheduling engine of the behavior link graph includes:

[0029] When the system receives a data scheduling request, it maps the requested target to the terminal node in the behavior path graph. Then, using a path search engine, it performs a comprehensive cost and risk assessment on all candidate paths to select the optimal path. A path evaluation function is introduced to combine the evaluation values ​​of the two functions and score the entire behavior path P, defined as follows:

[0030]

[0031] in:

[0032] Θ(P) represents the total evaluation score of path P, i.e., the comprehensive cost of the path; Φ(l ij ) is the cost function value of each hop link in the path; Λ(l) ij ) is the security weight value of each hop link in the path; γ is the security offset factor set by the system to balance the weight between performance and security: when γ increases, the system pays more attention to security; when γ decreases, it pays more attention to efficiency.

[0033] Furthermore, the method for constructing the polymorphic perception and difference recognition mechanism for heterogeneous behavioral states includes:

[0034] By continuously monitoring campus user behavior, user operation records in different systems are transformed into standardized behavioral event sequences. Each event is based on a timestamp and includes information such as behavior type, target audience, source of the scenario, and depth of the operation. Based on user category, time scenario, and teaching cycle, a predictive behavior chain library is constructed using historical behavioral data.

[0035] The link library describes the typical behavioral path that a certain type of user should present, and is used as a reference model for real-time behavior comparison.

[0036] Upon receiving a real-time behavioral event, the current behavioral trajectory is compared with the corresponding expected path, and the cumulative degree of behavioral deviation is calculated using an integral method. A deviation integral function is introduced to measure the degree of deviation of the user's overall behavioral trend, defined as follows:

[0037]

[0038] in:

[0039] Ψ(u,T) ​​represents the overall behavioral deviation intensity of user u within the time window T; A u (t) is the intensity function of the user's actual behavior at time t; E u (t) is the expected behavioral intensity function of the same user at time t, generated by the link library model; δ is the behavioral deviation index, with a value range of δ>1, used to amplify the impact of persistent or significant behavioral deviations; T is the behavioral observation period, which is 1 day, 1 week, or the length of the teaching task cycle.

[0040] Furthermore, the method for constructing the polymorphic perception and difference recognition mechanism for heterogeneous behavioral states includes:

[0041] Multi-dimensional features are extracted from the behaviors involved in the deviation events, including frequency changes, intensity fluctuations, and resource transfer information. All behavioral features are mapped to standard vectors and compared dimension-by-dimensionally with historical state vectors to construct a behavioral state deviation model. The following behavioral state difference function is introduced:

[0042]

[0043] in:

[0044] Ω(u) is the overall deviation score of user u's current behavioral state; The i-th behavioral feature value collected so far includes the number of course clicks and reading duration; Let η be the expected value of the i-th feature in a user's history, extracted from historical behavior; i represents the degree of fluctuation of the i-th feature in historical data, used to avoid oversensitivity to natural fluctuations; n is the total number of feature dimensions.

[0045] Furthermore, the method for constructing the polymorphic perception and difference recognition mechanism for heterogeneous behavioral states includes:

[0046] Based on multiple user behavior deviation scores, cluster modeling, outlier detection, and temporal evolution analysis are used to identify individual behavioral anomalies and overall strategic failures. When a large range of deviations are concentrated on a certain strategy path, the system needs to determine whether it is a strategy design failure. This is achieved by simulating slight strategy adjustments and observing the sensitivity response of behavioral states, defining the following strategy perturbation response function:

[0047]

[0048] in:

[0049] Σ(P) is the disturbance response sensitivity index of behavior path P; ΔΩ(u) represents the change in the state difference value of user u after the policy adjustment; ΔR(P,h) represents the minor policy disturbance operation implemented on path P, including fine-tuning the entry layout and delaying the intervention prompt time; h is the disturbance amplitude factor.

[0050] The smart campus data management method and system provided by this invention have the following beneficial effects: First, it achieves unified abstraction and semantic consistency modeling of multi-source heterogeneous data in the campus (such as academic affairs systems, access control systems, teaching evaluation platforms, library systems, etc.), integrating the originally scattered and inconsistent data into structured behavioral event units. By constructing spatiotemporal behavioral graphs and behavioral link graphs, it realizes the logical, temporal, and causal relationships between data, laying the foundation for subsequent intelligent scheduling and behavioral analysis. Second, it proposes a link cost function and a security weight function to finely evaluate the resource consumption and security risks of each cross-system data scheduling path. Through the path evaluation function, it achieves a controllable optimal balance between performance and security, significantly improving the efficiency and controllability of system data calls and reducing redundant access and risk exposure. Third, it innovatively constructs a multi-morphic perception and difference recognition mechanism for heterogeneous behavioral states. The system can not only identify the degree of deviation in individual user behavior (such as decreased login frequency, reduced learning time, etc.), but also perform group behavior clustering and policy-level deviation identification, promptly discovering potential policy design failures. Fourth, the system introduces a policy perturbation response function, which can accurately measure the sensitivity of user behavior response by slightly adjusting the policy path layout, rhythm, or trigger point without interrupting the original service process, thereby achieving self-learning, self-adjustment, and self-optimization of the policy. Fifth, the system forms a closed-loop mechanism of "behavioral data → state modeling → intelligent scheduling → policy optimization," possessing high real-time performance, high interpretability, and high adaptability. It significantly improves the level of intelligent data management in the smart campus environment, promotes the transformation of campus information systems from static integration to dynamic perception, prediction-driven, and policy evolution, and provides solid technical support for universities to achieve precise teaching, personalized services, and risk warning. Attached Figure Description

[0051] Figure 1 This invention provides a flowchart of a smart campus data management method and system.

[0052] Figure 2 This is a flowchart of the method for constructing a dynamic data flow scheduling engine based on the behavioral link graph of the present invention.

[0053] Figure 3 This is a flowchart of the method for constructing a multi-state perception and difference recognition mechanism for heterogeneous behavioral states according to the present invention. Detailed Implementation

[0054] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0055] Combined with appendix Figure 1 This invention discloses a smart campus data management method and system. Its core lies in achieving intelligent fusion and dynamic understanding of cross-system data. S1 employs data behavior modeling oriented towards semantic consistency, specifically including the following key operations: S1.1 The system unifies and abstracts structured or semi-structured data from multiple heterogeneous subsystems such as the academic affairs system, access control system, library system, and teaching platform. Instead of storing it in the traditional "data record" method, it abstracts it into "behavioral event units." Each event unit consists of five elements: the subject of the behavior (e.g., student, teacher), the action (e.g., course selection, book borrowing, attendance), the time of occurrence, the location of occurrence, and the object of action (e.g., course, classroom, etc.). Then, a "spatiotemporal behavior graph model" is used to construct dynamic connections between entities. Each node in the model represents a behavioral event, and the edges between nodes represent the temporal sequence and spatial location of the behavior. Relationships or logical dependencies form a computable, traceable, and analyzable behavioral link graph. The S1.2 system further performs semantic transformation on the node and edge information in the constructed behavioral graph, converting it into "semantic vectors" to support subsequent semantic computation, similarity analysis, and strategy reasoning. All semantic vectors are uniformly written into the "Campus Semantic Behavior Database" for centralized management. This database supports semantic access, semantic combination, and link tracing of behavioral events. At the same time, the system establishes a "standardized event rule base" to identify the boundaries and structures of various events, integrate and unify the naming of heterogeneous behaviors from different systems, and solve the problems of inconsistent naming, inconsistent granularity, and unclear behavioral semantics between different systems. This enables the identification, cross-domain alignment, and behavior fusion between events, thereby laying a semantically consistent foundation for subsequent data scheduling, intelligent analysis, and behavior prediction.

[0056] S2 is built using a dynamic data flow scheduling engine based on behavioral link graphs. Its aim is to achieve intelligent, efficient, and low-redundancy data retrieval across systems through structured modeling and dynamic computation of user behavior paths. In S2.1, the system first constructs a "behavioral path graph" based on modeled behavioral event nodes. This graph is a directed graph structure centered on user behavior and with event sequence and logical relationships as edges. Through mining and frequency analysis of historical behavioral data, the system extracts typical and representative high-frequency behavioral links, such as educational activity links like "course selection → entering the classroom → after-class evaluation → grade assessment." In the graph, each link not only represents the natural order of behavior occurrence but also includes two key attributes: first, "flow cost," which measures the resource overhead and processing latency of data transmission, access, and processing across systems in this path; and second, "security weight," used to measure the sensitivity and access permission requirements of the data involved in the path. All links are scored and ranked according to these two dimensions to provide parameter support for scheduling optimization. The S2.2 system further introduces a heuristic graph search algorithm to transform the data scheduling problem into a weighted path optimization search problem. By calculating the cost and security risks of all feasible paths, the optimal data call path is found. In actual operation, the system does not call the data of the entire link at once, but adopts an "on-demand activation mechanism". That is, according to the current business needs and user context, only the necessary event nodes and their related data in the link are activated to realize cross-system data linkage and minimal redundant data flow, avoiding unnecessary data access and system overhead, thereby achieving the goal of efficient scheduling, accurate access, and secure and controllable smart campus data call.

[0057] S3 employs a heterogeneous behavior state multi-state perception and difference recognition mechanism. Its core purpose is to enable the system to possess real-time perception and intelligent recognition capabilities for changes in behavior states under different users and contexts, thereby promoting dynamic optimization and precise intervention of management strategies. S3.1, based on a behavioral event flow and historical modeling-based expected behavior link library, continuously compares the user's actual behavior sequence with the typical behavior path it should present. By calculating the deviations between behavior trajectories in time, content, and structure, it identifies "behavioral deviation points" and quantifies the multi-dimensional characteristics of these deviations. These include frequency deviation (whether the number of times the behavior occurs decreases or increases sharply), intensity deviation (whether the level of participation or completion quality of the behavior declines), and associated object deviation (whether there is a shift in behavioral goals or a sudden change in resource access). Based on these characteristics, the system constructs a "behavioral state difference function," which compares the current behavioral state vector with the historical behavioral state vector dimension by dimension, outputting a behavioral state difference score to measure the severity and risk level of behavioral deviation. S3.2 utilizes multiple user... The system constructs a "behavioral state comparison model" based on state difference scoring. This model combines group behavior trends, individual behavior stability, and behavioral link context to perform clustering, classification, and anomaly identification, thereby distinguishing between individual sporadic anomalies and systemic strategy failures. If the system identifies a behavior path that exhibits high deviations simultaneously among a large number of users and is highly sensitive to system strategy fine-tuning, it determines that the path has a risk of insufficient strategy design or content mismatch. Based on this, the system will output a "difference data package," including the type of behavioral deviation, deviation intensity, associated path, and recommended response measures, as feedback input to the system. This drives the adjustment of service strategy parameters, generates user-end prompts, or provides intervention suggestions for managers, thereby realizing an intelligent closed-loop management mechanism of system perception—identification—response—reconstruction.

[0058] S4 employs an adaptive optimization and feedback update mechanism based on a strategy evolution graph, aiming to enable the system to possess the ability for strategy self-evolution. By driving strategy updates and optimizing management methods through behavioral data, it enhances the system's intelligent decision-making level and dynamic adaptability. S4.1 records each data scheduling, system response, and management intervention triggered by user behavior as a "strategy execution event" in the "strategy evolution graph." This graph is a dynamically updatable causal network used to track the entire process path of "behavior-response-feedback." For example, when the system pushes academic reminders to a student for continuous absences, feedback data such as whether the student returned to school on time, whether course grades improved, and whether resource usage rebounded will be continuously collected and written into the graph, forming a closed-loop link from intervention to effect feedback. Each node in the graph represents a strategy or feedback state, and each edge records the causal logical relationship between strategy execution and result. S4.2, based on the strategy evolution graph, introduces reinforcement learning algorithms to perform strategy scoring and value reward calculations on various strategy paths. The system adjusts the effectiveness of the strategy based on the actual feedback results. The evaluation process assesses the effectiveness of different strategy combinations across various user groups using a strategy benefit function. Each strategy path is assigned a dynamically changing value score, reinforcing strategies that deliver positive results and eliminating or adjusting those with weakened or even negative effects. Based on the strategy scoring results, the S4.3 system adaptively optimizes the entire behavioral response system. This includes optimizing parameter weights in behavioral links (e.g., assigning higher monitoring priority to certain behaviors), adjusting resource scheduling logic (e.g., reordering the priority of learning resource recommendations), and optimizing notification triggering rhythm (e.g., intelligently pushing information based on user response habits). This forms a systematic intelligent closed loop of "behavioral data → management response → effect evaluation → strategy evolution," enabling the smart campus data management system to continuously learn, adapt in real-time, and update itself, significantly improving the personalization, precision, and dynamic responsiveness of education management.

[0059] Example 1

[0060] Combined with appendix Figure 2In this embodiment, a university is running a new smart campus data management system based on this invention. The system integrates multiple subsystems such as the academic affairs system, access control system, and teaching evaluation platform, and uses "behavioral events" as the core data unit to drive data scheduling optimization. Student Xiaoming, a third-year student, entered the teaching building from his dormitory on Wednesday morning of the tenth week of teaching, swiped his access card to enter classroom 504, and attended the "Data Structures" course. After class, the system sent him a teaching evaluation reminder, and this evaluation data became part of the course's teaching quality analysis the following week. The system abstracts this series of user behaviors into the following behavioral event units: e1 is "selecting the course 'Data Structures'", e2 is "swiping the card to enter the teaching building", e3 is "attending class", e4 is "submitting post-class evaluation", and e5 is data preparation before "grade evaluation". The system constructs these events as a link in a behavioral path graph: e1→e2→e3→e4→e5. Each segment involves cross-system data scheduling and interaction, such as from the course selection system to the access control system, and from the attendance module to the evaluation platform. For the link e2→e3 (i.e., from "entering the teaching building" to "attending class"), the system needs to correlate access control data with attendance system data to verify whether the student has entered the classroom of the selected course on time. The resource consumption value r involved in this link is... 23 It mainly consists of two service nodes: an access control data parsing module and a sign-in verification module, with a resource assessment of r. 23 =3.2; meanwhile, the data structure complexity involved in this link is c. 23 =9.0 (because the check-in data contains five types of fields, including timestamp, GPS location, device ID, and course association, and the nested structure is quite deep); while the average access latency of this link in the past two weeks was t. 23 = 1.5 seconds. Substituting the above value into the link cost function defined in this invention:

[0061]

[0062] According to system evaluation criteria, when the link cost function Φ value is greater than 10, it indicates that the link call has a high impact on system performance. In this case, the system scheduling engine will trigger an "optimization suggestion mechanism," such as trying to preload access control data cache or modifying the attendance verification module for parallel processing to reduce the r value and thus optimize the Φ value. Another link is e4→e5 (post-class evaluation submission to grade assessment), which typically requires comprehensive processing of evaluation platform results, course numbers, teacher information, and historical teaching quality baseline data. The resource weight r involved in this link is... 45 =4.5, data structure complexity c 45 =4.0 (data structure is relatively simple), while the average response latency of the evaluation system interface is t. 45 = 3.0 seconds, substitute into the formula:

[0063]

[0064] The Φ value is significantly higher than the threshold, indicating that this cross-system call chain is putting considerable pressure on the data scheduling engine. To address this, the system can intelligently select a "delayed, low-frequency strategy," scheduling such evaluation-related data during off-peak data flow periods with lower real-time requirements, thereby reducing the risk of system resource congestion. In actual deployment, the system dynamically monitors and sorts the Φ values ​​of different links, prioritizing the path combination with the lowest Φ value to ensure that critical, high-real-time links are scheduled first, while links with high latency tolerance are processed later, thus achieving intelligent scheduling of the overall campus data flow. In actual projects, the system sets the r value to a range of 1.0–5.0, corresponding to calls from lightweight cached types to multi-level dependent computational calls; the c value generally ranges from 1–12, reflecting differences in data structure complexity; and the t value comes from historical interface performance monitoring, typically between 0.5 and 5 seconds.

[0065] To further explain the data usage security risk assessment part of the dynamic data flow scheduling engine of the behavior link graph proposed in this invention, specifically by introducing a security weight function Λ(l ij The system quantifies and manages the security of each link's access. In this system, each segment of the student behavior event chain e1→e2→e3→e4→e5 involves data access and retrieval. Data access risk no longer relies on fixed access control rules but is dynamically assessed based on the link's sensitivity, user permissions, and historical access patterns. The following example uses link e4→e5 (from after-class evaluation to grade assessment). In actual operation, this link needs to read data such as student evaluation scores, course ID, instructor identity, course content, and historical grade mapping relationships. Some of this data, such as "instructor evaluation scores," is evaluation-related privacy data, thus its sensitivity level is high. According to the system's definition of data sensitivity levels: 1 represents low sensitivity (e.g., course name), 2 represents medium sensitivity (e.g., attendance data), and 3 represents high sensitivity (e.g., grades and evaluations). The data sensitivity level involved in this link is denoted as s. 45 =3. In the current scheduling request, the operation was initiated by the system teaching analysis engine, not a single user; therefore, the permission level is set to the highest level to match u. 45 =5 indicates viewing and analysis permissions; for regular teachers, the setting is typically 3. According to historical logs, this data link is accessed frequently in the system, 31 times in the past week. To prevent excessive data exposure, the system records its access frequency index as d. 45 =31. The safety weight function of this invention is defined as follows:

[0066]

[0067] Substituting the data, we have α = 0.7 and β = 0.3. These coefficients are configured by the system administrator according to the system's operating mode. α is typically set in the range of 0.5 to 0.8, representing the level of importance placed on the sensitivity of the data itself. β is typically set in the range of 0.2 to 0.5, used to control the impact of caller permissions and historical access behavior. The sum of these two parameters is 1, facilitating standardized control. Substituting these values ​​into the calculation yields:

[0068]

[0069] Calculation result Λ(l 45 The value ≈ 0.1766 indicates that the security risk of this link is relatively low. This is mainly because although the data sensitivity level is high, the system caller's permissions are fully matched, and the access frequency is high but controllable. Therefore, the system assesses that this link can be called normally during the runtime without triggering the security delay mechanism. In contrast, if an ordinary student attempts to access the e4→e5 link, i.e., permission level u... 45 =1, the system will recalculate the security weights:

[0070]

[0071] Although the decrease is not significant compared to system-level permission requests, the system settings trigger permission verification logic by default when ordinary students access this link. Even if the security weight is not high, it will be blocked due to policy restrictions, demonstrating that the security weight function is used for risk assessment rather than direct authorization decisions. Another example is the link e2→e3 (access control to classroom → class attendance), where the data involves real-time location and course schedule, and the sensitivity level is set to s. 23 =2, the caller is the user, and the privilege level is u. 23 =3, historical access frequency d 23 =5, substitute into the calculation:

[0072]

[0073] The security weight of this link is relatively high. The system will prioritize the use of caching mechanisms during scheduling to avoid unnecessary security risk control pressure caused by repeated calls to real-time data. This shows that the function can not only be used for path priority sorting, but also dynamically guide the selection of scheduling modes, such as whether to enable caching, whether to segment scheduling, and whether to trigger user confirmation.

[0074] The entire process of student Xiaoming participating in the "Data Structures" course is recorded by the system as a complete behavioral chain. The event nodes are e1 (course selection), e2 (access to the teaching building), e3 (attendance), e4 (post-class evaluation), and e5 (grade assessment). This behavioral path P = {e1→e2, e2→e3, e3→e4, e4→e5} is constructed as a cross-system data scheduling path. When the system receives a scheduling request, such as the teaching quality analysis engine requesting a path to obtain the data chain of "course evaluation scores affecting the final grade," the system needs to select the optimal path among all feasible paths to achieve efficient and secure data scheduling. At this time, the path evaluation function Θ(P) needs to be introduced for comprehensive scoring. First, the system calls the link cost function Φ(l ij ) and safety weight function Λ(l ij As calculated above, the cost of link e2→e3 is Φ(l 23 )≈10.57, safety weight value Λ(l 23 The cost of link e4→e5 is Φ(l)≈0.2675. 45 )≈31.84, the safety weight is Λ(l 45 With a value approximately 0.1766, the system now receives a scheduling request, maps the path target to node e5, lists candidate paths (in this example, the primary path P1) using the path search engine, and evaluates them using the following path evaluation function:

[0075]

[0076] In this system, γ is a security offset factor used to dynamically adjust the emphasis on performance and security under different operational strategies. Before exam weeks, the system prioritizes stability and sets a higher value; during daily operations, it prioritizes efficiency and sets a lower value. Currently, for a normal teaching week, γ = 5.0, with a range of 1.0 to 10.0 set by the administrator and dynamically adjusted according to the campus security level strategy. The evaluation values ​​for each link in behavior path P1 are set as follows (some approximate values):

[0077] e1→e2:Φ(l 12 ) = 5.2, Λ(l 12 ) = 0.12

[0078] e2→e3:Φ(l 23 ) = 10.57, Λ(l 23 ) = 0.2675

[0079] e3→e4:Φ(l 34 ) = 7.6, Λ(l 34 ) = 0.19

[0080] e4→e5:Φ(l 45)=31.84,Λ(l 45 ) = 0.1766

[0081] Substitute the above values ​​into the path evaluation function for calculation:

[0082] Θ(P1)=(5.2+5·0.12)+(10.57+5·0.2675)+(7.6+5·0.19)+(31.84+5·0.1766)

[0083] Θ(P1)=(5.2+0.6)+(10.57+1.3375)+(7.6+0.95)+(31.84+0.883)

[0084] Θ(P1)=5.8+11.9075+8.55+32.723≈58.98

[0085] The total evaluation score for this path is Θ(P1)≈58.98. The system compares and ranks this score with other candidate paths, and selects this path as the optimal path for current data scheduling based on the lowest score. If there is a secondary path P2 with the structure e1→e6→e7→e5, which calls the historical analysis cache module instead of the real-time check-in system, its overall Φ value is lower but its security is slightly weaker. Θ(P2)≈62.15 is set. Since Θ(P1)<Θ(P2), the system still selects path P1 for execution. However, if the current time period switches to the final exam period, γ is set to 8.0, and the system's preference for security increases. In this case, the calculation is recalculated.

[0086] e2→e3: Φ=10.57, Λ=0.2675, the safety-weighted part becomes 8·0.2675=2.14

[0087] e4→e5: Φ=31.84, Λ=0.1766, safety weighted average is 8·0.1766=1.41

[0088] The overall score becomes:

[0089] Θ(P1)′≈5.2+0.96+10.57+2.14+7.6+1.52+31.84+1.41≈61.24

[0090] The secondary path P2 has a higher security risk, with a cumulative security weight of 3.1. It corresponds to a high score under high γ conditions, with a total score of Θ(P2)′≈65.8. In this case, P1 still wins, but the system will remind the administrator that this path is at an edge risk in a high-security scenario, and the security tolerance of key links should be improved. Through the above examples, it can be seen that the path evaluation function Θ(P) in this invention can achieve a comprehensive trade-off between performance and security of data call paths, enabling the smart campus system to achieve flexible and intelligent path selection under different operating strategies. At the same time, the value ranges of each parameter in the system (such as γ=1.0~10.0, r=1.0~5.0, s=0~3) all support dynamic configuration, improving the practical applicability and promotional value of the method, and meeting the needs of data management efficiency and security in real campus environments.

[0091] Example 2

[0092] Combined with appendix Figure 3 In this embodiment, set up in a smart campus application scenario where Xiaoming continues his participation in the "Data Structures" course, this system constructs a multi-morphic perception and difference recognition mechanism for heterogeneous behavioral states to achieve real-time tracking and deviation identification of his learning behavior in different subsystems, thereby assisting in teaching intervention and strategy optimization. Xiaoming, a third-year computer science student, exhibits normal behavior in the system including logging into the online teaching platform five times a week, signing in for each course once a week, borrowing a related book on average every two weeks, completing evaluations within 24 hours after class, and consistently maintaining a mid-to-high-achieving academic record. Based on historical big data, the system constructs an expected behavior link library for Xiaoming, a "high-performing student who values ​​courses," including nodes such as course learning, book borrowing, signing in, and evaluations, as well as their time patterns and behavioral intensity patterns. The system uses these links as a reference model for real-time behavior comparison. Entering the eleventh week, the system monitoring reveals that Xiaoming only logged into the platform once this week, borrowed zero books, was late for signing in twice, and failed to complete the "Data Structures" course evaluation, exhibiting significantly abnormal behavior. The system immediately aligns and analyzes his current behavioral trajectory with the expected behavior links, generating an actual behavior intensity function A. u (t) and the expected behavior intensity function E u (t), and uses the following behavioral bias integral function for quantitative evaluation:

[0093]

[0094] Where Ψ(u,T) ​​represents the intensity of Xiaoming's overall behavioral deviation within a one-week observation window; A u (t) represents the intensity of Xiaoming's actual behavior at time t (such as the number of times he accesses courses per minute, his attendance record, his borrowing behavior, etc.); E u(t) is the expected behavior model generated based on its profile type; δ is the deviation index, configured in the system according to the behavior stability type, usually between 1.5 and 3.0. In this example, Xiaoming's past behavior is highly regular, so δ = 2.5 is set to amplify the impact of his abnormal behavior on the result; the observation period T is 7 days (teaching cycle). Taking platform login behavior as an example, the expected login behavior E u (t) represents 5 times per week, distributed from Monday to Friday, with an average behavior intensity of 1.0 per instance, and actual behavior A. u (t) only appears on Mondays with an intensity of 0.8, and is 0 on the rest of the days. The system is discretely sampled and integrated every 24 hours. The integrated result of the login behavior bias is as follows:

[0095]

[0096] In the course evaluation process, Xiaoming failed to complete the originally planned evaluation of the "Data Structures" course, with an expected intensity of E. u (t) is 1.0 (the system is set to complete within 24 hours as the full result), but the actual value is 0.0, and the integral deviation is |0-1|. 2.5 =1. Regarding attendance, late arrivals occurred on Wednesdays and Thursdays (behavior intensity set at 0.5), while attendance was on time for the rest (behavior intensity 1.0). The expected daily intensity is 1.0. The points are calculated as follows:

[0097]

[0098] The expected number of books borrowed was 1, but the actual number was 0, resulting in a deviation of 1 point. The total deviation score for Xiaoming's behavior this week is:

[0099] Ψ total =Ψ login +Ψ sign +Ψ review +Ψ library =4.01 + 0.352 + 1 + 1 = 6.362

[0100] The system sets a threshold of 5.0. When Ψ(u,T) ​​> 5.0, it considers the user's behavior to deviate from the expected pattern, requiring strategic feedback or intervention. Therefore, the system marks Xiaoming as having "moderate behavioral deviation" and pushes a personalized prompt: "Your course participation has decreased this week. We suggest you complete the teaching evaluation and review the course content in a timely manner." Simultaneously, this difference value will be synchronized to the teacher's teaching behavior analysis module, prompting the teacher to pay attention to the student's change in learning status and, if necessary, provide one-on-one tutoring. Through this mechanism, the system not only achieves cross-platform and cross-behavioral type behavior perception but also achieves unified quantification of multidimensional deviations through the integral function Ψ(u,T). The δ value is set within the range of 1.5 to 3.0. The system parameters can be adjusted according to the student's behavioral stability type, and the behavior observation period T can be selected as 1 day, 3 days, 7 days, or the entire teaching task cycle, possessing extremely high adaptability, flexibility, and practical promotion value.

[0101] After the system identifies a moderate deviation in the overall behavioral trend using the integral function Ψ(u,T), it proceeds to the next stage of the difference identification mechanism: multi-dimensional feature extraction and detailed analysis of behavioral states. This quantifies the specific degree of deviation across different behavioral dimensions and provides a basis for subsequent strategy recommendations. Through a multi-modal perception mechanism of heterogeneous behavioral states, the system extracts five main behavioral features from Xiaoming's data over the past two weeks: f1: number of online course clicks, f2: total video learning time (minutes), f3: number of professional books borrowed, f4: class attendance rate, and f5: post-class evaluation submission rate. These features originate from heterogeneous subsystems such as the academic affairs system, library system, learning platform, and evaluation platform, and are uniformly transformed into a structured behavioral feature vector within the system. Based on historical records, Xiaoming's expected behavioral value is: During the current monitoring period, its actual behavior is as follows: Simultaneously, the system calculates volatility indicators based on its historical behavior fluctuation data: η1 = 5, η2 = 60, η3 = 0.3, η4 = 0.1, η5 = 0.05. These η values... i The values ​​represent the natural fluctuations in a user's past behavior across various dimensions, preventing the system from overreacting to normal fluctuations. The system substitutes the above data into the behavioral state difference function for calculation, the function being defined as:

[0102]

[0103] Substituting the data into the table, and assuming the total number of feature dimensions n = 5, the calculations are as follows:

[0104] Item 1 (Number of Clicks): (10-25) 2 / (5+1)2 =225 / 36 = 6.25

[0105] Item 2 (Study Time): (120-300) 2 / (60+1) 2 =32400 / 3721≈8.71

[0106] Item 3 (Borrowing Books): (0-1) 2 / (0.3+1) 2 =1 / 1.69≈0.59

[0107] Item 4 (Attendance Rate): (0.6-1) 2 / (0.1+1) 2 =0.16 / 1.21≈0.132

[0108] Item 5 (Teacher Evaluation Rate): (0-1) 2 / (0.05+1) 2 =1 / 1.1025≈0.907

[0109] Summing and dividing by 5 yields:

[0110]

[0111] This value represents Xiaoming's overall behavioral deviation score. The system presets the following difference threshold levels: Ω < 1.5 indicates mild deviation, 1.5–3.0 indicates moderate deviation, and Ω > 3.0 indicates severe deviation. Therefore, Xiaoming's current score Ω(u) ≈ 3.32 has reached the "severe behavioral deviation" level. The system then pushes the following response strategies: First, it sends a warning report to the counselor, including Xiaoming's key deviation characteristics: a 60% decrease in clicks, a 180-minute reduction in study time, and no feedback from teachers. It also suggests paying attention to factors such as learning pressure and lack of interest in the course. Second, the system simultaneously generates personalized push notifications, such as "This week's 'Data Structures' materials have been updated; it is recommended to view the accompanying videos for Chapter Four," to improve his behavioral recovery rate through gentle intervention. Furthermore, the behavioral score will serve as a training sample for the teaching analysis platform, used to dynamically optimize the teaching intervention model. Regarding parameter settings, η in the system... i The value range is typically between 0.05 and 100, generated based on the sensitivity of the behavior type. The behavior index Ω(u) range is generally set with threshold segments of 0, 1.5, 1.5, 3.0, 3.0, and ∞, and is dynamically adjusted according to the student's semester progress and behavioral stability. Through this mechanism, the system achieves a complete closed-loop perception of "individual behavior - offset features - dynamic profile" in terms of data, constructing a behavioral state difference analysis capability that is both granular and strategically valuable, greatly improving the level of refined teaching management and student support response efficiency in the smart campus environment.

[0112] A case where Xiaoming's behavior state shows a serious deviation in the smart campus system. This system not only conducts an intervention analysis on his individual deviation, but also further activates the most crucial part of the polymorphic perception and difference recognition mechanism for heterogeneous behavior states - the group analysis and dynamic feedback optimization at the strategy level. After the system detects that Xiaoming's score Ω(u) = 3.32 reaches the serious deviation level, it does not handle it in isolation. Instead, it inputs this deviation result together with the behavior deviation scores of a total of 428 students in the same major and grade into the behavior state clustering model. The system uses the K-Means algorithm to divide the students into 3 categories according to the behavior deviation feature vectors. Among them, the low-deviation group (Cluster1) accounts for about 62.1%, the moderate-deviation group (Cluster2) accounts for about 29.7%, and the high-deviation group (Cluster3) including Xiaoming accounts for 8.2%. The system further observes that among the high-deviation group, 93% of the students have not completed the after-class teaching evaluation task for the "Data Structure" course, and such a large-scale teaching evaluation missing rate has never occurred in the past 5 semesters for this course. The system initially judges that this phenomenon is not caused by the individual意愿 of the students, but rather源于 a problem in the design of a certain key node in the current teaching strategy path. So it activates the strategy perturbation sensitivity test mechanism. The system sets the current after-class teaching evaluation strategy path P for the "Data Structure" course as: Home page of the teaching platform → Course information page → Teaching evaluation reminder pop-up window (path P), and performs a mild perturbation operation on it: Move the teaching evaluation reminder entry 20px upward from the middle of the course information page, and delay the pop-up time from "5 minutes after class" to "10 minutes after class". The adjustment perturbation amplitude is set as h = 0.05, which belongs to the micro-perturbation interval defined by the system (value range: 0 < h ≤ 0.1, indicating that the relative operation or strategy weight adjustment does not exceed 10%). In the second teaching week after implementing the strategy perturbation, the system re-collects the behavior data of the high-deviation group. Xiaoming completed the teaching evaluation under the new strategy, and 41 other controlled students also completed the teaching evaluation. The average value of their behavior state scores decreased from Ω before = 3.17 to Ω after = 2.21, and the state difference change is ΔΩ(u) = 3.17 - 2.21 = 0.96. The corresponding strategy perturbation is ΔR(P,h) = h = 0.05. Substitute it into the strategy perturbation response function:

[0113]

[0114] This value is significantly higher than the system's set response sensitivity threshold of 8.0 (the recommended range is 5.0–10.0, flexibly configured according to the overall system behavior), indicating that path P is highly sensitive to minor policy adjustments and exhibits significant behavior recovery, further verifying the original design of the policy has a failure issue. Based on this conclusion, the system pushes a recommendation to the academic affairs decision module, suggesting optimizing the position of the evaluation reminder pop-up in all courses to the top of the information page and uniformly setting the delay reminder time to 10 minutes. Simultaneously, the priority of this policy path is increased by 15% through a behavior path priority weight enhancement mechanism. Furthermore, the policy perturbation response score Σ(P) = 19.2 is also recorded in the policy evolution graph for use in subsequent similar courses' policy simulation and fine-tuning suggestion model training. In summary, through the polymorphic perception of Xiaoming and the group of students' behaviors and the perturbation response evaluation of policy paths, the smart campus data management system in this invention can not only accurately identify individual behavioral deviations but also has the ability to discern overall policy failures from minor perturbations. Through refined calculation and policy feedback mechanisms, it achieves closed-loop optimization, thereby significantly improving the intelligence and dynamic adaptive capabilities of educational data governance.

[0115] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A smart campus data management method, characterized in that... Includes the following steps: S1. Employing data behavior modeling oriented towards semantic consistency: S1.

1. Abstract all system data into behavioral event units; use a spatiotemporal behavioral graph model to express the dynamic relationships between entities; S1.

2. The modeling results are uniformly converted into semantic vectors and written into the campus semantic behavior database; and an event standardization rule base is established to enable event recognition, fusion, and alignment. S2. Constructed using a dynamic data flow scheduling engine based on behavioral link graphs: S2.1 Construct a behavior path graph based on behavior event nodes, and extract high-frequency behavior links including course selection → entering the classroom → after-class evaluation → grade assessment; each link is assigned flow cost and security weight; S2.2 Introduce a graph search algorithm to calculate the data scheduling path; activate the link dataset on demand during runtime to enable cross-system data linkage and minimal redundancy flow; S3. Employ a multi-modal perception and difference recognition mechanism for heterogeneous behavioral states: S3.1, Capture the deviation between actual behavioral events and expected behavioral links; conduct multidimensional analysis of behavioral deviations, including frequency, intensity, and associated objects, and introduce a behavioral state difference function; S3.2 Construct a behavioral state comparison model to identify potential abnormal states or policy failure risks; The output difference data is used for system adjustments, user prompts, and policy restructuring. S4. Adaptive optimization and feedback update mechanism based on strategy evolution graph: S4.1 Each data scheduling and behavioral response is written into the strategy evolution graph as a strategy execution event; user behavior feedback, including changes in performance after intervention and increases in resource clicks, is collected to construct causal paths; S4.2 Introduce reinforcement learning algorithms to perform policy scoring and value reward calculation on the policy graph; S4.3, optimize link parameters, resource scheduling logic, and notification rhythm to form a closed loop of behavioral data → management response → effect evaluation → strategy evolution.

2. The smart campus data management method according to claim 1, characterized in that... The method for constructing the dynamic data flow scheduling engine of the behavior link graph includes: Raw data from various subsystems, including the academic affairs system, access control system, and teaching evaluation platform, are abstracted into event units with behavioral characteristics. Each behavioral event encapsulates the user's behavioral intent, occurrence time, behavioral location, and system interaction state. Through abstraction, data from different sources and in different formats have a unified semantic expression foundation. The behavioral events are constructed into a multi-dimensional behavioral graph, where nodes represent behavioral events and edges represent logical, temporal, or causal relationships between events. Based on the behavioral graph, the scheduling cost of each candidate path is evaluated.

3. The smart campus data management method according to claim 2, characterized in that... The method for constructing the dynamic data flow scheduling engine of the behavior link graph includes: Further assess the security risks of data usage on each path, considering user data permissions, information sensitivity, and historical call behavior; introduce a security weight function, expressed as follows: in: This represents the security risk weight value on this link; a higher value indicates a higher security requirement. This represents the sensitivity level of the data involved in the link; a higher value indicates more sensitive data. This indicates the degree of matching between the current user's permission level and the data type, including whether the user has view or modify permissions; This represents the historical access frequency of the link data. This is a risk adjustment coefficient set in the system, used to adjust the weight of sensitivity and authority risks in the assessment as needed.

4. The smart campus data management method according to claim 3, characterized in that... The method for constructing the dynamic data flow scheduling engine of the behavior link graph includes: When the system receives a data scheduling request, it maps the requested target to the terminal node in the behavior path graph. Then, using a path search engine, it performs a comprehensive cost and risk assessment on all candidate paths to select the optimal path. A path evaluation function is introduced to combine the evaluation values ​​of the two functions to evaluate the entire behavior path. The scoring is defined as follows: in: For path The total evaluation score, i.e., the comprehensive cost of the path; It is the cost function value of each hop link in the path; It is the security weight value of each hop link in the path; A safety offset factor is set for the system to balance performance and safety: when With increased size, the system places greater emphasis on security; when When reducing size, more attention is paid to efficiency.

5. The smart campus data management method according to claim 1, characterized in that... The method for constructing the multi-mode perception and difference recognition mechanism for heterogeneous behavioral states includes: By continuously monitoring campus user behavior, user operation records in different systems are transformed into standardized behavioral event sequences. Each event is based on a timestamp and includes information such as behavior type, target audience, source of the scenario, and depth of the operation. Based on user category, time scenario, and teaching cycle, a predictive behavior chain library is constructed using historical behavioral data. The link library describes the typical behavioral path that a certain type of user should present, which is used as a reference model for real-time behavior comparison. When a real-time behavior event is received, the current behavior trajectory is compared with the corresponding expected link, and the cumulative degree of behavior deviation is calculated by integration.

6. The smart campus data management method according to claim 5, characterized in that... The method for constructing the multi-mode perception and difference recognition mechanism for heterogeneous behavioral states includes: Multi-dimensional features are extracted from the behaviors involved in the deviation events, including frequency changes, intensity fluctuations, and resource transfer information. All behavioral features are mapped to standard vectors and compared dimension-by-dimensionally with historical state vectors to construct a behavioral state deviation model. The following behavioral state difference function is introduced: in: For users A comprehensive deviation score for the current behavioral state; The currently collected number These include behavioral characteristic values ​​such as the number of course clicks and reading time; For the first time in user history The expected value of each feature is extracted from historical behavior; For the first The degree of fluctuation of a feature in historical data is used to avoid oversensitivity to natural fluctuations; This represents the total number of feature dimensions.

7. A smart campus data management method according to claim 6, characterized in that... The method for constructing the multi-mode perception and difference recognition mechanism for heterogeneous behavioral states includes: Based on multiple user behavior deviation scores, cluster modeling, outlier detection, and temporal evolution analysis are used to identify individual behavioral anomalies and overall strategic failures. When a large range of deviations are concentrated on a certain strategy path, the system needs to determine whether it is a strategy design failure. This is achieved by simulating slight strategy adjustments and observing the sensitivity response of behavioral states, defining the following strategy perturbation response function: in: For behavior path The disturbance response sensitivity index; Indicates user The change in its state difference value after the strategy adjustment; Indicates path The minor strategic disruptions implemented included fine-tuning the entry layout and delaying the intervention notification time; This is the disturbance amplitude factor.

Images