Autonomous learning auxiliary system based on Chinese professional education

Abstract

The invention discloses an autonomous learning auxiliary system based on Chinese professional education, and the system comprises a multi-source data collection module which is used for collecting user learning behavior data, user knowledge mastering data and standardized Chinese learning resource data in a Chinese professional autonomous learning scene; the Chinese knowledge graph construction module is used for integrating Chinese professional knowledge points in the fields of Chinese characters, grammar, literature and writing, and has the advantages that by means of the Chinese knowledge graph construction module, a rule extraction method can be adopted for structured resources, a BERT entity recognition model can be adopted for unstructured resources, the knowledge points in the fields of Chinese characters, grammar, literature and writing can be accurately covered, and the knowledge points in the fields of Chinese characters, grammar, literature and writing can be accurately recognized. Then the association relationship of three types of semantics, logic and culture is defined, association strength is quantified by combining teacher annotation and text co-occurrence analysis, hierarchy is divided according to a'basic layer, a hierarchical layer and a research layer ', and scattered knowledge is connected in series to form a'node, association and hierarchy' complete network.

Description

Technical Field

[0001] This invention relates to the field of Chinese language education technology, specifically to a self-learning support system based on Chinese language education. Background Technology

[0002] In the field of Chinese language education, self-directed learning is a key link for students to build a knowledge system and improve their professional skills. With the advancement of educational informatization, various Chinese learning platforms have emerged, but existing technologies are insufficient to meet the needs of Chinese language learning for "systematic, personalized, and in-depth" learning.

[0003] First, the fragmentation of knowledge systems is a prominent issue, lacking structured connections. Chinese professional knowledge covers multiple fields such as character recognition, grammar rules, literary analysis, and professional writing. There are close logical, semantic, and cultural connections between knowledge points in each field. However, existing learning platforms mostly present content in the form of "single knowledge point resources," without systematically integrating and modeling the connections between knowledge points. This makes it difficult for students to form a complete knowledge network during the learning process, thus affecting their in-depth understanding and flexible application of knowledge. Second, learning paths are highly homogenized. Different students have significant differences in their Chinese knowledge foundation and learning style preferences, but existing platforms mostly adopt a "uniform course package plus fixed learning order" model. This model neither accurately pushes remedial content to students' weak knowledge points nor matches resource types with learning style preferences. As a result, students often face the problem of "learning knowledge they already know" but "using an unsuitable format" in self-study, significantly reducing learning efficiency and enthusiasm.

[0004] Furthermore, the evaluation of learning outcomes is limited by a single dimension and lacks a dynamic optimization mechanism. Current assessments of the effectiveness of self-directed learning in Chinese majors rely heavily on the "accuracy rate of objective questions," which only reflects students' mastery of basic knowledge points but neglects the core dimension of "subjective professional competence" in Chinese majors. At the same time, existing platforms do not provide real-time identification and intervention for abnormal student learning behaviors, nor can they dynamically adjust subsequent learning plans based on the learning outcomes at each stage. This makes it difficult to resolve problems that arise during the learning process in a timely manner, creating a vicious cycle of "problem accumulation - declining effectiveness." To address this, we propose a self-directed learning support system based on Chinese major education. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] This invention solves the above-mentioned technical problems by adopting the following technical solution: It provides a self-learning assistance system based on Chinese professional education, comprising:

[0007] The multi-source data acquisition module is used to collect user learning behavior data, user knowledge acquisition data, and standardized Chinese learning resource data in the context of self-directed learning in Chinese majors.

[0008] The Chinese knowledge graph construction module is used to integrate Chinese professional knowledge points in the fields of Chinese characters, grammar, literature, and writing to build a knowledge network that includes "knowledge point nodes, relationships, and hierarchical structure";

[0009] The personalized learning push module is used to calculate the user's mastery of knowledge points and learning style preferences through a quantitative model based on the user data collected by the multi-source data acquisition module and the knowledge network generated by the Chinese knowledge graph construction module, and to generate learning content and customized learning paths that are adapted to the user's weaknesses and preferences.

[0010] The learning effectiveness evaluation and optimization module is used to evaluate users' learning effectiveness from two dimensions: objective knowledge mastery ability and subjective professional competence. Based on the evaluation results, the learning plan is dynamically adjusted to form a self-learning closed loop of "data collection, knowledge modeling, content push, and effect optimization".

[0011] Preferably, the user learning behavior data collected by the multi-source data acquisition module specifically includes: learning format selection records, single browsing time and total time of Chinese knowledge points, answering time and submission interval of practice questions, Chinese professional keywords manually marked by the user in the notes, and Chinese professional questions actively submitted by the user.

[0012] Preferably, the user knowledge acquisition data collected by the multi-source data acquisition module specifically includes: answer records of Chinese character recognition and phonetic transcription, answer records of ancient Chinese function words and modern Chinese syntax application, literary works appreciation texts of no less than 300 words, and Chinese major writing assignments, and the writing assignments are accompanied by user modification traces and teacher annotation records.

[0013] Preferably, the construction process of the Chinese knowledge graph construction module includes:

[0014] S11. Knowledge point extraction: Rule extraction method is used for structured Chinese learning resources; BERT entity recognition model is used for unstructured Chinese learning resources.

[0015] S12. Relationship Construction: Define semantic relationship, logical relationship, and cultural relationship. Quantify the relationship strength through annotation by Chinese professional teachers and text co-occurrence analysis. The value range is 0 to 1. A relationship strength greater than or equal to 0.8 is considered a strong relationship.

[0016] S13. Hierarchical division: The basic level includes Chinese character recognition, basic grammar, and literary common sense; the advanced level includes advanced grammar and literary techniques; the research level includes academic viewpoints and cross-work comparisons.

[0017] Preferably, the formula for the knowledge point mastery measurement model in the personalized learning push module is:

[0018] In the formula, For users to learn Chinese professional knowledge points Mastery level (values ​​range from 0 to 100). For knowledge points Total number of practice sessions For knowledge points , For users to understand knowledge points The depth of interaction is scored as follows: detailed notes = 10 points, browsing only = 3 points, no interaction = 0 points. The maximum score for interaction depth is 10 points by default.

[0019] The personalized learning recommendation module is based on Push notification content: When At that time, it pushes basic knowledge point analysis documents and introductory-level practice questions; when At that time, push advanced case study videos and reinforcement exercises for knowledge points; when At that time, push knowledge points from the knowledge graph Materials on strongly related extended knowledge points.

[0020] Preferably, the personalized learning recommendation module further includes learning format preference calculation and content matching degree calculation, specifically:

[0021] S21. Calculation of learning format preference weights: using the following formula:

[0022]

[0023] in, For users' learning formats The preference weights, wherein the weights range from 0 to 100. Corresponding text format Corresponding video format, Corresponding audio format, The cumulative time a user spends using the corresponding learning format;

[0024] S22. Content matching degree calculation: The formula is as follows:

[0025]

[0026] in, For Chinese learning content The matching degree with the user, wherein the matching degree ranges from 0 to 100. For content Preference weights for different learning formats;

[0027] The personalized learning push module prioritizes push notifications. The learning content, when multiple contents When scores are the same, the score is determined by... Push notifications in descending order.

[0028] Preferably, the learning path planning steps of the personalized learning push module include:

[0029] S31. Determine the starting point for learning: using the formula:

[0030]

[0031] Calculation results These knowledge points serve as the user's initial learning starting point;

[0032] S32. Plan the learning order: Arrange the learning nodes according to the logical order in step S12 where the association strength is greater than or equal to 0.8. The user must complete the learning of the current node, and the node must be learned through the formula:

[0033]

[0034] Post-learning mastery of calculation Only after completing the above steps can you proceed to the next learning stage;

[0035] S33. Adapting to Learning Format: Based on the learning format preference weight formula:

[0036] ,

[0037] Calculated user preference weights ,choose The highest-ranking learning format matches the learning content format of the corresponding node; if the preference weight for video format is... If so, then a knowledge point analysis video will be matched for that node;

[0038] S34. Dynamically Adjust Path: If a user's learning node is accessed via the formula:

[0039]

[0040] Post-learning mastery of calculation If so, return to the basic content of that node and relearn; If so, the learning access to advanced knowledge points in the Chinese knowledge graph that are strongly associated with that node will be unlocked in advance.

[0041] Preferably, the comprehensive score evaluation formula for the learning effect evaluation and optimization module is as follows:

[0042]

[0043] In the formula, The user's overall learning score ranges from 0 to 100. This represents the total number of knowledge points covered in this learning cycle. For the knowledge points after learning The degree of mastery; Pre-learning knowledge points The degree of mastery, The subjective professional competence score ranges from 0 to 10. The subjective professional competence score is derived by analyzing the key point coverage of the user's literary appreciation text and the logical completeness of the writing assignment using NLP technology, combined with the annotation results of Chinese major teachers.

[0044] Preferably, the learning effect evaluation and optimization module also includes abnormal behavior judgment and scheme adjustment, specifically using an abnormal threshold formula to dynamically adjust the judgment criteria for answering behavior, the formula being:

[0045]

[0046] in, for An abnormal answer threshold is defined, with the value ranging from 0 to 60 seconds. for Threshold for abnormal answering at any time; for Time spent answering questions by users; This represents the average time spent answering questions on this knowledge point. It is a symbolic function;

[0047] when, When the answer is deemed too long, a lightweight summary text of key knowledge points will be sent; when... If the answer is deemed too short, a document summarizing the key knowledge points will be sent to the user.

[0048] Simultaneously based on and Optimization solution: When At that time, it pushes basic knowledge point analysis and introductory questions; when At that time, push out examples of easily mistaken knowledge points and advanced questions; when At that time, push academic materials at the research level; when At that time, a manual on literary appreciation methods and writing templates will be pushed out; when At the same time, it pushes cross-work comparative analysis data.

[0049] Compared with existing technologies, this invention provides a self-learning assistance system based on Chinese professional education, which has the following beneficial effects:

[0050] 1. This invention utilizes a Chinese knowledge graph construction module to extract structured resources using rule-based methods and employs the BERT entity recognition model for unstructured resources. This accurately covers knowledge points in the fields of Chinese characters, grammar, literature, and writing. Furthermore, by defining three types of relationships—semantic, logical, and cultural—and combining teacher annotations with text co-occurrence analysis to quantify the strength of these relationships, and dividing the knowledge into "basic, advanced, and research" levels, it connects scattered knowledge into a complete network of "nodes, relationships, and levels." This allows students to clearly perceive the internal connections between knowledge points, solves the problem of fragmented knowledge, and helps build a systematic Chinese professional knowledge system.

[0051] 2. This invention comprehensively collects learning behavior and knowledge mastery data through a multi-source data acquisition module, and the personalized push module accurately adapts through a quantitative model. Based on the knowledge point mastery formula, it divides the push gradient into "basic-advanced-expansion" levels, combines the learning format preference weight formula to match text, video, and audio formats, and plans the learning path according to the strong correlation logic of knowledge points. It dynamically adjusts in real time according to the mastery level, solving the problem of "learning knowledge that has already been mastered" but "using an unsuitable format", thereby improving learning efficiency and motivation.

[0052] 3. This invention achieves a dual-dimensional assessment of learning effectiveness by quantifying objective improvements through changes in knowledge mastery and introducing subjective professional competence scoring. Furthermore, it utilizes a comprehensive scoring formula to realize this "objective plus subjective" approach. In terms of intervention, it uses an anomaly threshold formula to identify abnormal answers in real time and develops tiered optimization plans based on the comprehensive score and subjective assessment. The assessment results are fed back to the push module to dynamically adjust the learning plan, thus promptly addressing learning problems, preventing "problem accumulation leading to declining effectiveness," and ensuring a steady improvement in self-directed learning outcomes. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the system module operation of the present invention;

[0054] Figure 2 This is a schematic diagram illustrating the construction process of the Chinese knowledge graph construction module of the present invention;

[0055] Figure 3 This is a schematic diagram of the learning path planning for the personalized learning push module of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Please see Figure 1-3 A self-directed learning support system based on Chinese language professional education, comprising:

[0058] The multi-source data acquisition module is used to collect user learning behavior data, user knowledge acquisition data, and standardized Chinese learning resource data in the context of self-directed learning in Chinese majors.

[0059] The Chinese knowledge graph construction module is used to integrate Chinese professional knowledge points in the fields of Chinese characters, grammar, literature, and writing to build a knowledge network that includes "knowledge point nodes, relationships, and hierarchical structure";

[0060] The personalized learning push module is used to calculate the user's mastery of knowledge points and learning style preferences through a quantitative model based on the user data collected by the multi-source data acquisition module and the knowledge network generated by the Chinese knowledge graph construction module, and to generate learning content and customized learning paths that are adapted to the user's weaknesses and preferences.

[0061] The learning effectiveness evaluation and optimization module is used to evaluate users' learning effectiveness from two dimensions: objective knowledge mastery ability and subjective professional competence. Based on the evaluation results, the learning plan is dynamically adjusted to form a self-learning closed loop of "data collection, knowledge modeling, content push, and effect optimization".

[0062] In this embodiment, the user learning behavior data collected by the multi-source data acquisition module specifically includes: learning format selection records, single browsing time and total time of Chinese knowledge points, answering time and submission interval of practice questions, Chinese professional keywords manually marked by the user in the notes, and Chinese professional question text actively submitted by the user. The user knowledge mastery data collected by the multi-source data acquisition module specifically includes: Chinese character recognition and phonetic answering records, ancient Chinese function words and modern Chinese syntax application answering records, literary work appreciation text of no less than 300 words, and Chinese professional writing assignments, with the writing assignments accompanied by user modification traces and teacher annotation records.

[0063] Specifically, user learning behavior data includes records of users' selection and switching between text, video, and audio learning formats during their Chinese major studies. It also covers the effective browsing time for each Chinese knowledge point (excluding invalid time such as page pauses and minimization) and the total browsing time within the learning cycle. Furthermore, it includes the time spent answering individual practice questions, the total time spent on major questions, and the submission interval between adjacent questions. Additionally, it collects Chinese major keywords manually marked in users' study notes, such as "causative usage," "the preordained alliance between wood and stone," and "extracting arguments from argumentative essays," as well as keywords actively submitted through the system's question function. The test includes Chinese professional questions covering topics such as Chinese character pronunciation, usage of classical Chinese function words, and analytical approaches to literary works. User knowledge data includes user answer records for exercises related to Chinese character recognition and pronunciation, answer records for questions on understanding classical Chinese function words and applying modern Chinese syntax, analytical texts of no less than 300 words focusing on specific literary works, and completed Chinese professional writing assignments. These assignments must include traces of user revisions (such as records of adding or deleting sentences and adjusting structure) and teacher annotations regarding content logic and accuracy of expression.

[0064] In this embodiment, the construction process of the Chinese knowledge graph construction module includes:

[0065] S11. Knowledge point extraction: Rule extraction method is used for structured Chinese learning resources; BERT entity recognition model is used for unstructured Chinese learning resources.

[0066] S12. Relationship Construction: Define semantic relationship, logical relationship, and cultural relationship. Quantify the relationship strength through annotation by Chinese professional teachers and text co-occurrence analysis. The value range is 0 to 1. A relationship strength greater than or equal to 0.8 is considered a strong relationship.

[0067] S13. Hierarchical division: The basic level includes Chinese character recognition, basic grammar, and literary common sense; the advanced level includes advanced grammar and literary techniques; the research level includes academic viewpoints and cross-work comparisons.

[0068] Specifically, the process begins with knowledge point extraction in step S11. For structured Chinese learning resources such as textbook catalogs and categorized question banks, a rule-based extraction method is used, leveraging pre-defined Chinese professional grammar rules, categorization logic rules, and keyword matching rules to efficiently extract explicitly marked or rule-compliant knowledge points. For unstructured Chinese learning resources such as academic papers and literary appreciation texts, a BERT entity recognition model incorporating corpora from the Chinese professional field (such as a database of classical Chinese function words and a database of literary techniques and terms) is used to accurately extract implicit knowledge points after word segmentation and part-of-speech tagging preprocessing. Next, the process moves to step S12, which involves constructing associations. Semantic associations (such as the association between "reading the Chinese character 'dental caries'" and "the definition of 'dental caries'") and logical associations (such as "basic...") are defined. The study identifies three types of relationships: the connection between "subject-verb-object structure" and "complex syntactic structure," and the cultural connection (such as the connection between "the bold and unrestrained style of Song Dynasty poetry" and "the background of Song Dynasty literati governing the country"). This is further combined with the experience of Chinese language teachers in manually annotating texts and the analysis of text co-occurrence of knowledge points in learning resources (statistical co-occurrence frequency and positional distance) to quantify the strength of the connection in the 0-1 range, with connections ≥0.8 being classified as strong connections. Finally, step S13 involves hierarchical division, based on the cognitive patterns and knowledge depth of Chinese language learners. Knowledge points are categorized into a basic layer (including Chinese character recognition, basic grammar, and literary common sense), an advanced layer (including advanced grammar and literary techniques), and a research layer (including academic viewpoints and cross-work comparisons), forming a clearly structured and interconnected network of Chinese language professional knowledge.

[0069] In this embodiment, the formula for the knowledge point mastery measurement model in the personalized learning push module is:

[0070] In the formula, For users to learn Chinese professional knowledge points Mastery level (values ​​range from 0 to 100). For knowledge points Total number of practice sessions For knowledge points , For users to understand knowledge points The depth of interaction is scored as follows: detailed notes = 10 points, browsing only = 3 points, no interaction = 0 points. This represents the maximum score for interaction depth, where the maximum score is 10 points by default.

[0071] Personalized learning push module based on Push notification content: When At that time, it pushes basic knowledge point analysis documents and introductory-level practice questions; when At that time, push advanced case study videos and reinforcement exercises for knowledge points; when At that time, push knowledge points from the knowledge graph Materials on strongly related extended knowledge points.

[0072] Specifically, in the formula above, the user's knowledge of Chinese professional knowledge points... The mastery score ranges from 0 to 100. This score is calculated by considering multiple key indicators, including knowledge points. The total number of practice sessions refers to the cumulative number of times a user has answered practice questions related to that knowledge point, including questions on Chinese character pronunciation and grammar application. It also includes the user's overall understanding of the knowledge point. The interaction depth metric is divided into different levels based on user learning behavior. If a user takes detailed notes on knowledge points, such as marking core test points and adding personal understanding, they can get 10 points. If they only browse the knowledge point analysis document but do not make any additional notes, they can get 3 points. If they do not click to view the knowledge point-related content and do not have any interaction behavior, they can get 0 points. In addition, there is a maximum score for interaction depth, which is set to 10 points by default. Through the comprehensive calculation of these metrics, the mastery assessment can reflect both the user's practice accumulation and their level of active learning engagement.

[0073] More specifically, the personalized learning recommendation module will also push corresponding content based on the user's mastery level calculated above. When the user understands the knowledge points... If a user's mastery level is less than 60 points, it indicates a weak foundation in that knowledge point. The system will then provide basic explanation documents, such as explanations of basic Chinese character strokes and introductory grammar rules, along with beginner-level practice questions. These questions are relatively easy and focus on the application of basic concepts, such as simple Chinese character pronunciation questions and basic syntax judgment questions, to help users solidify their foundation in that knowledge point. When a user's mastery level is 60 points or higher but less than 80 points, it indicates that the user has grasped the basic content of the knowledge point, but still needs to strengthen their application skills. The system will push advanced case video examples, such as analyzing grammar applications with examples from literary works, and breaking down the usage scenarios of literary techniques through videos. At the same time, it will push reinforcement practice questions. These questions are more difficult and focus on comprehensive application, such as complex sentence rewriting questions and literary technique appreciation questions, to help users deepen their understanding of the knowledge point. When the user's mastery level reaches 80 points or above, it indicates that the user has a solid grasp of the knowledge point, and the system will push relevant knowledge points from the Chinese knowledge graph. Extended knowledge materials with strong correlations, where strong correlation means a correlation strength greater than or equal to 0.8, such as expanding from "characteristics of Tang poetry genres" to "the influence of Tang Dynasty social culture on poetic style" or from "metaphorical rhetoric" to "differences and comparisons of metaphorical techniques in Chinese and foreign literary works," guide users to build a more systematic Chinese professional knowledge system.

[0074] In this embodiment, the personalized learning recommendation module further includes learning format preference calculation and content matching degree calculation, specifically:

[0075] S21. Calculation of learning format preference weights: using the following formula:

[0076]

[0077] in, For users' learning formats The preference weights range from 0 to 100. Corresponding text format Corresponding video format, Corresponding audio format, The cumulative time a user spends using the corresponding learning format;

[0078] S22. Content matching degree calculation: The formula is as follows:

[0079]

[0080] in, For Chinese learning content The degree of match with the user, with a value ranging from 0 to 100. For content Preference weights for different learning formats;

[0081] Personalized learning recommendation module prioritizes push notifications The learning content, when multiple contents When scores are the same, the score is determined by... Push notifications in descending order.

[0082] Specifically, by learning the formula for calculating formal preference weights in step S21, the user's... Regarding learning methods The preference weights, where users prefer different learning formats. Preference weights Values ​​range from 0 to 100. Corresponding text format Corresponding video format, For the corresponding audio format, the calculation is based on the user's cumulative time using the corresponding learning format. Based on this as the core criterion, the user's preference for different learning formats is quantified, and the content matching degree formula in step S22 is used. For Chinese learning content The match score with the user is between 0 and 100. It is content The module prioritizes pushing content based on the preference weight of the learning format. The learning content, if multiple contents If the scores are the same, then the preference weight will be used. Push notifications in descending order.

[0083] In this embodiment, the learning path planning steps of the personalized learning push module include:

[0084] S31. Determine the starting point for learning: using the formula:

[0085]

[0086] Calculation results These knowledge points serve as the user's initial learning starting point;

[0087] S32. Plan the learning order: Arrange the learning nodes according to the logical order in step S12 where the association strength is greater than or equal to 0.8. The user must complete the learning of the current node, and the node must pass the formula:

[0088]

[0089] Post-learning mastery of calculation Only after completing the above steps can you proceed to the next learning stage;

[0090] S33. Adapting to Learning Formats: Based on the learning format preference weighting formula:

[0091] ,

[0092] Calculated user preference weights ,choose The highest-ranking learning format matches the learning content format of the corresponding node; if the preference weight for video format is... If so, then a knowledge point analysis video will be matched for that node;

[0093] S34. Dynamically Adjust Path: If a user's learning node is accessed via the formula:

[0094]

[0095] Post-learning mastery of calculation If so, return to the basic content of that node and relearn; If so, the learning access to advanced knowledge points in the Chinese knowledge graph that are strongly related to that node will be unlocked in advance.

[0096] Specifically, the personalized learning push module's learning path planning unfolds through four consecutive steps. In step S31, when determining the learning starting point, the system uses the aforementioned specific formula to calculate the user's mastery of each Chinese professional knowledge point, selecting knowledge points whose mastery does not meet the basic standard as initial learning nodes. This accurately identifies the user's knowledge gaps, ensuring that learning starts from the weakest link. In step S32, when planning the learning order, learning nodes are arranged based on the strong association logic (association strength greater than or equal to 0.8) from step S12 of the Chinese knowledge graph construction module. Advancement conditions are also set: users must complete the current node's learning and calculate their mastery of that node using the specific formula to be greater than or equal to 80 points before proceeding to the next node, ensuring a sequential learning process. The learning process is progressive and systematic. In step S33, when adapting the learning format, the user's preference weight for text, video, and audio is calculated based on the learning format preference weight formula. The format with the highest weight is selected to match the corresponding node content. If the video format preference weight is greater than or equal to 60, a knowledge point analysis video is matched for that node to match the user's learning habits and improve absorption efficiency. In step S34, when dynamically adjusting the path, the mastery of the node after learning is evaluated in real time using a formula. If the mastery is less than 60 points, the system will return the basic content of that node for the user to relearn. If the mastery is greater than or equal to 90 points, the learning permissions for advanced knowledge points strongly associated with that node in the Chinese knowledge graph are unlocked in advance, flexibly adapting to the user's learning progress and ability and optimizing the learning path.

[0097] In this embodiment, the comprehensive score evaluation formula for the learning effect evaluation and optimization module is as follows:

[0098]

[0099] In the formula, The user's overall learning score, which ranges from 0 to 100. This represents the total number of knowledge points covered in this learning cycle. For the knowledge points after learning The degree of mastery; Pre-learning knowledge points The degree of mastery, Subjective professional competence is scored, with a value ranging from 0 to 10. The subjective professional competence score is derived by analyzing the key points coverage of the user's literary appreciation text and the logical completeness of the writing assignment using NLP technology, combined with the annotation results of Chinese major teachers.

[0100] Specifically, the user's overall learning score can be calculated using the comprehensive score evaluation formula described above. The value ranges from 0 to 100, and its calculation depends on the total number of knowledge points within this learning cycle. (i.e., the total amount of knowledge points in areas such as Chinese characters, grammar, literature, and writing that the user actually learns during this period), and each knowledge point after learning. degree of mastery , corresponding knowledge points before learning degree of mastery and subjective professional competence score Among them, the subjective professional competence score The value ranges from 0 to 10. It is necessary to first analyze the key point coverage of the literary appreciation texts submitted by users (no less than 300 words) using NLP technology (such as whether the core content such as the theme and artistic techniques of the work is accurately extracted) and the logical integrity of Chinese major writing assignments (such as the article structure and the coherence of arguments and evidence). Then, it is combined with the results of manual annotation of appreciation texts and writing assignments by Chinese major teachers. Finally, these parameters are integrated through formula to comprehensively and objectively quantify the user's learning effect, providing core data basis for subsequent dynamic adjustment of learning plans.

[0101] In this embodiment, the learning effect evaluation and optimization module also includes abnormal behavior judgment and scheme adjustment. Specifically, it uses an abnormal threshold formula to dynamically adjust the judgment criteria for answering behavior. The formula is as follows:

[0102]

[0103] in, for The threshold for abnormal answering at any given time is set, and the value of the abnormal answering threshold ranges from 0 to 60 seconds. for Threshold for abnormal answering at any time; for Time spent answering questions by users; This represents the average time spent answering questions on this knowledge point. It is a symbolic function;

[0104] when, When the answer is deemed too long, a lightweight summary text of key knowledge points will be sent; when... If the answer is deemed too short, a document summarizing the key knowledge points will be sent to the user.

[0105] Simultaneously based on and Optimization solution: When At that time, it pushes basic knowledge point analysis and introductory questions; when At that time, push out examples of easily mistaken knowledge points and advanced questions; when At that time, push academic materials at the research level; when At that time, a manual on literary appreciation methods and writing templates will be pushed out; when At the same time, it pushes cross-work comparative analysis data.

[0106] Specifically, precise intervention is achieved by combining dynamic thresholds with multi-dimensional scoring. Firstly, the module dynamically adjusts the criteria for judging answer behavior using an anomaly threshold formula. In the aforementioned formula... abnormal answer threshold at any time Values ​​range from 0 to 60 seconds. Time threshold Based on, combined User answering time Average answering time for this knowledge point and symbolic functions (The threshold is calculated by adjusting the trend of the difference between the answering time and the average time); based on this threshold, when If the answer is deemed too long, the system will push a lightweight summary of key knowledge points to help the user overcome comprehension bottlenecks. If the answer is deemed too short, a document summarizing key knowledge points will be sent to guide the user in developing a structured thought process; simultaneously, the module will be integrated with the overall learning score. Subjective professional competence score Optimization solution: When Time-based push notifications of basic knowledge point analysis and introductory questions, when... The system pushes out frequently missed knowledge point examples and advanced questions in shifts, when... Time-based delivery of research-level academic materials. Time-sharing delivery of literary appreciation method manuals and writing templates Time-based push of cross-work comparative analysis materials comprehensively adapts to users' learning status and needs.

[0107] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0108] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A self-directed learning support system based on Chinese language professional education, characterized in that, include: The multi-source data acquisition module is used to collect user learning behavior data, user knowledge acquisition data, and standardized Chinese learning resource data in the context of self-directed learning in Chinese majors. The Chinese knowledge graph construction module is used to integrate Chinese professional knowledge points in the fields of Chinese characters, grammar, literature, and writing to build a knowledge network that includes "knowledge point nodes, relationships, and hierarchical structure"; The personalized learning push module is used to calculate the user's mastery of knowledge points and learning style preferences through a quantitative model based on the user data collected by the multi-source data acquisition module and the knowledge network generated by the Chinese knowledge graph construction module, and to generate learning content and customized learning paths that are adapted to the user's weaknesses and preferences. The learning effectiveness evaluation and optimization module is used to evaluate users' learning effectiveness from two dimensions: objective knowledge mastery ability and subjective professional competence. Based on the evaluation results, the learning plan is dynamically adjusted to form a self-learning closed loop of "data collection, knowledge modeling, content push, and effect optimization".

2. The self-learning assistance system based on Chinese major education according to claim 1, characterized in that, The user learning behavior data collected by the multi-source data acquisition module specifically includes: learning format selection records, single browsing time and total time of Chinese knowledge points, answering time and submission interval of practice questions, Chinese professional keywords manually marked by users in notes, and Chinese professional questions actively submitted by users.

3. The self-learning assistance system based on Chinese major education according to claim 1, characterized in that, The user knowledge acquisition data collected by the multi-source data acquisition module specifically includes: answer records of Chinese character recognition and phonetic transcription, answer records of ancient Chinese function words and modern Chinese syntax application, literary works appreciation texts of no less than 300 words, and Chinese major writing assignments, with the writing assignments accompanied by user modification traces and teacher annotation records.

4. The self-learning assistance system based on Chinese major education according to claim 1, characterized in that, The construction process of the Chinese knowledge graph construction module includes: S11. Knowledge point extraction: Rule extraction method is used for structured Chinese learning resources; BERT entity recognition model is used for unstructured Chinese learning resources. S12. Relationship Construction: Define semantic relationship, logical relationship, and cultural relationship. Quantify the relationship strength through annotation by Chinese professional teachers and text co-occurrence analysis. The value range is 0 to 1. A relationship strength greater than or equal to 0.8 is considered a strong relationship. S13. Hierarchical division: The basic level includes Chinese character recognition, basic grammar, and literary common sense; the advanced level includes advanced grammar and literary techniques; the research level includes academic viewpoints and cross-work comparisons.

5. The self-learning assistance system based on Chinese major education according to claim 4, characterized in that, The formula for the knowledge point mastery quantification model in the personalized learning push module is: In the formula, For users to learn Chinese professional knowledge points Mastery level (values ​​range from 0 to 100). For knowledge points Total number of practice sessions For knowledge points , For users to understand knowledge points The depth of interaction is scored as follows: detailed notes = 10 points, browsing only = 3 points, no interaction = 0 points. The maximum score for interaction depth is 10 points by default. The personalized learning recommendation module is based on Push notification content: When At that time, it pushes basic knowledge point analysis documents and introductory-level practice questions; when At that time, push advanced case study videos and reinforcement exercises for knowledge points; when At that time, push knowledge points from the knowledge graph Materials on strongly related extended knowledge points.

6. The self-learning assistance system based on Chinese major education according to claim 5, characterized in that, The personalized learning recommendation module also includes learning format preference calculation and content matching degree calculation, specifically: S21. Calculation of learning format preference weights: using the following formula: in, For users' learning formats The preference weights, wherein the weights range from 0 to 100. Corresponding text format Corresponding video format, Corresponding audio format, The cumulative time a user spends using the corresponding learning format; S22. Content matching degree calculation: The formula is as follows: in, For Chinese learning content The matching degree with the user, wherein the matching degree ranges from 0 to 100. For content Preference weights for different learning formats; The personalized learning push module prioritizes push notifications. The learning content, when multiple contents When scores are the same, the score is determined by... Push notifications in descending order.

7. A self-learning support system based on Chinese major education according to claim 6, characterized in that, The learning path planning steps of the personalized learning push module include: S31. Determine the starting point for learning: using the formula: Calculation results These knowledge points serve as the user's initial learning starting point; S32. Plan the learning order: Arrange the learning nodes according to the logical order in step S12 where the association strength is greater than or equal to 0.

8. The user must complete the learning of the current node, and the node must be learned through the formula: Post-learning mastery of calculation Only after completing the above steps can you proceed to the next learning stage; S33. Adapting to Learning Format: Based on the learning format preference weight formula: ， Calculated user preference weights ,choose The highest-ranking learning format matches the learning content format of the corresponding node; if the preference weight for video format is... If so, then a knowledge point analysis video will be matched for that node; S34. Dynamically Adjust Path: If a user's learning node is accessed via the formula: Post-learning mastery of calculation If so, return to the basic content of that node and relearn; If so, the learning access to advanced knowledge points in the Chinese knowledge graph that are strongly associated with that node will be unlocked in advance.

8. The self-learning assistance system based on Chinese major education according to claim 1, characterized in that, The comprehensive score evaluation formula for the learning effectiveness evaluation and optimization module is as follows: In the formula, The user's overall learning score ranges from 0 to 100. This represents the total number of knowledge points covered in this learning cycle. For the knowledge points after learning The degree of mastery; Pre-learning knowledge points The degree of mastery, The subjective professional competence score ranges from 0 to 10. The subjective professional competence score is derived by analyzing the key point coverage of the user's literary appreciation text and the logical completeness of the writing assignment using NLP technology, combined with the annotation results of Chinese major teachers.

9. A self-learning support system based on Chinese major education according to claim 8, characterized in that, The learning effectiveness evaluation and optimization module also includes abnormal behavior detection and scheme adjustment. Specifically, it uses an abnormal threshold formula to dynamically adjust the criteria for judging answering behavior. The formula is: in, for An abnormal answer threshold is defined, with the value ranging from 0 to 60 seconds. for Threshold for abnormal answering at any time; for Time spent answering questions by users; This represents the average time spent answering questions on this knowledge point. It is a symbolic function; when, When the answer is deemed too long, a lightweight summary text of key knowledge points will be sent; when... If the answer is deemed too short, a document summarizing the key knowledge points will be sent to the user. Simultaneously based on and Optimization solution: When At that time, it pushes basic knowledge point analysis and introductory questions; when At the same time, push out examples of commonly mistaken knowledge points and advanced questions; At that time, push academic materials at the research level; when At that time, a manual on literary appreciation methods and writing templates will be pushed out; when At the same time, it pushes cross-work comparative analysis data.

Images