Multi-dimensional growth tree construction system and method based on knowledge point diagnosis

Abstract

The invention discloses a multi-dimensional growth tree construction system and method based on knowledge point diagnosis, and relates to the technical field of growth tree construction. The system comprises a knowledge point diagnosis module, a growth tree topology construction module, a self-adaptive path planning module and an internal and external dynamic change module. According to the method, state parameters are mastered and serve as direct input of instantiation nodes of a growth tree topology construction module, multi-dimensional attribute vectors are initialized and edge weights are dynamically adjusted, so that a structured initial growth tree is constructed, the personalized path matching degree is calculated according to the structured initial growth tree, dynamic adjustment is conducted, optimal dynamic adaptation of a learning path is achieved, and the learning efficiency is improved. Furthermore, by evaluating the degree of adaptation to internal and external dynamic changes and executing adaptive adjustment, the continuous effectiveness of path recommendation in a changing environment is guaranteed, then the effectiveness of multi-dimensional growth tree construction is improved, and the problem that in the prior art, path recommendation efficiency is poor is solved. The problem of low effectiveness of multi-dimensional growth tree construction caused by low accuracy and interpretability of a dynamic path recommendation function exists.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of growth tree construction, and particularly relates to a multi-dimensional growth tree construction system and method based on knowledge point diagnosis. BACKGROUND

[0002] Firstly, a full-semester knowledge point database is built according to an AI (Artificial Intelligence) large model and a big data technology, knowledge units are split according to levels such as disciplines, chapters and modules, a "understanding-understanding-mastery" gradient target is labeled, and capability dimension (such as logical reasoning and application transfer) and literacy dimension (such as learning habit and positive values) evaluation indexes are embedded, existing knowledge point graph technology and multi-element evaluation algorithm are fused, and comprehensive coverage of diagnosis dimensions is realized. Secondly, users select diagnosis granularity (such as knowledge points / chapters / semesters) by themselves, push hierarchical test questions (basic questions correspond to single knowledge points, medium questions fuse 2-3 knowledge points, and difficult questions combine complex situations) based on adaptive evaluation technology, and generate a multi-dimensional diagnosis report by combining answer results through behavior recognition technology to capture data such as test speed and abandonment frequency.

[0003] Then, data visualization technology is used to map the multi-dimensional diagnosis report results into the initial form of the growth tree, to label the completion degree of knowledge, capability and literacy modules with different colors, to synchronously embed a game-based positive incentive mechanism (such as unlocking new branches for knowledge point reaching the standard, and obtaining honor marks for stage achievements), and to improve user initiative by referring to the incentive logic of existing game-based learning platforms. Subsequently, personalized learning resources (such as wrong question analysis videos and targeted practice questions) are automatically pushed based on the diagnosis report, and the growth tree form is dynamically updated by continuously tracking user learning behaviors and retest results, and this process integrates existing personalized recommendation algorithm and learning behavior analysis technology.

[0004] Finally, academic evaluation data (such as high school entrance examination point matching degree) for student groups are connected, and certification information such as education background, work experience and honor for adult groups is connected, data authenticity is ensured through blockchain technology, a complete personal digital portrait is constructed, the growth tree becomes a comprehensive growth carrier with functions of visual display, capability diagnosis and positive guidance, and the whole process relies on existing AI diagnosis model, big data tracking and visualization technology.

[0005] For example, the Chinese invention patent with announcement number CN113393221B discloses a method and system for pushing enterprise ecosystem services based on online data, which includes: S1, allocating and storing corresponding in-stock party information according to pre-created service modules; wherein, the service modules include intellectual property modules, business and taxation modules, enterprise qualification certification modules, product testing and certification modules, policy subsidy application modules, and financing and loan modules; S2, obtaining and updating the current enterprise overview information of the accessed party; S3, constructing an enterprise growth tree GTR model based on the enterprise overview information; wherein, the enterprise growth tree GTR model generates growth nodes according to the service modules; S4, performing node analysis on the current growth nodes of the enterprise growth tree; S5, performing service push processing by combining the results of node association analysis and the service modules.

[0006] For example, Chinese invention patent CN113673943B discloses a method and system for assisting personnel appointment and removal decisions based on resume big data, including: S1, receiving personnel resume text and processing the personnel resume text through a Chinese resume text processing algorithm; S2, performing mixed short text multi-label classification based on expert rules and knowledge graphs; S3, constructing personnel's professional social network based on the organizational tree extracted from resume big data; S4, obtaining personnel information that meets the requirements based on query conditions and visualizing the personnel's professional social network.

[0007] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems: In existing technologies, recommendation algorithms, such as collaborative filtering or knowledge tracing models, are mostly derived from e-commerce or simple knowledge structure domains. When they are directly transferred to ability development scenarios with complex, multi-dimensional, and cross-functional characteristics, their underlying assumptions often fail. These algorithms typically make recommendations based on historical behavioral data of groups, failing to fully model and incorporate learners' specific cognitive styles, real-time emotional states, and non-linear development opportunities triggered by accidental cross-domain inspirations. This results in a significant deviation between the recommendation path and the individual's actual growth needs. At the same time, the recommendation process is usually a "black box," with opaque internal algorithm logic, making it difficult for educators and learners to understand the reasoning behind the recommendation results, severely weakening users' trust in the system and their willingness to adopt it. On the other hand, to achieve effective path optimization, continuous A / B testing and model updates based on learning performance data are necessary. However, this requires complex experimental design, long-term data accumulation, and expensive computing resources. A deeper problem is that existing systems are mostly designed as closed-loop systems, with their capability dimensions, node structures, and evolution rules often being rigid, lacking sufficient flexibility and scalability, making it difficult to quickly respond to dynamic changes in the capabilities required by external society and the workplace. Although the industry has proposed the idea of ​​interconnection through open APIs (Application Programming Interfaces), in practice, achieving secure data sharing and model interoperability across platforms and organizations still faces a series of non-technical fundamental obstacles, including strict data privacy protection regulations such as GDPR constraints and the lack of unified technical standards and interface specifications. There is also the problem of low accuracy and interpretability of dynamic path recommendation functions leading to low effectiveness in constructing multi-dimensional growth trees. Summary of the Invention

[0008] To address the technical problem of low effectiveness in multi-dimensional growth tree construction due to the low accuracy and interpretability of dynamic path recommendation functions in existing technologies, this invention provides a multi-dimensional growth tree construction system and method based on knowledge point diagnosis. The technical solution is as follows: On one hand, a multi-dimensional growth tree construction system based on knowledge point diagnosis is provided. This system includes: a knowledge point diagnosis module, a growth tree topology construction module, an adaptive path planning module, and a module for handling internal and external dynamic changes. The knowledge point diagnosis module is used to acquire and process the user's learning data, generating mastery status parameters for each knowledge point in the preset knowledge system through multi-dimensional analysis. The growth tree topology construction module is used to map the quantitative indicators corresponding to each growth dimension to the multi-dimensional attribute vectors of each node based on the mastery status parameters, with the root knowledge point of the preset knowledge system as the root node of the growth tree and each level of knowledge point as a child node. It also constructs directed weighted edges between nodes according to the preorder relationship and association strength of the preset knowledge system, thereby forming a structured initial growth tree. The adaptive path planning module... The module quantifies the degree of fit between the recommended path and the learner's needs, obtaining a personalized path matching degree. Based on the personalized path matching degree, it determines whether to perform adaptive adjustments to the path matching degree to achieve the best dynamic fit between the user's learning path and their personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization. The module for internal and external dynamic changes quantifies the response efficiency to changes in internal learning effect data optimization needs and external social and workplace ability needs, as well as the degree of fit for cross-platform interoperability, obtaining a degree of fit for internal and external dynamic changes. Based on the degree of fit for internal and external dynamic changes, it determines whether to perform adaptive adjustments to the change fit degree to achieve continuous and stable optimization and goal convergence under internal state evolution and external environmental disturbances, thereby ensuring the effectiveness of growth path recommendations.

[0009] On the other hand, a multi-dimensional growth tree construction method based on knowledge point diagnosis is provided. This method includes: acquiring and processing user learning data; generating mastery status parameters for each knowledge point in a pre-defined knowledge system through multi-dimensional analysis; based on these mastery status parameters, using the root knowledge point of the pre-defined knowledge system as the root node of the growth tree and knowledge points at each level as child nodes, mapping the quantitative indicators corresponding to each growth dimension to multi-dimensional attribute vectors of each node, and constructing directed weighted edges between nodes according to the preorder relationships and association strengths of the pre-defined knowledge system, thereby forming a structured initial growth tree; and quantifying the fit between the recommended path and the learner's needs to obtain... Personalized path matching degree is used to determine whether to perform adaptive adjustment of path matching degree to achieve the best dynamic adaptation between the user's learning path and personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization effect; the response efficiency to changes in internal learning effect data optimization needs and external social and workplace ability needs, as well as the degree of cross-platform interoperability, are quantified to obtain the adaptability to internal and external dynamic changes, and the degree of adaptability to change adaptation degree is determined based on the adaptability to internal and external dynamic changes to achieve continuous and stable optimization and goal convergence under the evolution of internal state and external environmental disturbances, thereby ensuring the effectiveness of growth path recommendation.

[0010] Beneficial effects The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. By acquiring and processing user learning data, multi-dimensional analysis is performed to generate mastery status parameters for each knowledge point within a pre-defined knowledge system. These parameters include at least a quantitative value of the mastery level, a weakness type identifier, and a description of error association characteristics. These parameters serve as direct inputs to the growth tree topology construction module for instantiating nodes, initializing their multi-dimensional attribute vectors, and dynamically adjusting the weights of directed weighted edges between nodes. Based on this, the root knowledge point of the pre-defined knowledge system is used as the root node of the growth tree, with each level of knowledge point as a child node. The quantitative indicators corresponding to each growth dimension are mapped to the multi-dimensional attribute vectors of the nodes. Directed weighted edges between nodes are constructed based on the priori relationships and association strengths within the knowledge system, thus forming a structured initial growth tree. Further, based on this growth tree, the degree of fit between the recommended path and the learner's needs is quantified to obtain a personalized path matching degree. The system determines whether to perform adaptive adjustment of the path matching degree based on this matching degree to achieve the best dynamic fit between the user's learning path and their individual cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization. Simultaneously, the system quantifies the response efficiency to changes in internal learning effectiveness data optimization needs and external social and workplace competency requirements, as well as the degree of cross-platform interoperability adaptation, obtaining the adaptability to internal and external dynamic changes. Based on this adaptability, the system determines whether to perform adaptive adjustments to achieve continuous and stable optimization and goal convergence under internal state evolution and external environmental disturbances, thereby ensuring the effectiveness, robustness, and timeliness of the growth path recommendation. Through a multi-level dynamic evaluation and adaptive adjustment mechanism, an intelligent learning path planning system that can continuously evolve and accurately match individual development needs with changes in the external environment is constructed. This improves the effectiveness of multi-dimensional growth tree construction and solves the problem in existing technologies where the low accuracy and interpretability of dynamic path recommendation functions lead to low effectiveness in multi-dimensional growth tree construction.

[0011] 2. The frequency of user interaction behavior is dynamically adjusted based on the density of high-frequency interactive knowledge points. By calculating the concentration and intensity of user interaction behavior on specific knowledge point clusters in real time (i.e., the density of high-frequency interactive knowledge points), the frequency and rhythm of subsequent interaction prompts, feedback, and task pushes are dynamically adjusted accordingly. This aims to achieve intelligent interaction management that adaptively synchronizes with the user's real-time cognitive load and focus of interest: When high-density interaction focus is detected, the system actively reduces non-core interference, allowing the user to focus deeply and improve cognitive processing efficiency; when the interaction density is low, guidance and incentives are appropriately increased to maintain learning continuity. This mechanism not only maintains long-term engagement by preventing cognitive overload and interaction fatigue, but also dynamically optimizes the user's attention allocation and cognitive resource investment by forming a data-driven personalized interaction rhythm model, thereby improving knowledge acquisition and problem-solving overall. The system also dynamically adjusts the proportion of time spent on recommended paths based on the task interval duration, analyzing the recommended paths... The system dynamically adjusts the proportion of planned stop times at nodes along the path based on the expected transfer intervals and their distribution characteristics. This aims to achieve adaptive optimization of path timing planning in relation to the user's actual travel pressure and physiological and psychological rhythm. When there are consecutive long-interval task transfers along the path, the system reduces the overall proportion of stop times and prioritizes concentrating stop resources on key nodes before and after the long intervals. This ensures travel efficiency while providing necessary buffers and recovery for users before and after high-pressure travel. When task intervals are short and dense, the overall proportion of stop times is increased to avoid rushed operations and cognitive fatigue caused by an excessively fast pace. By transforming stop times from static allocation to a flexible resource dynamically scheduled based on travel pressure, this mechanism not only optimizes the overall efficiency of time utilization but also significantly improves the sustainability, comfort, and final completion quality of users when following recommended paths to perform complex tasks through predictive rhythm management. This, in turn, enhances the effectiveness of multi-dimensional growth tree construction.

[0012] 3. Dynamically adjust the efficiency of demand priority determination based on the average cycle of new capability node identification and deployment. By continuously tracking the entire lifecycle data of historical new capability nodes from identification to deployment, calculate their average implementation cycle and stability indicators, and dynamically adjust the efficiency of the priority determination process for new demand items accordingly. This aims to build an adaptive demand decision-making gating mechanism that matches the organization's actual delivery capabilities in real time: When the average deployment cycle is long or fluctuates little, the determination process adopts a deep analysis mode, ensuring that limited resources are invested in high-value demands through thorough argumentation, avoiding resource waste caused by hasty decisions; when the average cycle shortens or fluctuations intensify, switch to an agile or fast-track mode to compress the determination time, preventing the decision-making process from becoming a bottleneck in responding to market changes, and maintaining a dynamic balance between demand flow rate and backend delivery throughput. This mechanism, by transforming demand determination efficiency from a fixed process to a flexible variable based on organizational delivery performance feedback, not only significantly improves the end-to-end flow efficiency from demand identification to value delivery, but also systematically reduces the risk of resource mismatch by ensuring that the priority determination rhythm is aligned with actual production capacity, enhancing the organization's strategic responsiveness and return on resource investment in a changing environment. The gray-scale verification cycle of rules is dynamically adjusted based on the iteration cycle of evolutionary rules (such as growth tree upgrade logic and path recommendation rules). By parsing metadata such as the preset iteration cycle and change magnitude of the evolutionary rules themselves, the gray-scale verification cycle before application to the production environment is dynamically calculated and adjusted. This aims to achieve an adaptive optimal match between verification resource investment and rule evolution risks and business impacts: For stable rules with long iteration cycles and small change magnitudes, a longer verification cycle is allocated to allow for sufficient observation and robustness testing to ensure the reliability of changes; for high-change rules with frequent iterations or logical refactoring, the verification cycle is compressed but supplemented by higher-frequency indicator monitoring and circuit breaker mechanisms to strictly control the risk exposure window while quickly verifying core value. This mechanism transforms the fixed verification duration into an elastic cycle dynamically calculated based on the inherent rhythm and change intensity of rule evolution. This not only significantly improves the overall rate of rule iteration and experimental efficiency, but also maximizes the agility of rule optimization and the continuity of business value delivery while ensuring system stability through precise risk-speed balance. This gives the evolution process of the system rule base itself the intelligent characteristic of continuous self-optimization, thereby improving the effectiveness of multi-dimensional growth tree construction. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1A schematic diagram of the structure of a multi-dimensional growth tree construction system based on knowledge point diagnosis provided in this application embodiment; Figure 2 This is a flowchart illustrating the optimization of dwell time percentage in a multi-dimensional growth tree construction system based on knowledge point diagnosis, provided in an embodiment of this application. Figure 3 A flowchart for optimizing the priority determination efficiency of a multi-dimensional growth tree construction system based on knowledge point diagnosis, provided in an embodiment of this application; Figure 4 A flowchart illustrating the multi-dimensional growth tree construction method based on knowledge point diagnosis provided in this application embodiment. Detailed Implementation

[0015] The technical solution provided in this application will now be described in conjunction with the accompanying drawings.

[0016] To facilitate understanding of the embodiments of this application, the following points will be explained first: First, in this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates an "or" relationship between the preceding and following related objects, but it does not exclude the possibility of indicating an "and" relationship; the specific meaning can be understood in context. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a and b and c. Here, a, b, and c can be single or multiple.

[0017] Second, the use of prefixes such as "first" and "second" in this application is solely for the purpose of distinguishing and describing different things belonging to the same category, and does not constrain the order, size, or quantity of things. For example, "first message" and "second message" are simply different messages, and there is no chronological, size, or priority relationship between them.

[0018] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0019] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0020] likeFigure 1 The diagram shown is a structural schematic of a multi-dimensional growth tree construction system based on knowledge point diagnosis provided in this application embodiment. The multi-dimensional growth tree construction system based on knowledge point diagnosis provided in this application embodiment includes: a knowledge point diagnosis module, a growth tree topology construction module, an adaptive path planning module, and a module for handling internal and external dynamic changes.

[0021] As the first module of this system, the knowledge point diagnosis module is used to acquire and process the user's learning data. Through multi-dimensional analysis, it generates mastery status parameters of each knowledge point in the preset knowledge system. The mastery status parameters are the direct inputs in the growth tree topology construction module to instantiate each node and initialize its multi-dimensional attribute vector, as well as to dynamically adjust the weights of the directed weighted edges between nodes. The mastery status parameters include at least the quantitative value of the mastery level of the knowledge point, the weak type identifier, and the description of error association features.

[0022] It should be understood that a pre-defined knowledge system refers to a knowledge model with a complete logical framework, constructed through a standardized structure during the initialization phase, based on a specific discipline, field, or skill category. Essentially, it is a systematically organized knowledge graph, clearly defining three core elements: knowledge nodes, which are the basic knowledge units constituting the system (such as concepts, principles, and skill points), each node having a unique identifier and content definition; hierarchical structure, which displays the subordinate and aggregate relationships of knowledge through parent-child relationships (such as "chapter-knowledge point") or classification systems, forming a tree-like or network topology; and association rules, including the pre-order relationships between knowledge nodes (i.e., prerequisite knowledge that must be mastered before learning a certain knowledge point), association relationships (such as the similarity, comparison, or application scenario coupling of knowledge points), and difficulty progression relationships. The pre-defined knowledge system provides the system with an objective reference system for assessing the user's cognitive state and an underlying logical framework for constructing personalized growth paths, ensuring that all diagnostic, modeling, and recommendation processes are built upon the system's knowledge logic foundation.

[0023] It needs to be explained that by collecting various types of raw data generated by users during the learning process (such as answer records, operation logs, and dwell time), and using a multi-dimensional analysis engine to clean, attribute, and model the data, a set of refined mastery status parameters is generated for each knowledge point in the pre-set knowledge system. This set of parameters is not a single score, but a structured state vector, which includes at least: a quantitative value of mastery level, which objectively represents the user's current understanding level of the knowledge point with continuous numerical values ​​or discrete levels; a weakness type identifier, which classifies the specific nature of the cognitive gap based on error pattern analysis (such as "conceptual misunderstanding", "formula misuse", or "reasoning jump"); and an error association feature description, which reveals possible erroneous links or confusion patterns between knowledge points in the user's erroneous cognition (such as "often misapplying the features of knowledge point A to knowledge point B"). These mastery state parameters then serve as the key driving inputs for the growth tree topology construction module: The module first instantiates each knowledge point as a node in the growth tree based on the hierarchical structure of the knowledge system, and immediately maps the aforementioned state parameters to the multi-dimensional attribute vector of that node, ensuring that each node carries a personalized capability state from its inception. Simultaneously, the module establishes directed weighted edges between nodes based on the inherent logical relationships (preorder and association) of the knowledge system, and dynamically initializes or adjusts the weights of these edges based on the mastery state parameters, particularly the characteristics of weak types and error associations. For example, the weights of the associated edges for a pair of knowledge points that users frequently confuse are specially marked or adjusted for special attention in subsequent path planning. Therefore, the output of the diagnostic module directly and in real-time defines the "state" of each node in the growth tree and the "strength" of the connections between nodes, transforming the static knowledge structure into a personalized intelligent model that dynamically maps to an individual's cognitive state.

[0024] The second module of this system, the growth tree topology construction module, is used to construct a structured initial growth tree based on the mastery state parameters, with the root knowledge point of the preset knowledge system as the root node of the growth tree, and the knowledge points at each level as child nodes. It maps the quantitative indicators corresponding to each growth dimension to the multi-dimensional attribute vector of each node, and constructs directed weighted edges between nodes according to the preorder relationship and association strength of the preset knowledge system.

[0025] It's important to explain that this involves integrating an abstract knowledge system with individualized learning states to construct a structured, computable, and visual model. The model uses the pre-defined hierarchy and logical relationships of the knowledge system as its static skeleton: starting from the root knowledge point as the root node of the growth tree, child nodes are generated layer by layer according to the subordinate and related relationships of knowledge points, forming a basic tree topology. Simultaneously, the module dynamically incorporates mastery status parameters output from the knowledge point diagnostic module as the model's dynamic flesh and blood: it transforms the quantified mastery level, weakness type identification, and error association characteristics corresponding to each knowledge point into standardized attribute vectors covering multiple dimensions such as knowledge mastery, ability development, and learning behavior through predefined mapping rules, and binds them to the corresponding nodes. Thus, each node not only represents a knowledge object but also carries the learner's ability status regarding that object across multiple dimensions. Furthermore, the module establishes directed weighted edges between nodes based on the pre-order dependencies and logical association strengths defined in the knowledge system; the direction of the edges reflects the learning order, while the initial weights are determined jointly by the association strength and diagnostic parameters. For example, if the diagnostics reveal a significant imbalance in a user's mastery of two knowledge points with a priori relationship, the weight of that priori edge will be reduced to reflect the actual obstacles in the learning path. Through this process, the module integrates the static knowledge structure, the dynamic individual state, and the complex relationships between knowledge points into a unified, structured initial growth tree model, providing a logical and personalized data foundation for subsequent path planning and dynamic optimization.

[0026] The third module of this system, the adaptive path planning module, is used to quantify the degree of fit between the recommended path and the learner's needs and responses, so as to obtain the personalized path matching degree. Based on the personalized path matching degree, it determines whether to perform adaptive adjustment of the path matching degree to achieve the best dynamic fit between the user's learning path and personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization effect.

[0027] It is important to understand that the adaptive path planning module includes: an adaptive path planning quantization unit and a path matching degree control unit.

[0028] First, the adaptive path planning quantification unit is used to obtain the personalized path matching degree based on the matching degree parameters, which include the planned path compliance rate, path adjustment frequency, and deviation path regression time. It needs to be explained that the planned path compliance rate refers to the proportion of the number of knowledge points in the user's actual learning sequence that overlap with the recommended sequence within a preset period, relative to the total number of knowledge points in the recommended sequence. This is achieved by using the knowledge point IDs and timestamps (accurate to the minute) of the user's actual clicks and interactions within the system to form the user's actual learning sequence. The user's completion status for each knowledge point (e.g., answer submission, resource completion, task acceptance, etc.) is recorded with binary labels. By associating the recommended knowledge point sequence with the user's actual learning sequence, the number of overlapping and completed knowledge points is counted, and then divided by the total number of knowledge points in the recommended path to obtain the planned path compliance rate. The path adjustment frequency refers to the number of times the user actively adjusts the learning path (e.g., changing the order, adding / removing modules) within a preset period. Within the preset period, the user's path operation log is traversed, and the total number of all active and passive adjustments is counted, which is the path adjustment frequency within that preset period. Among them, proactive adjustment behavior data represents records of path adjustment operations initiated by users, including adjustment trigger timestamps, adjustment types (changing the order of knowledge points, adding / deleting knowledge point modules, switching learning difficulty gradients, changing learning goals), and path differences before and after adjustment (such as the original path knowledge point sequence and the adjusted knowledge point sequence); passive adjustment behavior data represents records of path corrections automatically triggered based on users' real-time learning data (such as answer error rate, excessive cognitive load, and insufficient knowledge point mastery), including adjustment trigger conditions (such as an answer accuracy rate of <60% for a certain knowledge point), adjustment logic (such as inserting supplementary knowledge points, reducing the difficulty of subsequent tasks), and adjustment execution timestamps; deviation path regression time refers to the average time it takes for a user to return to the recommended path after leaving it; by comparing the user's actual learning sequence with the recommended path sequence, the system identifies the user's first departure from the recommended path, records the deviation trigger timestamp, continuously monitors user learning behavior, and when the user clicks again to learn an unfinished knowledge point (or subsequent knowledge point) in the recommended path, it is determined as a regression behavior, and the regression trigger timestamp is recorded. The difference between the regression trigger timestamp and the deviation trigger timestamp is recorded as the deviation path regression time.

[0029] Secondly, the compliance rate proportion factor is multiplied by the result of the analysis of the proportion of planned path compliance rate to the baseline compliance rate to obtain the compliance rate compensation value; the adjustment frequency proportion factor is multiplied by the result of the analysis of the proportion of path adjustment frequency to obtain the adjustment frequency compensation value; the regression duration proportion factor is multiplied by the result of the analysis of the proportion of regression duration to deviation from the path to obtain the regression duration compensation value; the compliance rate compensation value, adjustment frequency compensation value, and regression duration compensation value are then superimposed to obtain the personalized path matching degree. The specific constraint expression for obtaining the personalized path matching degree is as follows: ; In the formula, R represents the personalized path matching degree; c1 represents the compliance rate ratio factor obtained from the growth tree construction database; c2 represents the adjustment frequency ratio factor obtained from the growth tree construction database; c3 represents the regression duration ratio factor obtained from the growth tree construction database; S0 represents the compliance rate baseline value obtained from the growth tree construction database; K0 represents the adjustment frequency baseline value obtained from the growth tree construction database; T0 represents the regression duration baseline value obtained from the growth tree construction database; S represents the planned path compliance rate; K represents the path adjustment frequency; and T represents the deviation path regression duration.

[0030] It should be understood that a higher planned path adherence rate indicates a higher accuracy in matching the path with the user's learning needs, cognitive habits, and knowledge base, and a lower frequency of path adjustments. A higher planned path adherence rate means that even if a user deviates from the path due to accidental factors (such as temporary interest or accidental operation), they will quickly realize that deviating has no additional value because the core content of the path matches their needs (such as efficiently solving learning pain points and conforming to cognitive rhythm), and thus return to the path in a short time, resulting in a shorter return time. A higher path adjustment frequency indicates that the initial path and the path after multiple adjustments still do not fully match the user's needs, the user's trust and dependence on the path are lower, and after deviating from the path, they are more inclined to continue exploring non-recommended content (such as independently finding suitable resources) rather than quickly returning to the path with adaptation defects, resulting in a longer return time. Meanwhile, there is a positive correlation between planned path adherence rate and personalized path matching degree. The higher the planned path adherence rate, the more efficiently users can acquire value in learning (such as quickly understanding knowledge points and solving their own weaknesses), and naturally they are more willing to follow the recommended path to advance their learning, and will not easily deviate or give up, resulting in a higher personalized path matching degree. There is a negative correlation between path adjustment frequency and personalized path matching degree. The higher the path adjustment frequency, the more frequently users initiate proactive adjustments (such as reducing difficulty or changing resource types) to adapt to themselves. At the same time, the system will trigger passive adjustments based on data such as low adherence rate and high error rate (such as inserting supplementary learning content and correcting knowledge connections), resulting in a lower personalized path matching degree. There is also a negative correlation between deviation path regression time and personalized path matching degree. The longer the deviation path regression time, the more likely users are to turn to more suitable non-recommended content (such as cross-platform resources and self-exploration directions) after deviating from the path, forming a continuous exploration behavior, resulting in a lower personalized path matching degree.

[0031] Finally, the path matching degree adjustment unit is used to compare the personalized path matching degree with the path matching degree benchmark value. If the personalized path matching degree is greater than or equal to the path matching degree benchmark value, the path matching degree adaptive adjustment is not performed. Otherwise, the interaction behavior frequency optimization and dwell time ratio optimization are performed according to the matching degree correction amount. The matching degree correction amount represents the negative difference between the personalized path matching degree and the path matching degree benchmark value.

[0032] It should be further explained that the specific process for optimizing the frequency of interaction behavior is as follows: S100, in response to the continuous interaction sequence generated within the current session window, extracts the knowledge point identifiers associated with all interaction behaviors in the continuous interaction sequence in real time, and constructs and dynamically updates a real-time knowledge point density map within the session based on a preset time decay function and spatial proximity weight; wherein, the time decay function is used to give the knowledge points interacted within the current session window an initial weight, and the spatial proximity weight is used to give weight gain to the knowledge points co-occurring in neighboring interaction behaviors according to the semantic correlation of the knowledge points in the preset knowledge topology.

[0033] S101, perform cluster analysis on the real-time knowledge point density map within the session to identify knowledge point clusters whose cumulative knowledge point weights are greater than the knowledge point cluster benchmark value, and mark them as high-frequency interactive knowledge point clusters in the current session; calculate the comprehensive density value of the high-frequency interactive knowledge point cluster, where the comprehensive density value represents a function of the sum of the weights of all knowledge points in the knowledge point cluster, the number of knowledge points in the cluster, and the average correlation weight between knowledge points in the cluster.

[0034] Optimization of the frequency of interactive behavior execution also includes: S102, the comprehensive density value of the high-frequency interactive knowledge point cluster is mapped to the corresponding interactive behavior frequency adjustment parameter through a nonlinear mapping function. The nonlinear mapping function is configured as follows: when the comprehensive density value is lower than the lower limit of the knowledge point cluster reference value, an interactive behavior frequency gain coefficient is output based on the matching degree correction amount and the current comprehensive density value. The current user interactive behavior frequency and the interactive behavior frequency gain coefficient are combined to obtain the next user interactive behavior frequency. When the comprehensive density value is within the knowledge point cluster reference range, the current user interactive behavior frequency is maintained, and no interactive behavior frequency optimization is performed. When the comprehensive density value is higher than the upper limit of the knowledge point cluster reference value, an interactive behavior frequency suppression coefficient is output based on the matching degree correction amount and the current comprehensive density value. The current user interactive behavior frequency and the interactive behavior frequency suppression coefficient are combined to obtain the next user interactive behavior frequency. The knowledge point cluster reference range represents the closed interval formed by the lower limit and upper limit of the knowledge point cluster reference value.

[0035] It is important to understand that the time decay function is in the form of exponential decay: ; Among them, T time -λ represents the time decay function; -λ represents the decay coefficient; Δt is the time difference between the current time and the time when the interaction occurs; the spatial proximity weight is determined based on the pre-built knowledge graph by calculating the inverse of the shortest path between knowledge point pairs that co-occur in the same interaction unit or adjacent interaction units.

[0036] It should be understood that, firstly, the system responds in real time to the user's continuous interactive behavior within the current session window, extracts the knowledge point identifiers associated with each interaction, and dynamically constructs and updates a real-time knowledge point density graph within the session based on a preset time decay function (using an exponential decay form to give higher weight to recent interactions) and spatial proximity weight (determined based on the reciprocal of the shortest path between knowledge points in the knowledge graph, strengthening the co-occurrence weight of knowledge points with close semantic connections). This graph can accurately capture the user's immediate cognitive focus and thought process.

[0037] Subsequently, cluster analysis was performed on the aforementioned density map to identify high-frequency interactive knowledge point clusters whose cumulative weights exceeded a preset benchmark value, and their comprehensive density value was calculated (this value is a function of the sum of the weights of knowledge points within the cluster, their quantity, and the average correlation strength between knowledge points). The core technical effect of this step is that the system not only identifies the knowledge topics that users are intensively processing, but also provides an objective basis for subsequent frequency adjustment by quantifying the cognitive cohesion of these topics.

[0038] Furthermore, a nonlinear mapping function maps the overall density value to specific interaction frequency adjustment parameters: when the density value is below the lower reference limit, a gain coefficient is output to moderately increase the interaction frequency to maintain user engagement; when the density value is within the normal reference range, the current frequency is maintained to avoid unnecessary interference; when the density value is above the upper reference limit, a suppression coefficient is output to actively reduce the interaction frequency to prevent cognitive overload. The parameters are finely calibrated using a matching degree correction and combined with the current frequency to generate the final optimized interaction frequency for the next interaction.

[0039] In this embodiment, the interactive rhythm and the user's real-time cognitive load are adaptively synchronized: it can reduce interruptions to ensure the continuity of thinking when the user is deeply focused, and provide appropriate guidance during the divergent thinking period. Thus, while improving the cognitive efficiency of a single interaction, it significantly optimizes the long-term learning experience and knowledge internalization effect.

[0040] It is necessary to understand that, such as Figure 2 The diagram shows a flowchart of the dwell time ratio optimization process for a multi-dimensional growth tree construction system based on knowledge point diagnosis provided in this application embodiment. The specific process is as follows: Starting with "parse the recommended path and extract temporal features," the system first parses the recommended path generated by the user and containing ordered task nodes. Then, it calculates the transition interval between task nodes in the path, generating an interval duration sequence. Based on this sequence, the system further identifies and extracts path-level temporal features. Subsequently, the process enters the "mapping and obtaining adjustment coefficients" stage. In this stage, the temporal features extracted in the previous step and a "matching degree correction amount" are used as input to query a predefined "interval duration-dwell time ratio mapping table." This mapping table is configured to output two key coefficients: a "global dwell time adjustment coefficient" for overall adjustment; and a "local enhancement coefficient," specifically used to enhance dwell nodes located near "long interval clusters" (i.e., a set of consecutive long intervals with cumulative values ​​exceeding a preset threshold). Finally, the process enters the "redistributing the total dwell time budget" stage. First, based on the output "global dwell time adjustment coefficient," the preset total dwell time of the route baseline is scaled to calculate the dynamically adjusted "total available dwell time of the route." This completes the dynamic allocation process of dwell time budget based on route temporal characteristics.

[0041] It should be further explained that the specific process for optimizing the dwell time ratio is as follows: S201, parsed as the recommended path generated for the current user, the recommended path contains task nodes arranged in order; calculate the expected transfer interval between every two adjacent task nodes, forming an interval sequence {G1, G2, G3, ..., G...} (n-1) Based on the interval duration sequence, path-level temporal features are identified and extracted, including at least: maximum interval duration, average interval duration, interval duration variance, and long interval clusters.

[0042] S202, input the path-level temporal features and matching degree correction amount into a predefined interval duration-stay duration ratio mapping table. The interval duration-stay duration ratio mapping table is configured to output the global stay duration adjustment coefficient and the local enhancement coefficient for stay nodes near long interval clusters. Long interval clusters are defined as the set of interval durations that are cumulatively higher than the preset interval threshold. Stay nodes near long interval clusters refer to the task nodes that are adjacent to the long interval cluster in the path sequence order and are planned as stay points. Specifically: the preceding stay node is the stay node that is immediately before the start point of the long interval cluster; the following stay node is the stay node that is immediately after the end point of the long interval cluster.

[0043] S203, Based on the coefficients output in step S201, the total reserved stay time budget in the recommended path is reallocated: Based on the global dwell time adjustment coefficient, the total dwell time of the path baseline is scaled to determine the dynamically adjusted total dwell time available for the path (i.e., the total dwell time available for the path represents the result of combining the global dwell time adjustment coefficient and the total dwell time of the path baseline). Within the dynamically adjusted total dwell time available for the path, the individual allocation time of dwell nodes in neighboring long-interval clusters is weighted and gained according to the local enhancement coefficient to ensure that users have relatively more preparation time after long movement intervals.

[0044] It should be understood that, firstly, the system parses the user-generated recommended path (containing N task nodes arranged in sequence), calculates the estimated transfer interval between adjacent nodes to form a sequence, and extracts key path temporal features, including the maximum interval duration, average interval duration, variance, and long interval clusters (i.e., the set of interval durations that continuously exceed a preset threshold). The technical effect of this step is that the system not only quantifies the overall time pressure distribution of the path but also accurately locates high-pressure continuous movement segments in the journey, providing structured insights for subsequent resource allocation.

[0045] Subsequently, these temporal features and the real-time calculated matching degree correction are input into a predefined interval duration-dwell duration ratio mapping table. This mapping table is configured to output two key coefficients: a global dwell duration adjustment coefficient and a local enhancement coefficient for dwell nodes near long-interval clusters. Here, "nearby" specifically refers to the preceding dwell node immediately before the start point and the following dwell node immediately after the end point of the long-interval cluster in the path sequence. The core technical effect of this step is that the system implements a two-layer elastic strategy: the global coefficient responds to the overall rhythm of the path (e.g., if the average interval is long, the total dwell ratio is reduced to improve efficiency; if the average interval is short, the ratio is increased to ease the rhythm), while the local coefficient provides precise resource allocation for high-voltage road sections.

[0046] Finally, the total dwell time budget is redistributed based on the output coefficients: first, the baseline total dwell time is scaled using global coefficients to determine the dynamically adjusted total dwell time available for the path; then, within this total dwell time framework, local coefficients are used to weight the individual dwell times of preceding and following dwell nodes near long-interval clusters. This achieves a fundamental shift in dwell time from static allocation to dynamic scheduling. By proactively adding "buffer pads" before and after long-interval movements, it ensures that users are fully prepared before high-pressure transfers and can effectively recover after fatigue accumulation. Furthermore, global scaling avoids unnecessary expansion of the total dwell time, thereby optimizing overall time utilization efficiency and significantly improving the sustainability, comfort, and task completion quality of users executing complex paths.

[0047] In this embodiment, by dynamically analyzing the expected transfer interval distribution of task nodes in the recommended path, and intelligently adjusting the global dwell time ratio and the local dwell time at key nodes before and after long-interval clusters, this technology achieves adaptive matching between path timing planning and user travel pressure and physiological rhythm: it optimizes time utilization efficiency in the overall journey, avoids resource waste, and provides enhanced buffers for key nodes before and after high-pressure continuous movement sections, effectively preventing fatigue accumulation and hasty operation. Thus, while ensuring the smoothness of path execution, it significantly improves user experience comfort, task continuous execution capability, and final completion quality.

[0048] The fourth module of this system, the Internal and External Dynamic Change Module, is linked with the Adaptive Path Planning Module. It is used to quantify the response efficiency to changes in internal learning effect data optimization needs and external social workplace ability needs, as well as the degree of cross-platform interoperability. It obtains the adaptability to internal and external dynamic changes and determines whether to perform adaptive adjustment based on the adaptability to achieve continuous and stable optimization and target convergence under internal state evolution and external environmental disturbances, thereby ensuring the effectiveness of growth path recommendation.

[0049] It should be explained that the module for handling internal and external dynamic changes includes: a quantification unit for internal and external dynamic changes and a change adaptation control unit.

[0050] First, the internal and external dynamic change quantification unit is used to obtain the adaptability to internal and external dynamic changes based on dynamic change parameters. These parameters include path matching correction value, node structure adjustment response speed, and node association adaptation time. It's important to explain that the path matching correction value refers to the personalized path matching degree that is re-acquired if adaptive path matching adjustment has been performed; otherwise, the current personalized path matching degree is used. The node structure adjustment response speed refers to the total time taken from clarifying the node structure adjustment needs (e.g., internally identifying overly coarse knowledge point breakdowns or externally adding new workplace skills requiring new nodes) to completing all adjustments and officially launching the system. This is achieved by recording the final launch completion time (the time when the adjusted node structure is deployed and fully available to users) and the time recorded for requirement identification completion (e.g., the system automatically identifies node defects, manual intervention, etc.). The difference between the time to confirm the requirement for a new node (accurate to the minute) is recorded as the node structure adjustment response speed; the node association adaptation time refers to the total time spent optimizing the association relationship between the node and related nodes in the existing node system (such as prerequisite dependencies, subsequent expansion, and parallel association logic) after completing the node structure adjustment (adding, deleting, or modifying nodes) until the association is smooth and conforms to the knowledge / capability connection rules; the difference between the recorded association adaptation completion time (the time when the association logic is verified and the node is seamlessly connected with the existing system) and the recorded association adaptation start time (the time when the association logic optimization is started after the node structure development is completed) is recorded as the node association adaptation time.

[0051] Secondly, the path matching compensation value is obtained by multiplying the path matching ratio factor by the result of the analysis of the proportion of the path matching correction value to the path matching baseline value; the response speed compensation value is obtained by multiplying the response speed compensation factor by the result of the analysis of the proportion of the node structure adjustment response speed to the response speed baseline value; the adaptation time compensation value is obtained by multiplying the adaptation time compensation factor by the result of the analysis of the proportion of the node association adaptation time; the path matching compensation value, response speed compensation value, and adaptation time compensation value are then superimposed to obtain the adaptation degree to internal and external dynamic changes. The specific constraint expression for the adaptation degree to internal and external dynamic changes is as follows: ; In the formula, E represents the adaptability to internal and external dynamic changes; k1 represents the path matching ratio factor obtained from the growth tree construction database; k2 represents the response speed ratio factor obtained from the growth tree construction database; k3 represents the adaptation time ratio factor obtained from the growth tree construction database; G0 represents the path matching baseline value obtained from the growth tree construction database; L0 represents the response speed baseline value obtained from the growth tree construction database; M0 represents the adaptation time baseline value obtained from the growth tree construction database; G represents the path matching correction value; L represents the node structure adjustment response speed; and M represents the node association adaptation time.

[0052] It's important to understand that a higher path matching correction value indicates a smaller matching deviation in the original path, eliminating the need for deep adjustments to the node structure (such as adding large-scale supplementary learning nodes, splitting and merging core nodes, or reconstructing node association logic), resulting in a faster response time for node structure adjustments. A higher path matching correction value also indicates that node adjustments are only minor local adjustments (such as adjusting the difficulty of a single node or changing the order of a small number of nodes), involving fewer associated nodes and simpler association logic, leading to a shorter node association adaptation time. A higher response time for node structure adjustments also indicates that the adjustment task is only a local and simple optimization (such as adjusting the order of a small number of nodes or modifying the attributes of a single node), corresponding to a smaller amount of association adaptation tasks and simpler logic, requiring no long-term verification, resulting in a shorter node association adaptation time. Meanwhile, the path matching correction value is positively correlated with the adaptability to internal and external dynamic changes. The higher the path matching correction value, the more accurately it can capture the optimization needs of internal learning effect data feedback (such as the deviation of users' knowledge point mastery) and the changes in external social and workplace ability requirements (such as the requirement for new digital skills), and adjust the path design in advance so that the recommended path always maintains a high degree of fit with dynamic needs, and the higher the adaptability to internal and external dynamic changes. The node structure adjustment response speed is also positively correlated with the adaptability to internal and external dynamic changes. The higher the node structure adjustment response speed, the faster the node adjustment can be completed, avoiding the expansion of deviations. Externally, when new capability requirements (such as low-carbon management skills) appear, a fast response speed can quickly add corresponding nodes and adjust the node hierarchy relationship, so that the system capability framework can quickly adapt to external changes, and the higher the adaptability to internal and external dynamic changes. The node association adaptation time is negatively correlated with the adaptability to internal and external dynamic changes. The longer the node association adaptation time, the more likely the adjusted nodes will be incompatible with the existing system for a long time, resulting in the "node isolation" problem, which cannot solve internal deviations or meet external needs, and the lower the adaptability to internal and external dynamic changes.

[0053] Finally, the change adaptation control unit is used to compare the internal and external dynamic change adaptation with the change adaptation benchmark value. If the internal and external dynamic change adaptation is greater than or equal to the change adaptation benchmark value, the change adaptation adaptive adjustment is not performed. Otherwise, the analysis cycle optimization and priority determination efficiency optimization are performed according to the change adaptation correction amount. The change adaptation correction amount represents the negative difference between the internal and external dynamic change adaptation and the change adaptation benchmark value.

[0054] It should be further explained that the specific steps for performing analysis cycle optimization are as follows: Collect full lifecycle data of historical closed-loop capability nodes. The data for each capability node should include at least: identification timestamp, requirement freeze timestamp, and development and launch completion timestamp. Periodically calculate and update global key indicators, including: the average analysis cycle from identification to requirement freeze, the average implementation cycle from requirement freeze to development and launch, and the total average launch cycle of the sum of the two, and calculate the standard deviation of each period indicator.

[0055] When a new capability requirement identification signal is received, the system matches a preset priority determination efficiency mode based on the current total average uptime, uptime standard deviation, and change adaptability correction amount. Specifically: If the total average online period is greater than the upper limit of the online period reference, then the in-depth analysis mode is matched; If the total average online period is within the online period reference range, then the standard agile judgment mode is matched. The online period reference range represents the closed interval formed by the lower limit of the online period reference and the upper limit of the online period reference. If the total average online period is less than the lower limit of the online period reference, then the fast channel determination mode is matched.

[0056] Optimization of the analysis cycle also includes: When matching the deep analysis mode, the current total average online period, online period standard deviation, and change adaptability correction amount are input into a predefined analysis period mapping table, and the analysis period gain coefficient is output. The current average analysis period and the analysis period gain coefficient are combined to obtain the target average analysis period, so as to generate highly reliable and resistant priority decisions. When matching the standard agile judgment mode, the current average analysis cycle is maintained, and priority judgment efficiency optimization is not performed; When matching the fast channel determination mode, the current total average online period, online period standard deviation, and change adaptation correction amount are input into a predefined analysis period mapping table, and the analysis period suppression coefficient is output. The current average analysis period and the analysis period suppression coefficient are combined to obtain the target average analysis period. In the face of rapidly changing or highly uncertain environments, decision iteration is completed first.

[0057] It should be understood that, firstly, continuous collection of full lifecycle data for historically closed-loop capability nodes (including key timestamps such as identification, requirement freezing, and development / deployment completion) is performed, and global key indicators are periodically calculated and updated. These include the average analysis cycle from identification to requirement freezing, the average implementation cycle from requirement freezing to development / deployment, the total average deployment cycle (the sum of the two), and the standard deviation of each cycle indicator. This constructs an objective and quantitative organizational delivery performance measurement system that not only reflects the overall delivery speed but also reveals its stability through standard deviation, providing a data foundation for dynamic decision-making.

[0058] When a new capability requirement identification signal is received, the system intelligently matches a preset priority judgment efficiency mode based on the current total average deployment cycle, the standard deviation of the deployment cycle, and the real-time calculated change adaptation adjustment: if the cycle is greater than the reference upper limit, the system matches the in-depth analysis mode; if the cycle is within the normal reference range, the system matches the standard agile judgment mode; if the cycle is less than the reference lower limit or the standard deviation is too high (indicating a fast delivery pace or large fluctuations), the system matches the fast track judgment mode. The technical effect of this matching mechanism is that it aligns the rigor of the requirement judgment process with the organization's actual delivery capabilities in real time, avoiding resource misallocation caused by hasty decisions when delivery is slow, or innovation bottlenecks caused by lengthy processes when delivery is agile.

[0059] Subsequently, a differentiated decision-making process is implemented: In the in-depth analysis mode, the gain coefficient is output through the analysis cycle mapping table, and the analysis cycle is appropriately extended to support multiple rounds of review and detailed demonstration, aiming to generate highly reliable and resistant priority decisions and ensure that limited resources are invested in high-value needs; in the standard agile mode, the current cycle is maintained to balance quality and speed; in the fast track mode, the inhibition coefficient is output through the mapping table to significantly compress the analysis cycle and prioritize the completion of decision iterations to quickly respond to market changes or uncertainties.

[0060] In this embodiment, demand determination is transformed from a fixed process into a flexible adjustment process based on feedback from the organization's actual throughput capacity. This achieves a dynamic balance between demand flow rate and backend delivery capacity, which not only significantly improves the end-to-end flow efficiency from demand identification to value realization, but also systematically reduces the risk of resource waste by avoiding mismatch between processes and capabilities, thereby enhancing the organization's strategic response agility and return on investment in dynamic environments.

[0061] It is necessary to understand that, such as Figure 3The diagram shows a priority determination efficiency optimization flowchart for a multi-dimensional growth tree construction system based on knowledge point diagnosis provided in this application embodiment. The specific process is as follows: First, the target rule requiring gray-scale verification is obtained, and metadata, including a preset iteration cycle and iteration granularity coefficient, is extracted. Next, the core decision point is comparing the extracted iteration cycle with a preset reference value. If the rule iteration cycle is less than or equal to the reference value, the current gray-scale verification cycle is maintained without adjustment, and the process ends. Conversely, if the rule iteration cycle is greater than the reference value, the optimization process is required. The system takes the iteration cycle, granularity coefficient, and a change adaptation correction amount as input and queries a preset mapping table. The mapping table calculates and outputs a "verification cycle coefficient" based on these inputs. Finally, the current verification cycle is multiplied by this coefficient to calculate the new, optimized target gray-scale verification cycle, and the update is completed.

[0062] The specific steps for prioritizing and optimizing efficiency are as follows: Obtain the metadata definition of the target evolution rule to be grayscale verified, and extract its preset iteration cycle. The preset iteration cycle is the maximum time interval inherent in the rule design logic that forces it to update its content or parameters. Its representation forms include: the standard time required to complete a complete level stage in the growth tree system, the forced refresh cycle of the global rule base in the path recommendation system, or the reset cycle of statistical rules based on time windows; extract its iteration granularity coefficient from the rule definition. The iteration granularity coefficient is used to quantify the allowable rule change range in a single iteration. If the iteration cycle of the evolution rule is less than or equal to the reference value of the iteration cycle, then the gray-scale verification cycle optimization will not be performed, and the current rule gray-scale verification cycle will be maintained. If the iteration cycle of the evolution rule is greater than the reference value of the iteration cycle, the preset iteration cycle, iteration granularity coefficient, and change adaptability correction amount are input into a predefined iteration cycle-verification cycle mapping table, and the output is the verification cycle coefficient assigned to the rule. The current rule grayscale verification cycle is combined with the verification cycle coefficient assigned to the rule to obtain the target rule grayscale verification cycle. The iteration cycle-verification cycle mapping table is configured such that: the longer the preset iteration cycle of the rule, the longer the theoretical value of the grayscale verification cycle assigned to it; the greater the iteration change amplitude of the rule, the shorter the theoretical value of the grayscale verification cycle assigned to it.

[0063] It should be understood that obtaining the metadata definition of the evolution rule of the target to be verified involves extracting its preset iteration cycle (i.e., the maximum interval of forced updates in the rule logic, such as the duration of the growth tree system level stage, the forced refresh cycle of the rule base, etc.) and iteration granularity coefficient (quantifying the magnitude of a single rule change). The technical effect of this step is that the system captures the two key attributes of the rule's inherent evolution rhythm and change intensity from the source of rule design, providing a fundamental basis for subsequent precise control of verification resources. If the iteration cycle is less than or equal to the reference threshold, it indicates that the rule iterates frequently and has a short lifespan. In this case, the current verification cycle is maintained to avoid unnecessary process extension for high-frequency iterative rules, thus ensuring the experimental iteration speed. If the iteration cycle is greater than the reference threshold, the optimization process is initiated. In this process, the system inputs the preset iteration cycle, iteration granularity coefficient, and real-time calculated change adaptation correction amount into a predefined iteration cycle-verification cycle mapping table. The core configuration logic of this mapping table is: the longer the preset iteration cycle of a rule, the longer the theoretical value of the gray-scale verification cycle allocated to it; the greater the magnitude of the rule's iterative change, the shorter the theoretical value of the gray-scale verification cycle allocated to it. Based on this logic, the mapping table outputs a verification cycle coefficient, which the system combines with the current verification cycle to obtain the final optimized target rule grayscale verification cycle.

[0064] In this embodiment, an adaptive optimal match is achieved between verification resource investment and the risks and value of rule evolution: for stable rules with slow iteration and small changes, a longer verification period is allocated to allow for sufficient observation and robustness testing, ensuring the reliability of changes; for highly evolving rules with fast iteration or large changes, the verification period is shortened (especially when changes are large) and supplemented with more intensive monitoring, so as to quickly verify the core value while strictly controlling the risk exposure window. This transforms the verification duration from a fixed value to an elastic variable dynamically calculated based on the rule's inherent attributes, thereby maximizing the overall efficiency of rule optimization and the continuity of business value delivery while ensuring system stability, and enabling the evolution process of the system rule base itself to have intelligent adaptive optimization capabilities.

[0065] like Figure 4The diagram shown is a flowchart of a multi-dimensional growth tree construction method based on knowledge point diagnosis provided in this application embodiment. Specifically, it includes: acquiring and processing user learning data; generating mastery status parameters for each knowledge point in a preset knowledge system through multi-dimensional analysis; based on the mastery status parameters, using the root knowledge point of the preset knowledge system as the root node of the growth tree, and each level of knowledge point as a child node, mapping the quantitative indicators corresponding to each growth dimension to multi-dimensional attribute vectors of each node, and constructing directed weighted edges between nodes according to the preorder relationship and association strength of the preset knowledge system, thereby forming a structured initial growth tree; and quantifying the adaptation process of the recommended path and learner demand response. The system quantifies the degree of personalized path matching to determine whether to perform adaptive adjustments based on the personalized path matching degree. This aims to achieve the best dynamic fit between the user's learning path and their personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization. It also quantifies the responsiveness to changes in internal learning performance data and external social and workplace competency requirements, as well as the degree of cross-platform interoperability. This yields the degree of adaptability to internal and external dynamic changes. Based on this degree of adaptability, the system determines whether to perform adaptive adjustments to achieve continuous and stable optimization and goal convergence under internal state evolution and external environmental disturbances, thus ensuring the effectiveness of growth path recommendations.

[0066] In this embodiment, a fundamental leap from static knowledge transmission to dynamic capability development is achieved by constructing an intelligent learning framework driven by an individual's real-time cognitive state and capable of co-evolving with the internal and external environment. Specifically, the user's learning behavior is first transformed into multi-dimensional, fine-grained knowledge point mastery status parameters, and a dynamic growth tree is constructed based on this, reflecting both the logic of the knowledge system and accurately depicting the individual's cognitive state. Furthermore, the learning path is continuously evaluated and dynamically optimized based on this growth tree, ensuring a high degree of alignment with the user's current cognitive level, learning pace, and long-term development goals, thereby significantly improving learning efficiency and the depth of knowledge internalization. More importantly, by monitoring the evolution of internal learning effects and changes in external capability requirements in real time, the recommendation logic is dynamically adjusted to not only adapt to the individual user's growth trajectory but also synchronously respond to changes in social and workplace demands and the interoperability requirements of cross-platform ecosystems. Ultimately, this ensures the long-term effectiveness, environmental robustness, and strategic foresight of the recommended growth path, forming a personalized lifelong learning support system with self-evolving capabilities.

[0067] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A multi-dimensional growth tree construction system based on knowledge point diagnosis, characterized in that, include: The module includes a knowledge point diagnosis module, a growth tree topology construction module, an adaptive path planning module, and a module for handling internal and external dynamic changes. The knowledge point diagnosis module is used to acquire and process the user's learning data, and generate mastery status parameters of the user's knowledge points in the preset knowledge system through multi-dimensional analysis. The growth tree topology construction module is used to map the quantitative indicators corresponding to each growth dimension to the multi-dimensional attribute vector of each node based on the mastery state parameters, with the root knowledge point of the preset knowledge system as the root node of the growth tree, and the knowledge points at each level as child nodes. It also constructs directed weighted edges between nodes according to the preorder relationship and association strength of the preset knowledge system, thereby forming a structured initial growth tree. The adaptive path planning module is used to quantify the degree of fit between the recommended path and the learner's needs and response, so as to obtain the personalized path matching degree. Based on the personalized path matching degree, it determines whether to perform adaptive adjustment of the path matching degree to achieve the best dynamic fit between the user's learning path and personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization effect. The module for handling internal and external dynamic changes is used to quantify the response efficiency to changes in internal learning effect data optimization needs and external social workplace ability needs, as well as the degree of cross-platform interoperability adaptation. It obtains the degree of adaptation to internal and external dynamic changes and determines whether to perform adaptive adjustment based on the degree of adaptation to achieve continuous and stable optimization and target convergence under internal state evolution and external environmental disturbances, thereby ensuring the effectiveness of growth path recommendation.

2. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 1, characterized in that, The adaptive path planning module includes: an adaptive path planning quantization unit and a path matching degree control unit; The adaptive path planning quantization unit is used to obtain a personalized path matching degree based on the matching degree parameters, wherein the matching degree parameters include the planned path compliance rate, the path adjustment frequency, and the deviation path regression time. The compliance rate compensation value is obtained by multiplying the compliance rate ratio factor by the result of the analysis of the proportion of the planned path compliance rate to the compliance rate benchmark value. The adjustment frequency compensation value is obtained by multiplying the adjustment frequency scaling factor by the adjustment frequency benchmark value and the result of the path adjustment frequency ratio analysis. The regression duration ratio factor is multiplied by the regression duration baseline value and the result of the deviation path regression duration ratio analysis to obtain the regression duration compensation value. The compliance rate compensation value, adjustment frequency compensation value, and regression duration compensation value are superimposed to obtain the personalized path matching degree; The path matching degree adjustment unit is used to compare the personalized path matching degree with the path matching degree benchmark value. If the personalized path matching degree is greater than or equal to the path matching degree benchmark value, the path matching degree adaptive adjustment is not performed. Otherwise, the interaction behavior frequency optimization and dwell time ratio optimization are performed according to the matching degree correction amount. The matching degree correction amount represents the degree of negative deviation between the personalized path matching degree and the path matching degree benchmark value.

3. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 2, characterized in that, The specific process for optimizing the frequency of interactive behavior is as follows: S100, in response to the continuous interaction sequence generated within the current session window, extracts the knowledge point identifiers associated with all interaction behaviors in the continuous interaction sequence in real time, and constructs and dynamically updates a real-time knowledge point density map within the session based on a preset time decay function and spatial proximity weight; wherein, the time decay function is used to give the knowledge points interacted within the current session window an initial weight, and the spatial proximity weight is used to give weight gain to the knowledge points co-occurring in neighboring interaction behaviors based on the semantic correlation of the knowledge points in the preset knowledge topology; S101, perform cluster analysis on the real-time knowledge point density map within the session to identify knowledge point clusters whose cumulative knowledge point weights are greater than the knowledge point cluster benchmark value, and mark them as high-frequency interactive knowledge point clusters in the current session; calculate the comprehensive density value of the high-frequency interactive knowledge point cluster, where the comprehensive density value represents a function of the sum of the weights of all knowledge points in the knowledge point cluster, the number of knowledge points in the cluster, and the average correlation weight between knowledge points in the cluster.

4. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 3, characterized in that, The optimization of the frequency of execution interaction behavior also includes: S102, the comprehensive density value of the high-frequency interactive knowledge point cluster is mapped to the corresponding interactive behavior frequency adjustment parameter through a nonlinear mapping function; the nonlinear mapping function is configured as follows: when the comprehensive density value is lower than the lower limit of the knowledge point cluster reference value, an interactive behavior frequency gain coefficient is output based on the matching degree correction amount and the current comprehensive density value, and the current user interactive behavior frequency and the interactive behavior frequency gain coefficient are combined to obtain the next user interactive behavior frequency; when the comprehensive density value is within the knowledge point cluster reference interval, the current user interactive behavior frequency is maintained, and no interactive behavior frequency optimization is performed; when the comprehensive density value is higher than the upper limit of the knowledge point cluster reference value, an interactive behavior frequency suppression coefficient is output based on the matching degree correction amount and the current comprehensive density value, and the current user interactive behavior frequency and the interactive behavior frequency suppression coefficient are combined to obtain the next user interactive behavior frequency, wherein the knowledge point cluster reference interval represents the closed interval formed by the lower limit of the knowledge point cluster reference value and the upper limit of the knowledge point cluster reference value.

5. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 2, characterized in that, The specific process for optimizing the dwell time ratio is as follows: S201, parse the recommended path generated for the current user, the recommended path contains task nodes arranged in order; calculate the expected transfer interval between every two adjacent task nodes to form an interval duration sequence; based on the interval duration sequence, identify and extract path-level temporal features; S202, input the path-level temporal features and matching degree correction amount into a predefined interval duration-stay duration ratio mapping table. The interval duration-stay duration ratio mapping table is configured to output a global stay duration adjustment coefficient and a local enhancement coefficient for stay nodes near long interval clusters. The long interval cluster is defined as a set of interval durations that are cumulatively higher than a preset interval threshold. S203, Based on the coefficients output in step S201, the total reserved stay time budget in the recommended path is reallocated: Based on the global dwell time adjustment coefficient, the total dwell time of the path baseline is scaled to determine the dynamically adjusted total dwell time available for the path. Within the dynamically adjusted total dwell time available for the path, the individual dwell time allocation of the dwell nodes in the adjacent long-interval clusters is weighted and gained according to the local enhancement coefficient to ensure that users have relatively more preparation time after long movement intervals.

6. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 1, characterized in that, The module for handling internal and external dynamic changes includes: a quantification unit for internal and external dynamic changes and a change adaptation control unit. The internal and external dynamic change quantification unit is used to obtain the internal and external dynamic change adaptability based on the dynamic change parameters, wherein the dynamic change parameters include path matching correction value, node structure adjustment response speed and node association adaptation time. The path matching compensation value is obtained by multiplying the results of the path matching ratio factor and the path matching correction value by the path matching baseline value. The response speed compensation value is obtained by multiplying the response speed proportional factor by the result of the analysis of the ratio of the node structure adjustment response speed to the response speed benchmark value. The adaptation time ratio factor is multiplied by the adaptation time baseline value and the results of the node-related adaptation time ratio analysis to obtain the adaptation time compensation value. The path matching compensation value, response speed compensation value, and adaptation time compensation value are superimposed to obtain the degree of adaptation to internal and external dynamic changes. The change adaptability control unit is used to compare the internal and external dynamic change adaptability with the change adaptability benchmark value. If the internal and external dynamic change adaptability is greater than or equal to the change adaptability benchmark value, then no adaptive adjustment of change adaptability is performed. Otherwise, the analysis cycle optimization and priority determination efficiency optimization are performed according to the change adaptability correction amount. The change adaptability correction amount represents the degree of negative deviation between the internal and external dynamic change adaptability and the change adaptability benchmark value.

7. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 6, characterized in that, The specific steps for optimizing the execution analysis cycle are as follows: Collect full lifecycle data of historical closed-loop capability nodes, periodically calculate and update global key indicators, including: the average analysis cycle from identification to requirement freezing, the average implementation cycle from requirement freezing to development and launch, and the total average launch cycle of the sum of the two, and calculate the standard deviation of each cycle indicator. When a new capability requirement identification signal is received, the system matches a preset priority determination efficiency mode based on the current total average uptime, uptime standard deviation, and change adaptability correction amount. Specifically: If the total average online period is greater than the upper limit of the online period reference, then the in-depth analysis mode is matched; If the total average online period is within the online period reference range, then the standard agile judgment mode is matched. The online period reference range represents the closed interval formed by the lower limit of the online period reference and the upper limit of the online period reference. If the total average online period is less than the lower limit of the online period reference, then the fast channel determination mode is matched.

8. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 7, characterized in that, The optimization of the execution analysis cycle also includes: When matching the deep analysis mode, the current total average online period, online period standard deviation, and change adaptability correction amount are input into a predefined analysis period mapping table, and the analysis period gain coefficient is output. The current average analysis period and the analysis period gain coefficient are combined to obtain the target average analysis period, so as to generate highly reliable and resistant priority decisions. When matching the standard agile judgment mode, the current average analysis cycle is maintained, and priority judgment efficiency optimization is not performed; When matching the fast channel determination mode, the current total average online period, online period standard deviation, and change adaptation correction amount are input into a predefined analysis period mapping table, and the analysis period suppression coefficient is output. The current average analysis period and the analysis period suppression coefficient are combined to obtain the target average analysis period. In the face of rapidly changing or highly uncertain environments, decision iteration is completed first.

9. The multi-dimensional growth tree construction system based on knowledge point diagnosis as described in claim 6, characterized in that, The specific steps for optimizing the priority determination efficiency are as follows: Obtain the metadata definition of the target evolution rule to be grayscale verified, extract its preset iteration cycle from it, and extract its iteration granularity coefficient from the rule definition. The iteration granularity coefficient is used to quantify the rule change range allowed in a single iteration. If the iteration cycle of the evolution rule is less than or equal to the reference value of the iteration cycle, then the gray-scale verification cycle optimization will not be performed, and the current rule gray-scale verification cycle will be maintained. If the iteration cycle of the evolution rule is greater than the reference value of the iteration cycle, the preset iteration cycle, iteration granularity coefficient, and change adaptability correction amount are input into a predefined iteration cycle-verification cycle mapping table, and the output is the verification cycle coefficient assigned to the rule. The current rule grayscale verification cycle is combined with the verification cycle coefficient assigned to the rule to obtain the target rule grayscale verification cycle. The iteration cycle-verification cycle mapping table is configured such that: the longer the preset iteration cycle of the rule, the longer the theoretical value of the grayscale verification cycle assigned to it; the greater the iteration change amplitude of the rule, the shorter the theoretical value of the grayscale verification cycle assigned to it.

10. The method for constructing a multi-dimensional growth tree based on knowledge point diagnosis as described in any one of claims 1-9, characterized in that, include: Acquire and process user learning data, and generate mastery status parameters of users for each knowledge point in the preset knowledge system through multi-dimensional analysis; Based on the mastery of state parameters, the root knowledge point of the preset knowledge system is used as the root node of the growth tree, and the knowledge points at each level are used as child nodes. The quantitative indicators corresponding to each growth dimension are mapped to the multi-dimensional attribute vectors of each node. The directed weighted edges between nodes are constructed according to the preorder relationship and correlation strength of the preset knowledge system, thereby forming a structured initial growth tree. The degree of fit between the recommended path and the learner's needs is quantified to obtain the personalized path matching degree. Based on the personalized path matching degree, it is determined whether to perform adaptive adjustment of the path matching degree to achieve the best dynamic fit between the user's learning path and personal cognitive state and development goals, thereby maximizing learning efficiency and knowledge internalization effect. The system quantifies the response efficiency to changes in internal learning effectiveness data optimization needs and external social and workplace competency requirements, as well as the degree of cross-platform interoperability adaptation. This yields the degree of adaptation to internal and external dynamic changes. Based on this degree of adaptation, the system determines whether to perform adaptive adjustments to achieve continuous and stable optimization and goal convergence under internal state evolution and external environmental disturbances, thereby ensuring the effectiveness of growth path recommendations.

Images