CN116860978B - Primary school Chinese personalized learning system based on knowledge graph and large model - Google Patents
Primary school Chinese personalized learning system based on knowledge graph and large model Download PDFInfo
- Publication number
- CN116860978B CN116860978B CN202311109073.XA CN202311109073A CN116860978B CN 116860978 B CN116860978 B CN 116860978B CN 202311109073 A CN202311109073 A CN 202311109073A CN 116860978 B CN116860978 B CN 116860978B
- Authority
- CN
- China
- Prior art keywords
- knowledge
- learning
- language
- students
- graph
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000005065 mining Methods 0.000 claims abstract description 27
- 238000010276 construction Methods 0.000 claims abstract description 25
- 238000000034 method Methods 0.000 claims description 54
- 238000004458 analytical method Methods 0.000 claims description 25
- 230000004927 fusion Effects 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 21
- 239000000463 material Substances 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 20
- 239000013598 vector Substances 0.000 claims description 17
- 230000006399 behavior Effects 0.000 claims description 16
- 238000005516 engineering process Methods 0.000 claims description 14
- 238000011160 research Methods 0.000 claims description 14
- 230000006698 induction Effects 0.000 claims description 13
- 238000007418 data mining Methods 0.000 claims description 10
- 238000007619 statistical method Methods 0.000 claims description 10
- 230000005540 biological transmission Effects 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 9
- 230000002776 aggregation Effects 0.000 claims description 8
- 238000004220 aggregation Methods 0.000 claims description 8
- 238000003058 natural language processing Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000003066 decision tree Methods 0.000 claims description 5
- 230000001939 inductive effect Effects 0.000 claims description 5
- 230000009193 crawling Effects 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 description 12
- 230000008569 process Effects 0.000 description 10
- 239000010410 layer Substances 0.000 description 7
- 230000008901 benefit Effects 0.000 description 6
- 238000001914 filtration Methods 0.000 description 6
- 238000007405 data analysis Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 3
- 230000008520 organization Effects 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000007499 fusion processing Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000007786 learning performance Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000009897 systematic effect Effects 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Animal Behavior & Ethology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a primary school Chinese personalized learning system based on a knowledge graph and a large model, which comprises a Chinese knowledge graph construction unit, a learning scheme making unit and a Chinese knowledge system construction unit; the Chinese knowledge graph construction unit is used for forming a knowledge system by analyzing and researching Chinese literature data, identifying and summarizing knowledge points and displaying the knowledge system in a graph form; the learning scheme making unit is used for determining learning requirements through mining and analyzing the historical data of students, and matching and recommending knowledge points according to the learning requirements; the language knowledge system construction unit is used for extracting language knowledge through unsupervised learning on a large-scale corpus and fusing the language knowledge to a language knowledge map to establish a language knowledge system. The invention supports the construction of personalized Chinese learning schemes according to different demands and interests of students.
Description
Technical Field
The invention relates to the technical field of personalized learning, in particular to a primary school Chinese personalized learning system based on a knowledge graph and a large model.
Background
The primary school Chinese education is always a hot topic in the education field, and how to effectively improve the Chinese learning effect and interest of students is a problem which is always required to be perfected. Under traditional education mode, primary school's chinese education mostly adopts the mode of one to many, can't fully satisfy the demand of different students ' personalized study. Meanwhile, teachers often cannot find learning difficulties and problems of students in time, so that teaching effects are difficult to improve. Therefore, how to improve the individuation and the intelligent degree of Chinese teaching becomes an important research direction.
A knowledge graph is a semantic-based knowledge representation method, which is a directed graph composed of a plurality of entities and relationships, wherein the entities represent things in the real world, and the relationships represent links between the entities. Knowledge maps can describe and express rich semantic information of knowledge fields, such as attributes, categories, levels and the like of entities. Compared with the traditional relational database, the knowledge graph can better process semantic information and complex relations, and has stronger expression and reasoning capacity.
The large model is a neural network model formed by a large number of parameters, and has strong language understanding and generating capacity by performing unsupervised learning on a large-scale prediction library. These models have achieved very good results in natural language processing, language learning, and other language-dependent tasks.
In the aspect of primary school Chinese learning, the existing online learning platform and teaching software generally only provide static learning content and exercise, and cannot conduct personalized recommendation and coaching according to learning conditions and feedback information of students, so that personalized learning in a real sense is difficult to achieve. In addition, most of the existing language learning systems only consider grammar and vocabulary knowledge of a language surface layer, and cannot analyze and understand semantics and emotion of the language in depth, so that learning effect is limited.
For the problems in the related art, no effective solution has been proposed at present.
Disclosure of Invention
Aiming at the problems in the related art, the invention provides a primary school Chinese personalized learning system based on a knowledge graph and a large model, so as to overcome the technical problems in the prior art.
For this purpose, the invention adopts the following specific technical scheme:
the primary school language personalized learning system based on the knowledge graph and the large model comprises a language knowledge graph construction unit, a learning scheme making unit and a language knowledge system construction unit;
the Chinese knowledge graph construction unit is used for forming a knowledge system by analyzing and researching Chinese literature data, identifying and summarizing knowledge points and displaying the knowledge system in a graph form;
The learning scheme making unit is used for determining learning requirements through mining and analyzing the historical data of students, and matching and recommending knowledge points according to the learning requirements;
the language knowledge system construction unit is used for extracting language knowledge through unsupervised learning on a large-scale corpus and fusing the language knowledge to a language knowledge map to establish a language knowledge system.
Further, the Chinese knowledge graph construction unit comprises a research acquisition module, a generalization classification module and a verification display module;
the research acquisition module is used for carrying out deep analysis and research on Chinese teaching materials and identifying core knowledge points in the teaching materials;
the induction classification module is used for formulating an induction classification principle by utilizing linguistics and ontology, and inducing and organizing knowledge points according to the induction classification principle to form a clear knowledge system;
the verification display module is used for verifying and correcting the knowledge system in a mode of expert discussion, literature research and teacher feedback, and displaying the knowledge system in a map mode.
Further, the method for formulating the generalization classification principle by using linguistics and ontologies and generalizing and organizing the knowledge points according to the generalization classification principle, so as to form a clear knowledge system includes:
The concept system and the classification relation of knowledge points are clarified through ontology theory;
analyzing the language structure and the language rule in the knowledge points by applying the theory and the method of linguistics, and clearing the grammar characteristics and the semantic relation among the knowledge points;
according to the attribute and the characteristic of the knowledge points, a proper induction classification rule is formulated;
and classifying and organizing the knowledge points according to the generalized classification rules to form a clear knowledge system.
Further, the concept system and the classification relation of the knowledge points through ontology theory comprise:
determining the basis and concept required for constructing a knowledge system according to the characteristics of Chinese subject and related literature data;
analyzing the hierarchical structure and the subordinate relation of the Chinese knowledge, and establishing the hierarchical structure of the concepts by utilizing the inheritance relation and the classification relation among the concepts;
defining attributes and relationships for concepts, describing features and interrelationships between concepts;
based on the attribute and relation defined in the ontology, utilizing logic and an inference algorithm to infer the linguistic knowledge, and defining an inference classification rule based on an inference result;
and comparing the concept system with the Chinese teaching materials and subject standards, determining whether the concept system and the classification relationship in the ontology are consistent with the actual Chinese relationship, and updating the ontology regularly according to the actual Chinese knowledge.
Further, the learning scheme making unit comprises a data mining module, an analysis determining module and a matching module;
the data mining module is used for mining learning history data, learning behavior data and interest preference of students;
the analysis and determination module is used for analyzing the mined data to obtain the learning requirement analysis of the students and determining the personalized characteristics of the students;
the matching module is used for selecting and matching knowledge points suitable for students from the established Chinese knowledge graph based on the learning requirements and the personalized characteristics of the students.
Further, the selecting and matching knowledge points suitable for the students from the established Chinese knowledge graph based on the learning requirements and the personalized characteristics of the students comprises:
based on the learning requirements and individuation characteristics of students, recommending knowledge points according to the learning history and behaviors of the students through a depth model;
according to the learning requirements and individuation characteristics of students, the matched learning resources are screened and recommended from the knowledge graph, and the learning characteristics and the resource content of the students are recommended through a recommendation algorithm and a natural language processing technology.
Further, the Chinese knowledge system construction unit comprises a learning module, a representation fusion module and a recommendation module;
The learning module is used for performing unsupervised learning on a large amount of data through the GLM model;
the expression fusion module is used for extracting the language model, representing the knowledge graph and fusing the expression of the language model and the expression of the Chinese knowledge graph by using a fusion technology;
the recommendation module is used for mining and analyzing the learning data of the students and carrying out matching recommendation on knowledge points and learning resources in the knowledge graph according to learning requirements and interests of the students.
Further, the unsupervised learning of the large amount of data by the GLM model includes:
the method comprises the steps of obtaining rich language data as a training data set by crawling language data, a corpus, teaching material texts and other various sources on the Internet;
preprocessing the collected training data by using a text processing technology;
and inputting the processed training data set into the GLM model for basic training, and extracting the characteristic representation of the language knowledge.
Further, the extracting the language model and expressing the knowledge graph, and using the fusion technology to fuse the expression of the language model and the expression of the Chinese knowledge graph includes:
Using a trained GLM model to represent text data in language data by using hidden layer output, word vectors, sentence vectors and other different layers of the model;
converting a graph structure in the Chinese knowledge graph into a low-vector representation by adopting a graph embedding method, and capturing semantic relations of nodes in the graph;
taking the nodes of the knowledge graph as nodes of the graph and the characteristic representation of the large model as characteristic vectors of the nodes to combine and construct a graph structure;
iterative propagation and aggregation are carried out on the graph structure through GCN, and the characteristics of the nodes and the characteristics of the adjacent nodes are fused and updated;
fusing node features in the knowledge graph with features in the large model through the GCN during information transmission;
through multiple rounds of information propagation and feature fusion, the GCN aggregates the features of the current node and the features of the neighbor nodes, updates the features of the current node, and gradually and iteratively updates the feature representation of the node;
and inputting the fused features into a learning scheme making unit, matching and recommending knowledge points and learning resources by utilizing the fused features, and generating a personalized learning scheme according to the matching and recommending results.
Further, the mining and analyzing the learning data of the students, and performing matching recommendation according to learning requirements and interests of the students and knowledge points and learning resources in the knowledge graph comprises:
Cleaning and converting the collected student learning data, removing abnormal values and missing values, converting the data into a format which can be used for statistical analysis, and carrying out descriptive analysis among learning features on the data by using a statistical method;
selecting the characteristic with distinction degree for classifying the student learning types from the statistical analysis learning characteristics through the coefficient of the foundation, and constructing a classification model of the student learning types by adopting a decision tree algorithm based on the selected characteristic;
converting learning data of the students into transaction type data sets required by association rule mining, wherein the transaction type data sets represent learning records of the students, and the learning records represent a set of learning resources;
excavating frequent item sets among learning resources through an Apriori algorithm, and determining the frequent item sets through a minimum support threshold;
generating an association rule based on the frequent item set, calculating the confidence coefficient and the support degree of the association rule, and sequencing the association rule obtained by mining according to the confidence coefficient and the support degree;
based on the sorting result, learning resources corresponding to the association rules with high confidence are recommended to students.
The beneficial effects of the invention are as follows:
the invention has higher application value and technical advantage in the field of Chinese learning, better Chinese learning experience and effect can be provided for students by combining a GLM model with a knowledge map and combining a personalized learning scheme and a recommendation system method, a Chinese knowledge system is constructed by using the knowledge map, chinese knowledge can be organically organized, knowledge points inside and outside the class can be clearly connected in series, and particularly, the abstract and induction are carried out through the related knowledge points of the Chinese, so that a comprehensive and systematic Chinese knowledge map is established, and the students are supported to carry out targeted knowledge learning.
According to the invention, a personalized Chinese learning scheme is constructed according to different demands and interests of students, specifically, the learning demands and interests of the students can be analyzed and mined through an intelligent recommendation algorithm in the system, then related knowledge points and learning resources are matched and recommended, and meanwhile, the system can be adjusted and optimized in real time according to learning performance and historical books of the students, so that the learning effect is improved.
The invention carries out unsupervised learning on a large-scale corpus, thereby having strong language understanding and generating capability, abstracts and generalizes language knowledge by utilizing a large model, builds a finer and high-level language knowledge system, and extracts the essential characteristics and rules of the language knowledge by training and learning a large amount of language data, thereby realizing the automatic expression and reasoning of the language knowledge.
According to the invention, through fusing the GLM model and the knowledge graph and combining with a customized learning scheme, a more comprehensive and fine language knowledge system is constructed by fusing the advantages of the language model and the knowledge graph, the language model can learn the essential characteristics and rules of language from a large amount of language data, the language model has strong language modeling and expression capability, and the knowledge graph provides structured language knowledge expression and relationship, so that the organization and recommendation of learning resources are more accurate and effective.
According to the invention, the data analysis and the mining are carried out on the learning history, the behaviors and the interests of the students, so that a customized learning scheme is realized, the students can acquire knowledge points and learning resources matched with the students according to own learning requirements and interests, and the personalized learning scheme can better meet the learning requirements of the students and improve the learning effect and interests.
According to the invention, in the process of constructing a Chinese knowledge system, an unsupervised learning method is utilized, a large amount of language data is trained, the unsupervised learning can automatically find the essential characteristics and rules of language knowledge without manually marked labels, the information of the large-scale language data can be fully utilized, the expression capacity and reasoning effect of the Chinese knowledge are improved, and the learning requirement and interest of students are matched and recommended with knowledge points and learning resources in a Chinese knowledge graph by utilizing a recommendation system method.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic block diagram of a primary school Chinese personalized learning system based on knowledge graph and large model in accordance with an embodiment of the invention.
In the figure:
1. a Chinese knowledge graph construction unit; 2. a learning scheme making unit; 3. and a Chinese knowledge system construction unit.
Description of the embodiments
For the purpose of further illustrating the various embodiments, the present invention provides the accompanying drawings, which are a part of the disclosure of the present invention, and which are mainly used for illustrating the embodiments and for explaining the principles of the operation of the embodiments in conjunction with the description thereof, and with reference to these matters, it will be apparent to those skilled in the art to which the present invention pertains that other possible embodiments and advantages of the present invention may be practiced.
According to the embodiment of the invention, a primary school Chinese personalized learning system based on a knowledge graph and a large model is provided.
The invention is further described with reference to the drawings and the specific embodiments, as shown in fig. 1, the primary school language personalized learning system based on a knowledge graph and a large model according to the embodiment of the invention comprises a language knowledge graph construction unit 1, a learning scheme formulation unit 2 and a language knowledge system construction unit 3;
the Chinese knowledge graph construction unit 1 is used for forming a knowledge system by analyzing and researching Chinese literature data, identifying and summarizing knowledge points and displaying the knowledge system in a graph form;
the establishment of the Chinese knowledge graph comprises the steps of abstracting and summarizing the Chinese knowledge points to establish a comprehensive and systematic Chinese knowledge graph, and specifically, the step can be used for analyzing and summarizing literature data such as Chinese teaching materials, chinese subject standards, teaching outlines and the like, so that a complete Chinese knowledge system is established. Then, by summarizing and classifying the knowledge points, a Chinese knowledge framework based on the knowledge graph is established.
In one embodiment, the Chinese knowledge graph construction unit 1 includes a research acquisition module, a generalization classification module and a verification display module;
the research acquisition module is used for carrying out deep analysis and research on Chinese teaching materials and identifying core knowledge points in the teaching materials.
Specifically, first, literature data such as Chinese teaching materials, chinese subject standards, teaching outlines and the like are subjected to deep analysis and research to obtain core knowledge points and teaching points of the Chinese subject.
Based on these literature, different levels and components of the chinese discipline, such as words, sentences, chapters, etc., as well as associated grammar rules, tutorial techniques, composition skills, etc., are identified.
The generalization and classification module is used for formulating a generalization and classification principle by utilizing linguistics and ontology, and generalizing and organizing knowledge points according to the generalization and classification principle to form a clear knowledge system.
The method for creating the summary classification principle by using linguistics and ontologies, and summarizing and organizing knowledge points according to the summary classification principle, so as to form a clear knowledge system comprises the following steps:
the concept system and the classification relation of knowledge points are clarified through ontology theory;
analyzing the language structure and the language rule in the knowledge points by applying the theory and the method of linguistics, and clearing the grammar characteristics and the semantic relation among the knowledge points;
according to the attribute and the characteristic of the knowledge points, a proper induction classification rule is formulated;
and classifying and organizing the knowledge points according to the generalized classification rules to form a clear knowledge system.
In building a knowledge hierarchy, we will generalize and categorize each knowledge point so that the system can better understand and apply the knowledge.
Inductive classification can be based on different principles such as grammatical features, language functions, chapter structures, etc. For example, grammar rules may be categorized by tense, language, sentence pattern, etc.; the techniques of the repair are classified according to metaphors, exaggerations, comparisons, and the like.
In the process of summarizing classification, the related theory and method of linguistics and education can be used as reference. For example, category theory may be used to analyze the categorization of grammar rules and chapter analysis theory may be used to analyze the categorization of chapter structures.
Classification of grammar rules: category theory is a theoretical framework for grammar analysis, which classifies grammar rules into different categories or categories, category theory is based on category concepts, wherein categories represent basic semantic concepts in grammar, grammar rules can be classified and generalized by mapping different grammar rules into proper categories, and category theory can help understand the structure and composition of language, so that a knowledge system of grammar rules is established.
The classifying method of the chapter structure comprises the following steps: the chapter analysis theory provides a method for analyzing and understanding the chapter structure, the chapters can be classified according to the organization structure, information flow and semantic relation, the common chapter structure classification method comprises a linear structure, a hierarchical structure, a causal relation structure, a problem solving structure and the like, and a knowledge system of the chapter structure can be established and related with other Chinese knowledge points through analyzing and classifying the chapters.
Inductive classification of language styles and repair techniques: the language style and the method of the congratulation are the common expression mode and the congratulation mode in the Chinese, and the language style can be classified according to the characteristics of expression, the style of words, the structure of sentences and the like, such as image, abstract, elegant, naive and the like. The method can be classified according to the purpose and effect of expression, such as metaphor, anthropomorphic, exaggeration, reverse question, etc., and can help students understand and apply different expression modes by generalizing and classifying language style and the method of the repair.
Classification of topics and literaries of literary works: literary works can be classified according to the themes and the literaries. The subject classification refers to the subject matter and content related to the work, such as love, nature, society, etc. The genre classification refers to the expression form and structure of the work, such as prose, poems, novels, etc. By summarizing and classifying the topics and the literaries of the literary works, a knowledge system of the literary works can be established, and students can be helped to understand and appreciate different types of literary works.
In the process of establishing the Chinese knowledge graph and the knowledge point induction classification, related theories and methods such as knowledge graph, ontology, linguistic, machine learning, natural language processing technology and the like can be applied.
The knowledge graph, as a graph-like structure, can be used to represent and organize knowledge points and their relationships. The expression capability of the knowledge graph can be utilized to construct a Chinese knowledge system with a hierarchical structure and an association relationship.
Ontology is a method for researching concepts and relationships between concepts, and can help us to understand the concept system and classification relationships of Chinese knowledge. From the analysis of ontologies we can determine the appropriate concept classification and categorization principles.
The following steps may be implemented using ontologies to implement conceptual systems and classification relationships of the theory of Chinese knowledge:
specifically, the concept system and the classification relation of the knowledge points through ontology theory comprise:
determining the basic and concepts required for constructing a knowledge system according to the characteristics of Chinese subject and related literature data, wherein the basic and concepts can comprise concepts such as language, literature, chapters, congratulations and the like;
analyzing the hierarchical structure and the subordinate relation of the Chinese knowledge, and establishing the hierarchical structure of the concepts by utilizing the inheritance relation and the classification relation among the concepts; for example, the language knowledge may be divided into two major classes, namely language knowledge and literature knowledge, and then further subdivided into subclasses such as grammar, thesaurus, literature, and the like.
For concepts to define attributes and relationships, features and interrelationships between concepts are described, attributes may be features or descriptions of concepts, such as grammatical concepts, attributes may include parts of speech, syntactic functions, etc., and relationships may be associations or relations between concepts, such as relationships between literature topics and literature works.
Based on the attribute and relation defined in the ontology, the language knowledge is inferred by utilizing logic and an inference algorithm, and an inference classification rule is defined based on an inference result, so that the language knowledge is inferred and inferred, and students are helped to understand and apply the language knowledge.
And comparing the concept system with the Chinese teaching materials and subject standards, determining whether the concept system and the classification relationship in the ontology are consistent with the actual Chinese relationship, and updating the ontology regularly according to the actual Chinese knowledge.
In the process of establishing the ontology, verification and updating of knowledge are required. Through comparison and verification with Chinese teaching materials, subject standards and the like, the concept system and the classification relationship in the ontology are ensured to be consistent with actual Chinese knowledge. Meanwhile, with the continuous development and change of the Chinese field, the ontology needs to be updated regularly, so that the consistency of the ontology and the latest knowledge is maintained.
Linguistics is the discipline of studying linguistic structure and linguistic rules, which provides a theoretical basis for analyzing and classifying linguistic knowledge. For example, using knowledge of sentence laws, we can analyze and classify sentence structures; using semantic concepts, we can study semantic relationships and semantic features between terms.
In the knowledge point induction classification process, we can also use machine learning and natural language processing techniques to assist. For example, language features and rules may be extracted from a large corpus using text mining and automatic labeling techniques to support generalization and classification of knowledge points.
In general, a complete Chinese knowledge system is established by comprehensively considering a plurality of information sources, and through verification and correction processes, the Chinese knowledge materials can be deeply analyzed to identify core knowledge points and important concepts in the teaching materials, and simultaneously, the subject standard and the teaching outline are referred to obtain the Chinese knowledge points and skills which the students of corresponding grades should master.
In the process of analysis and summarization, related theories and methods such as concept analysis, ontology, linguistics and the like can be applied to assist in managing the concept system and classification relation of Chinese knowledge.
The verification display module is used for verifying and correcting the knowledge system in a mode of expert discussion, literature research and teacher feedback, and displaying the knowledge system in a map mode.
The establishment of the knowledge system needs to be verified and corrected in the modes of expert discussion, literature study, teacher feedback and the like so as to ensure the accuracy and the integrity of the knowledge system.
On this basis, by induction and summarization, the knowledge points are organized into an organic knowledge system, and the knowledge system is represented and displayed in the form of a conceptual diagram, a tree structure or a map.
The learning scheme making unit 2 is used for determining learning requirements by mining and analyzing the history data of students, and matching and recommending knowledge points according to the learning requirements;
the unit comprises the steps of analyzing and mining the learning requirement and interest of the student, and matching and recommending relevant knowledge points and learning resources according to the personalized requirement of the student, wherein the learning characteristics and requirements of the student can be determined by analyzing and mining the learning history, learning behaviors, interest preference and other data of the student. The system then matches and recommends relevant knowledge points and learning resources based on the student's learning needs and interests. In the step of supporting the customized learning scheme, the learning requirements and interests of the students are analyzed and mined, and relevant knowledge points and learning resources are matched and recommended according to the personalized requirements of the students.
In one embodiment, the learning scheme making unit 2 includes a data mining module, an analysis determining module, and a matching module;
The data mining module is used for mining learning history data, learning behavior data and interest preference of students;
the analysis and determination module is used for analyzing the mined data to obtain the learning requirement analysis of the students and determining the personalized characteristics of the students;
specifically, learning requirement analysis: and analyzing the superiority and inferiority of the students on different knowledge points by integrating the learning history data and the learning behavior data of the students, and identifying the learning requirement and improvement points of the students.
And (3) personalized characteristics determination: and determining personalized learning characteristics of the students, such as learning styles, interest preferences, knowledge point mastering conditions and the like, according to learning historical data, learning behavior data and interest preference analysis results.
The matching module is used for selecting and matching knowledge points suitable for students from the established Chinese knowledge graph based on the learning requirements and the personalized characteristics of the students.
Specifically, based on the learning requirement and individuation characteristics of the students, selecting and matching knowledge points suitable for the students from the established Chinese knowledge graph comprises:
based on the learning requirements and individuation characteristics of students, recommending knowledge points according to the learning history and behaviors of the students through a depth model;
According to the learning requirements and individuation characteristics of students, the matched learning resources are screened and recommended from the knowledge graph, and the learning characteristics and the resource content of the students are recommended through a recommendation algorithm and a natural language processing technology.
Knowledge point matching recommendation: based on the learning requirement and individuation characteristics of the students, knowledge points suitable for the students are selected and matched from the established Chinese knowledge graph. Recommendation algorithms (e.g., collaborative filtering, content filtering, deep learning models, etc.) may be used to recommend knowledge points based on the learning history and behavior of the students.
The foregoing is merely illustrative of some of the common theories and algorithms, and the actual implementation may be adapted according to specific needs and technical choices.
Learning resource recommendation: according to the learning requirements and individuation characteristics of students, learning resources suitable for the students, such as textbook chapters, exercises, reading materials, learning videos and the like, are screened and recommended from a learning resource library. Recommendation algorithms and natural language processing techniques can be used to make recommendations based on the learning characteristics of the student and the content of the resource.
The language knowledge system construction unit 3 is configured to extract language knowledge by performing unsupervised learning on the large-scale corpus, and fuse the language knowledge to the language knowledge map to construct a language knowledge system.
The GLM (Generative Language Model) model language knowledge system of the Qinghai open source is utilized: the method comprises the steps of training and learning a large amount of language data, extracting the essential characteristics and rules of language knowledge, thereby constructing a finer and higher-level language knowledge system, and particularly, the method can realize automatic representation and reasoning of the language knowledge by performing unsupervised learning on a large amount of text data. Then, by fusing the language knowledge of the large model and the language knowledge graph, a more comprehensive and fine language knowledge system is established.
In order to construct a finer and higher-level Chinese knowledge system, the step utilizes an open-source GLM (Generative Language Model) model of the Qinghai university to perform unsupervised learning, and the language knowledge of the large model is fused with a Chinese knowledge graph.
In one embodiment, the language knowledge system construction unit 3 includes a learning module, a representation fusion module and a recommendation module;
the learning module is used for performing unsupervised learning on a large amount of data through the GLM model;
specifically, the unsupervised learning of a large amount of data by the GLM model includes:
The method comprises the steps of obtaining rich language data as a training data set by crawling language data, a corpus, teaching material texts and other various sources on the Internet;
preprocessing the collected training data by using a text processing technology;
and inputting the processed training data set into the GLM model for basic training, and extracting the characteristic representation of the language knowledge.
Specifically, data collection: a large amount of language data is obtained as a training data set by crawling various sources such as language data, a corpus, text of teaching materials and the like on the Internet.
Data preprocessing: preprocessing the collected data, including word segmentation, part-of-speech tagging, stop word removal, unknown word processing and the like. These preprocessing steps may utilize conventional text processing techniques such as word segmentation tools, part-of-speech markers, etc.
GLM training: the GLM model of the open source of the Qinghua university is selected as a basic model for training, the GLM model is generally based on a transducer architecture, has strong language modeling capability, and can learn the probability distribution of the language and the characteristic representation of the language through iterative training.
The expression fusion module is used for extracting the language model, representing the knowledge graph and fusing the expression of the language model and the expression of the Chinese knowledge graph by using a fusion technology;
Specifically, the extracting the language model and representing the knowledge graph, and using the fusion technology to fuse the representation of the language model and the representation of the Chinese knowledge graph includes:
using a trained GLM model to represent text data in language data by using hidden layer output, word vectors, sentence vectors and other different layers of the model;
converting a graph structure in the Chinese knowledge graph into a low-vector representation by adopting a graph embedding method, and capturing semantic relations of nodes in the graph;
taking the nodes of the knowledge graph as nodes of the graph and the characteristic representation of the large model as characteristic vectors of the nodes to combine and construct a graph structure;
iterative propagation and aggregation are carried out on the graph structure through GCN, and the characteristics of the nodes and the characteristics of the adjacent nodes are fused and updated;
fusing node features in the knowledge graph with features in the large model through the GCN during information transmission;
through multiple rounds of information propagation and feature fusion, the GCN aggregates the features of the current node and the features of the neighbor nodes, updates the features of the current node, and gradually and iteratively updates the feature representation of the node;
and inputting the fused features into a learning scheme making unit, matching and recommending knowledge points and learning resources by utilizing the fused features, and generating a personalized learning scheme according to the matching and recommending results.
Graph Convolutional Networks (graph roll-up network, GCN) is a deep learning model for processing graph data. GCN is a variant based on a graph neural network (Graph Neural Network) aimed at performing tasks such as node classification, graph classification, and graph generation using relationships between nodes, in which the representation of each node is obtained by aggregating its own features with the features of neighboring nodes. This aggregation operation is similar to the convolution operation in a conventional convolutional neural network, but is performed in a graph structure. By performing an aggregation operation on each node, the GCN can capture global and local graph structure information and use it for node representation learning and task solution.
The core operation of the GCN is a graph roll stacking, which updates the node representation based on the node's neighbor relationships and characteristics. By stacking multiple graph convolution layers, the GCN is able to aggregate graph structure information layer by layer and produce progressively richer node representations. These node representations may be used for various graph analysis tasks such as node classification, link prediction, graph generation, and the like.
The following is a specific example of a fusion process using GCN to fuse knowledge maps with large models:
Chinese knowledge graph representation: and converting the nodes and edges in the Chinese knowledge graph into vector representations. Graph embedding methods such as GraphSAGE, node2Vec and the like can be adopted to convert the graph structure into low-dimensional vector representation and capture the semantic relationship among nodes;
the language model representation and the language knowledge graph representation are fused by using a fusion technology, and the graph neural network (Graph Neural Networks) model, such as Graph Convolutional Networks (GCN), graph Attention Networks (GAT) and the like, can be used for carrying out joint training on the language model representation and the knowledge graph representation, so that the characteristics of the language model representation and the knowledge graph representation are fused, the prior knowledge of the language knowledge can be added on the basis of the language model, and the understanding and the representing capability of the language knowledge can be improved.
Constructing a knowledge graph: firstly, constructing a knowledge graph according to a concept system and a classification relation of Chinese knowledge. The knowledge graph may be represented as a directed graph, in which nodes represent concepts of Chinese knowledge and edges represent relationships between the concepts, and each node may contain attribute information of the concepts, such as names, definitions, examples, and the like.
Constructing a large model: the language data is trained and learned by using a Qinghua open-source GLM model or other suitable large model, the characteristic representation of language knowledge is extracted, the large model can convert the language data into a low-dimensional vector representation, and the semantic and grammar characteristics of the language are captured.
Constructing a graph structure: taking the nodes of the knowledge graph as nodes of the graph and the characteristic representation of the large model as characteristic vectors of the nodes to combine and construct a graph structure;
information transmission is carried out: the characteristics of the nodes and the characteristics of the adjacent nodes are fused and updated through iterative propagation and aggregation of the GCN on the graph structure, and in general, the GCN is utilized for information propagation and characteristic fusion. The GCN fuses and updates the characteristics of the nodes and the characteristics of the adjacent nodes by iterative propagation and aggregation on the graph, and in each round of propagation, the GCN considers the information of the adjacent nodes of the nodes and performs information aggregation and transmission according to the graph structure;
feature fusion: and fusing the node characteristics in the knowledge graph with the characteristics in the large model through the GCN during information transmission, and fusing the node characteristics in the knowledge graph with the characteristics in the large model through the GCN in each round of information transmission, so as to gradually fuse the information of the node characteristics and the characteristics into the same characteristic space. The fused features may include more comprehensive and rich language knowledge and language features.
Iterative propagation: through multiple rounds of information propagation and feature fusion, the GCN aggregates the features of the current node and the features of the neighbor nodes, updates the features of the current node, and gradually and iteratively updates the feature representation of the node;
Feature application: and inputting the fused features into a learning scheme making unit, matching and recommending knowledge points and learning resources by utilizing the fused features, and generating a personalized learning scheme according to the matching and recommending results. In general, the fused features can be applied to the individual learning needs and recommendations of the student. Through analyzing the learning history, behaviors and interests of the students, the individual learning requirements and the recommendation of the students can be utilized, the fused features are applied to a customized learning scheme, and specifically, the fused features can be used for matching and recommending relevant knowledge points and learning resources so as to meet the learning requirements and interests of the students;
matching and recommending: based on the learning requirement and interest of students, the knowledge points and learning resources are matched and recommended by utilizing the fused features, the similarity between the students and the knowledge points or learning resources can be calculated, the content with high matching degree with the characteristics of the students is selected for recommendation, and various measurement methods such as cosine similarity, euclidean distance and the like can be used for similarity calculation.
Personalized learning scheme: and generating a personalized learning scheme according to the matching and recommending results, wherein the personalized learning scheme comprises recommending knowledge points, learning materials and learning paths suitable for students. The personalized learning scheme can be dynamically adjusted and updated according to learning progress and feedback of students so as to provide learning experience which meets the demands of the students.
In the whole fusion process, GCN plays a key role in fusing the knowledge graph with the characteristic representation of the large model, and combining the information of the knowledge graph and the characteristic representation of the large model step by step through iterative propagation and characteristic fusion. The language knowledge structure in the knowledge graph and the language features in the large model can be fully utilized, more comprehensive and rich language knowledge representation and personalized learning recommendation are provided, and specific fusion steps comprise building graph structures, information transmission, feature fusion, iterative transmission, feature application and the like.
In the implementation process, adjustment and optimization are needed according to specific requirements and data conditions, such as selecting proper GLM model parameters, fusion algorithm structures, super parameters and the like. In addition, care is taken to reasonably sample and train the data to take full advantage of the limited computing resources and time.
The recommendation module is used for mining and analyzing the learning data of the students and carrying out matching recommendation on knowledge points and learning resources in the knowledge graph according to learning requirements and interests of the students.
The recommendation module specifically includes:
data analysis and mining: data such as learning history, learning behaviors, interest preferences and the like of students are analyzed and mined. Data mining and machine learning techniques, such as clustering, classification, association rule mining, etc., may be used from which the learning characteristics and needs of students are found.
Learning demand matching and recommendation: according to the learning requirement and interest of students, matching and recommending the learning requirement and interest with knowledge points and learning resources in the Chinese knowledge graph. The method of the recommendation system, such as content-based recommendation, collaborative filtering recommendation and the like, can be used for providing personalized learning recommendation for students by combining learning preference of the students and the structure of the Chinese knowledge graph.
Specifically, the mining and analyzing the learning data of the students, and performing matching recommendation according to learning requirements and interests of the students and knowledge points and learning resources in the knowledge graph includes:
the collected student learning data is cleaned and converted to remove abnormal values and missing values, the data is converted into a format which can be used for statistical analysis, and descriptive analysis among learning features is carried out on the data by applying a statistical method, for example, average learning time of students, average scores of mastering conditions of knowledge points, clicking times of the students on different types of learning resources and the like are calculated.
From the study characteristics of statistical analysis, selecting the characteristic with distinction degree for classifying the study types of the students through the coefficient of the foundation, such as the correlation between the study time and the student performance, constructing a classification model of the study types of the students by adopting a decision tree algorithm (such as ID3, C4.5 and the like) based on the selected characteristic, and carrying out node division on the decision tree by recursively selecting the optimal characteristic to finally form a tree structure for predicting the study types of the students;
Converting learning data of the students into transaction type data sets required by association rule mining, wherein the transaction type data sets represent learning records of the students, and the learning records represent a set of learning resources;
excavating frequent item sets among learning resources through an Apriori algorithm, and determining the frequent item sets through a minimum support threshold;
generating an association rule based on a frequent item set, calculating the confidence coefficient and the support degree of the association rule, and sorting the association rule obtained by mining according to the confidence coefficient and the support degree to set a minimum confidence coefficient threshold value and screen out a rule with certain association;
based on the sorting result, recommending learning resources corresponding to the association rule with high confidence to the students, and based on the association rule obtained by mining, recommending the learning resources related to the current learning situation of the students to the students. The association rules can be ordered according to indexes such as confidence level, support level and the like, and learning resources corresponding to the association rules with high confidence level are recommended.
For data analysis and mining, methods such as statistical analysis, machine learning, deep learning and the like, such as clustering algorithms (e.g. K-means, hierarchical clustering), classification algorithms (e.g. decision trees, support vector machines), association rule mining (e.g. Apriori algorithm) and the like, can be used to mine learning characteristics and requirements of students.
For learning demand matching and recommendation, methods of the recommendation system, such as Content-based recommendation (Content-based Recommendation), collaborative filtering recommendation (Collaborative Filtering), and the like, may be used. The content-based recommendation can be performed according to learning requirements and interests of students, the content-based recommendation can be matched with knowledge points and learning resources in a Chinese knowledge graph, and the recommendation is performed by using feature similarity or similarity represented by vectors. Collaborative filtering recommendation can discover a common mode and similar interests of learning by analyzing behavior data of students and behavior data of other students, and recommend learning resources of similar students for the students.
The implementation of these steps can be adjusted and optimized according to actual situations and requirements, such as selecting a proper data analysis and mining algorithm, recommending a system method, adjusting recommended strategies and weights, and the like.
In summary, by means of the technical scheme, the method has higher application value and technical advantage in the field of Chinese learning, better Chinese learning experience and effect can be provided for students by combining the GLM model with the knowledge map and combining a personalized learning scheme with a recommendation system method, a Chinese knowledge system is constructed by using the knowledge map, chinese knowledge can be organically organized, knowledge points inside and outside the class can be clearly connected in series, and particularly, the Chinese knowledge map is built comprehensively and systematically by abstracting and inducing the Chinese related knowledge points, so that the students are supported to carry out targeted knowledge learning. According to the invention, a personalized Chinese learning scheme is constructed according to different demands and interests of students, specifically, the learning demands and interests of the students can be analyzed and mined through an intelligent recommendation algorithm in the system, then related knowledge points and learning resources are matched and recommended, and meanwhile, the system can be adjusted and optimized in real time according to learning performance and historical books of the students, so that the learning effect is improved. The invention carries out unsupervised learning on a large-scale corpus, thereby having strong language understanding and generating capability, abstracts and generalizes language knowledge by utilizing a large model, builds a finer and high-level language knowledge system, and extracts the essential characteristics and rules of the language knowledge by training and learning a large amount of language data, thereby realizing the automatic expression and reasoning of the language knowledge.
According to the invention, through fusing the GLM model and the knowledge graph and combining with a customized learning scheme, a more comprehensive and fine language knowledge system is constructed by fusing the advantages of the language model and the knowledge graph, the language model can learn the essential characteristics and rules of language from a large amount of language data, the language model has strong language modeling and expression capability, and the knowledge graph provides structured language knowledge expression and relationship, so that the organization and recommendation of learning resources are more accurate and effective. According to the invention, the data analysis and the mining are carried out on the learning history, the behaviors and the interests of the students, so that a customized learning scheme is realized, the students can acquire knowledge points and learning resources matched with the students according to own learning requirements and interests, and the personalized learning scheme can better meet the learning requirements of the students and improve the learning effect and interests. According to the invention, in the process of constructing a Chinese knowledge system, an unsupervised learning method is utilized, a large amount of language data is trained, the unsupervised learning can automatically find the essential characteristics and rules of language knowledge without manually marked labels, the information of the large-scale language data can be fully utilized, the expression capacity and reasoning effect of the Chinese knowledge are improved, and the learning requirement and interest of students are matched and recommended with knowledge points and learning resources in a Chinese knowledge graph by utilizing a recommendation system method.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (6)
1. The primary school language personalized learning system based on the knowledge graph and the large model is characterized by comprising a language knowledge graph construction unit, a learning scheme making unit and a language knowledge system construction unit;
the Chinese knowledge graph construction unit is used for forming a knowledge system by analyzing and researching Chinese literature data, identifying and summarizing knowledge points and displaying the knowledge system in a graph form;
the learning scheme making unit is used for determining learning requirements through mining and analyzing the historical data of students, and matching and recommending knowledge points according to the learning requirements;
the language knowledge system construction unit is used for extracting language knowledge through unsupervised learning on a large-scale corpus and fusing the language knowledge to a language knowledge map to establish a language knowledge system;
the Chinese knowledge system construction unit comprises a learning module, a representation fusion module and a recommendation module;
The learning module is used for performing unsupervised learning on a large amount of data through the GLM model;
the expression fusion module is used for extracting the language model, representing the knowledge graph and fusing the expression of the language model and the expression of the Chinese knowledge graph by using a fusion technology;
the recommendation module is used for mining and analyzing the learning data of the students and carrying out matching recommendation on knowledge points and learning resources in the knowledge map according to the learning requirements and interests of the students;
the unsupervised learning of a large amount of data by the GLM model includes:
the method comprises the steps of obtaining rich language data as a training data set by crawling language data, a corpus and teaching material texts on the Internet;
preprocessing the collected training data by using a text processing technology;
inputting the processed training data set into a GLM model for basic training, and extracting the characteristic representation of language knowledge;
the extracting the language model, representing the knowledge graph, and fusing the representation of the language model and the representation of the Chinese knowledge graph by using a fusion technology comprises the following steps:
using a trained GLM model to represent text data in language data by using hidden layer output, word vectors and sentence vectors of the model;
Converting a graph structure in the Chinese knowledge graph into a low-vector representation by adopting a graph embedding method, and capturing semantic relations of nodes in the graph;
taking the nodes of the knowledge graph as nodes of the graph and the characteristic representation of the large model as characteristic vectors of the nodes to combine and construct a graph structure;
iterative propagation and aggregation are carried out on the graph structure through GCN, and the characteristics of the nodes and the characteristics of the adjacent nodes are fused and updated;
fusing node features in the knowledge graph with features in the large model through the GCN during information transmission;
through multiple rounds of information propagation and feature fusion, the GCN aggregates the features of the current node and the features of the neighbor nodes, updates the features of the current node, and gradually and iteratively updates the feature representation of the node;
inputting the fused features into a learning scheme making unit, matching and recommending knowledge points and learning resources by utilizing the fused features, and generating a personalized learning scheme according to the matching and recommending results;
the mining and analyzing the learning data of the students, and carrying out matching recommendation on knowledge points and learning resources in the knowledge graph according to learning requirements and interests of the students comprises the following steps:
Cleaning and converting the collected student learning data, removing abnormal values and missing values, converting the data into a format which can be used for statistical analysis, and carrying out descriptive analysis among learning features on the data by using a statistical method;
selecting the characteristic with distinction degree for classifying the student learning types from the statistical analysis learning characteristics through the coefficient of the foundation, and constructing a classification model of the student learning types by adopting a decision tree algorithm based on the selected characteristic;
converting learning data of the students into transaction type data sets required by association rule mining, wherein the transaction type data sets represent learning records of the students, and the learning records represent a set of learning resources;
excavating frequent item sets among learning resources through an Apriori algorithm, and determining the frequent item sets through a minimum support threshold;
generating an association rule based on the frequent item set, calculating the confidence coefficient and the support degree of the association rule, and sequencing the association rule obtained by mining according to the confidence coefficient and the support degree;
based on the sorting result, learning resources corresponding to the association rules with high confidence are recommended to students.
2. The primary school language personalized learning system based on the knowledge graph and the large model according to claim 1, wherein the language knowledge graph construction unit comprises a research acquisition module, a generalization classification module and a verification display module;
The research acquisition module is used for carrying out deep analysis and research on Chinese teaching materials and identifying core knowledge points in the teaching materials;
the induction classification module is used for formulating an induction classification principle by utilizing linguistics and ontology, and inducing and organizing knowledge points according to the induction classification principle to form a clear knowledge system;
the verification display module is used for verifying and correcting the knowledge system in a mode of expert discussion, literature research and teacher feedback, and displaying the knowledge system in a map mode.
3. The knowledge graph and large model based primary school language personalized learning system according to claim 2, wherein the steps for formulating a generalization classification rule using linguistics and ontologies and generalizing and organizing knowledge points according to the generalization classification rule to form a clear knowledge system comprise:
the concept system and the classification relation of knowledge points are clarified through ontology theory;
analyzing the language structure and the language rule in the knowledge points by applying the theory and the method of linguistics, and clearing the grammar characteristics and the semantic relation among the knowledge points;
according to the attribute and the characteristic of the knowledge points, a proper induction classification rule is formulated;
And classifying and organizing the knowledge points according to the generalized classification rules to form a clear knowledge system.
4. The knowledge-graph-and-large-model-based primary school language personalized learning system according to claim 3, wherein the conceptual system and classification relation for learning knowledge points through ontology theory comprises:
determining the basis and concept required for constructing a knowledge system according to the characteristics of Chinese subject and related literature data;
analyzing the hierarchical structure and the subordinate relation of the Chinese knowledge, and establishing the hierarchical structure of the concepts by utilizing the inheritance relation and the classification relation among the concepts;
defining attributes and relationships for concepts, describing features and interrelationships between concepts;
based on the attribute and relation defined in the ontology, utilizing logic and an inference algorithm to infer the linguistic knowledge, and defining an inference classification rule based on an inference result;
and comparing the concept system with the Chinese teaching materials and subject standards, determining whether the concept system and the classification relationship in the ontology are consistent with the actual Chinese relationship, and updating the ontology regularly according to the actual Chinese knowledge.
5. The knowledge graph and large model-based primary school language personalized learning system according to claim 1, wherein the learning scheme making unit comprises a data mining module, an analysis determining module and a matching module;
The data mining module is used for mining learning history data, learning behavior data and interest preference of students;
the analysis and determination module is used for analyzing the mined data to obtain the learning requirement analysis of the students and determining the personalized characteristics of the students;
the matching module is used for selecting and matching knowledge points suitable for students from the established Chinese knowledge graph based on the learning requirements and the personalized characteristics of the students.
6. The knowledge-graph and large-model-based primary school language personalized learning system according to claim 5, wherein selecting and matching knowledge points suitable for students from the established Chinese knowledge graph based on learning requirements and personalized features of the students comprises:
based on the learning requirements and individuation characteristics of students, recommending knowledge points according to the learning history and behaviors of the students through a depth model;
according to the learning requirements and individuation characteristics of students, the matched learning resources are screened and recommended from the knowledge graph, and the learning characteristics and the resource content of the students are recommended through a recommendation algorithm and a natural language processing technology.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311109073.XA CN116860978B (en) | 2023-08-31 | 2023-08-31 | Primary school Chinese personalized learning system based on knowledge graph and large model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311109073.XA CN116860978B (en) | 2023-08-31 | 2023-08-31 | Primary school Chinese personalized learning system based on knowledge graph and large model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116860978A CN116860978A (en) | 2023-10-10 |
| CN116860978B true CN116860978B (en) | 2023-11-21 |
Family
ID=88228911
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311109073.XA Active CN116860978B (en) | 2023-08-31 | 2023-08-31 | Primary school Chinese personalized learning system based on knowledge graph and large model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116860978B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117114104B (en) * | 2023-10-24 | 2024-01-30 | 徐州医科大学 | Personalized knowledge graph generation method based on knowledge feature division and multidimensional evaluation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110289101A (en) * | 2019-07-02 | 2019-09-27 | 京东方科技集团股份有限公司 | A kind of computer equipment, system and readable storage medium storing program for executing |
| CN113361285A (en) * | 2021-06-30 | 2021-09-07 | 北京百度网讯科技有限公司 | Training method of natural language processing model, natural language processing method and device |
| CN114707000A (en) * | 2021-08-09 | 2022-07-05 | 北京智源人工智能研究院 | Knowledge graph-based question-answer library generation method and device, electronic equipment and storage medium |
| CN114861069A (en) * | 2022-06-07 | 2022-08-05 | 安徽农业大学 | Knowledge graph-based network learning resource analysis and personalized recommendation method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210157312A1 (en) * | 2016-05-09 | 2021-05-27 | Strong Force Iot Portfolio 2016, Llc | Intelligent vibration digital twin systems and methods for industrial environments |
| CN115699050A (en) * | 2019-11-05 | 2023-02-03 | 强力价值链网络投资组合2019有限公司 | Value chain network control tower and enterprise management platform |
| EP4323940A2 (en) * | 2021-04-16 | 2024-02-21 | Strong Force VCN Portfolio 2019, LLC | Systems, methods, kits, and apparatuses for digital product network systems and biology-based value chain networks |
-
2023
- 2023-08-31 CN CN202311109073.XA patent/CN116860978B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110289101A (en) * | 2019-07-02 | 2019-09-27 | 京东方科技集团股份有限公司 | A kind of computer equipment, system and readable storage medium storing program for executing |
| CN113361285A (en) * | 2021-06-30 | 2021-09-07 | 北京百度网讯科技有限公司 | Training method of natural language processing model, natural language processing method and device |
| CN114707000A (en) * | 2021-08-09 | 2022-07-05 | 北京智源人工智能研究院 | Knowledge graph-based question-answer library generation method and device, electronic equipment and storage medium |
| CN114861069A (en) * | 2022-06-07 | 2022-08-05 | 安徽农业大学 | Knowledge graph-based network learning resource analysis and personalized recommendation method |
Non-Patent Citations (6)
| Title |
|---|
| Integrative generalized convex clustering optimization and feature selection for mixed multi-view data;Minjie Wang 等;《The Journal of Machine Learning Research》;第22卷(第1期);2498–2571 * |
| Junyi Gao 等.PopNet: Real-Time Population-Level Disease Prediction with Data Latency.《WWW '22: Proceedings of the ACM Web Conference 2022》.2022,2552–2562. * |
| 基于知识图谱的主动交互推荐算法研究;赵梦媛;《中国优秀硕士学位论文全文数据库 信息科技辑》(第07期);I138-299 * |
| 基于知识图谱的承包商智能推荐系统的研究与设计;邢威;《中国优秀硕士学位论文全文数据库 信息科技辑》(第03期);I138-656 * |
| 面向复杂网络的异常检测研究进展;苏江军 等;《控制与决策》;第36卷(第06期);1293-1310 * |
| 面向智能招聘的数据挖掘方法及其应用;秦川;《中国博士学位论文全文数据库 信息科技辑》(第09期);I138-24 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116860978A (en) | 2023-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111737495B (en) | Middle-high-end talent intelligent recommendation system and method based on domain self-classification | |
| CN107230174B (en) | Online interactive learning system and method based on network | |
| CN104794212B (en) | Context sensibility classification method and categorizing system based on user comment text | |
| Mondal et al. | A course recommendation system based on grades | |
| Pong-Inwong et al. | Sentiment analysis in teaching evaluations using sentiment phrase pattern matching (SPPM) based on association mining | |
| CN113886567A (en) | Teaching method and system based on knowledge graph | |
| Lalata et al. | A sentiment analysis model for faculty comment evaluation using ensemble machine learning algorithms | |
| Wang et al. | Topic sentiment analysis in online learning community from college students | |
| CN116860978B (en) | Primary school Chinese personalized learning system based on knowledge graph and large model | |
| Chen et al. | ADOL: a novel framework for automatic domain ontology learning | |
| CN117875412A (en) | Method for constructing computer education knowledge graph based on knowledge graph | |
| Hassan | Designing a flexible system for automatic detection of categorical student sentiment polarity using machine learning | |
| Ross et al. | A case-based reasoning system for conflict resolution: design and implementation | |
| Marivate et al. | An intelligent multi-agent recommender system for human capacity building | |
| CN117216221A (en) | Intelligent question-answering system based on knowledge graph and construction method | |
| Khanam | Sentiment analysis of user reviews in an online learning environment: analyzing the methods and future prospects | |
| Abdussalam et al. | BERT implementation on news sentiment analysis and analysis benefits on branding | |
| CN113326348A (en) | Blog quality evaluation method and tool | |
| CN111563162A (en) | MOOC comment analysis system and method based on text emotion analysis | |
| Singh et al. | Analysis of Student Sentiment Level Using Perceptual Neural Boltzmann Machine Learning Approach for E-learning Applications | |
| Charnine et al. | Optimal automated method for collaborative development of universiry curricula | |
| Qamar et al. | Text mining | |
| Tamrakar et al. | Student sentiment analysis using classification with feature extraction techniques | |
| Zhao et al. | Research on emotion-embedded design flow based on deep learning technology | |
| Ngo et al. | Exploration and integration of job portals in Vietnam |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |