Personalized learning recommendation method based on personalized knowledge graph

Abstract

The invention discloses a personalized learning recommendation method based on a personalized knowledge graph. The method comprises the steps that the knowledge graph is constructed, and a learner fills in learning style self-description and learning target expected content; the current learning condition of the learner is subjected to primary research evaluation through the enrollment test module; an initial learner portrait model is established, a learner enters a database for learning, and a learning evaluation module performs staged evaluation on the learning condition of the learner; updating a learner portrait model, and generating a learning path and learning resources adaptive to the learner by using a knowledge graph technology and a reinforcement learning DQN network; updating the learning path database; the new learning path is recommended to the learner; according to the method, the learning data of the students are analyzed to determine the personal basic conditions of the students, the learner portrait model is constructed, the knowledge graph technology and the learning recommendation algorithm are utilized to adapt the personalized learning features to the learning knowledge element sequence, and a personalized learning push path recommendation mode is constructed to assist the students in improving the learning effect.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and in particular relates to a personalized learning recommendation method based on a personalized knowledge graph. Background Art

[0002] As lifelong learning becomes more popular, most people are improving their abilities through continuous learning. In order to help users learn better, there are many training courses set up for various exams on the market. However, most training courses set up courses for most users. However, each user has different learning interests, cognitive levels, and learning abilities, which will lead to different acceptance levels and learning progress during the learning process, making each user's learning situation uneven, affecting the user's learning efficiency, and also reflecting that the courses or teaching content recommended by such a learning model cannot meet the actual needs of users, affecting their learning effects;

[0003] Even if some platforms are able to launch personalized learning courses, they are unable to make timely adjustments based on the learning situation of customers at different time periods, which will lead to the problem that the teaching content does not match the actual situation of users, seriously affecting their learning effects;

[0004] However, in the knowledge management systems of various platforms, the extraction and organization of knowledge points often rely on manual operations, which results in a large workload, low manual organization efficiency, and is easily affected by subjective judgment during manual processing. With the development of artificial intelligence technology, large AI models such as BERT and GPT have made significant progress in the field of natural language processing, providing the possibility for automated knowledge extraction. Although some models are trained through deep learning algorithms, they can understand and process large amounts of text data and identify entities, concepts, attributes, and the relationships between them in the text. The existing technical solutions have the following problems:

[0005] 1. Currently, most knowledge management systems use keyword matching and expert systems to extract and organize knowledge points.

[0006] 2. The keyword matching method relies on a predefined keyword list and identifies relevant knowledge points by searching for keywords in documents. However, this method often fails to understand the context and easily ignores information that does not contain keywords but is still important.

[0007] 3. Expert systems rely on the knowledge and rules of domain experts to extract and organize knowledge points through a series of customized rules. However, this method requires a lot of manual participation and is difficult to adapt to the ever-changing knowledge system.

[0008] 4. Another method is to use traditional relational databases to store knowledge points, but this method has limitations in dealing with complex relationships and dynamically updating knowledge graphs because it cannot effectively represent complex relationships and attributes between entities.

[0009] At the same time, there are still some limitations in the existing technology in data processing:

[0010] Existing technologies often require a lot of manual intervention when processing large amounts of unstructured data, which not only increases costs but also reduces efficiency.

[0011] It is difficult for existing technologies to accurately capture the complex relationships between knowledge points, especially in cross-domain and cross-document situations. The construction and maintenance of knowledge graphs often require professional knowledge and skills, which limits their application in a wider range of fields; existing technical solutions perform poorly in the dynamic update of knowledge graphs and are difficult to adapt to the rapidly changing information environment. In summary, the extraction of knowledge points in the existing technical data processing process is highly dependent on manual labor, with low work efficiency and high costs; and the relationship between knowledge points is difficult to accurately capture, which increases the difficulty of knowledge point extraction and processing. Even if some existing technologies use knowledge graphs, the subsequent dynamic update and maintenance are more difficult. Therefore, the existing technologies have problems such as inaccurate knowledge point extraction, time-consuming knowledge graph construction and difficulty in maintenance. . Summary of the invention

[0012] Purpose of the invention: In order to overcome the above shortcomings, the purpose of the present invention is to provide a personalized learning recommendation method based on personalized knowledge graph, which builds a learner portrait model based on the user's current actual learning situation and the set goals, combined with learning ability, learning level, learning interest, and learning style, records the learning process data through the learning process database, and conducts timely evaluation through the learning evaluation module, and adjusts the learning content in a timely manner according to the evaluation results.

[0013] Analyze the learning data generated by students during the learning process on the platform to clarify their basic personal information and build a learner portrait model. By combining subject knowledge graphs, using knowledge graph technology and learning recommendation algorithms, the personalized learning features are adapted to the learning knowledge element sequence, search and optimize the sequence combination of subject knowledge elements, and build a learning path recommendation model to provide different personalized learning recommendations to different students, helping them improve their learning outcomes.

[0014] Technical solution: In order to achieve the above object, the present invention provides a personalized learning recommendation method based on personalized knowledge graph, comprising:

[0015] S1): To build a knowledge graph, first pre-process the data source, then automatically identify entities in the text through a dedicated knowledge point extraction fine-tuning model, and extract the relationship between entities to obtain pre-processed knowledge points and relationships; then disambiguate the knowledge points and relationships; establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges, and then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph;

[0016] S2): When learners first enter the personalized learning platform through the learner interaction interface, they must first provide basic personal information and complete registration;

[0017] Afterwards, learners enter the learning style determination module and learning goal setting module respectively under the guidance of the personalized learning platform wizard to complete their own learning style assessment and learning goal setting, including selectively completing the learner's learning style scale questionnaire, filling in a self-description of the learning style and the expected content of the learning goal; and taking the entrance test module to conduct an initial assessment of the learner's current learning situation;

[0018] S3): Initial learner portrait model establishment, that is, the personalized learning recommendation platform builds a preliminary learner portrait model through the learner portrait model establishment module according to the collected user information; the learner portrait model establishment module processes the collected information and stores it in the user model database;

[0019] S4): The learner enters the database to study, and the learning process recording module records the data of the learner's entire learning process, records the sequence of the learner's accessed pages, learning materials, access time information in real time, and instantly updates the user learning record database;

[0020] S5): The learning assessment module conducts periodic assessments on the learners’ learning status. That is, after each unit of knowledge is learned, the learning assessment module assesses the learners’ learning status recorded by the learning process recording module, and sends the assessment results to the learner portrait model and the personalized learning recommendation engine;

[0021] S6): updating the learner portrait model, that is, updating the learner portrait model according to the evaluation results of the learning evaluation module, and storing the updated learner model in the learner model database;

[0022] S7): Based on the updated learner portrait model and learning process records, the personalized learning recommendation engine calculates the learner's mastery of knowledge elements in real time, generates a learner portrait based on this, and calculates the user's comprehensive feature vector. Then, the knowledge graph technology and the reinforcement learning DQN network are used to adaptively and self-organize the learning path and learning resources that are suitable for the learner based on the calculated user comprehensive feature vector, providing users with accurate personalized learning guidance services;

[0023] S8): Update the learning path database, and store the learning path suitable for the learner newly generated in S7) into the learning path database;

[0024] S9): Personalized learning content recommendation recommends new learning paths to learners from the learning path database;

[0025] S10): Then the learner repeats step 4) to step 9) until all the knowledge points to be learned are completed.

[0026] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific construction method of the knowledge graph in S1) is as follows:

[0027] S11): data preprocessing, i.e. processing the data source of the knowledge points to be extracted, wherein the data source of the knowledge points to be extracted includes structured data and unstructured data, the structured data is directly input into the mapper, and the text data in the unstructured data is cleaned;

[0028] S12): extracting knowledge points and relationships from the data source, first building a dedicated knowledge point extraction fine-tuning model, inputting the data after data preprocessing into the knowledge point extraction fine-tuning model, automatically identifying entities in the text through the knowledge point extraction fine-tuning model, and extracting the relationships between the entities to obtain preprocessed knowledge points and relationships;

[0029] S13): Knowledge point and relationship disambiguation, that is, identification and mapping of different descriptions of the same knowledge point; the specific disambiguation process is as follows:

[0030] S131): Establishing a standard knowledge point library, that is, first constructing a standard set of knowledge point entities, and using the established standard set of knowledge point entities as a mapping target set to disambiguate knowledge points extracted from the text;

[0031] S132): Similarity calculation and mapping, i.e., using text embedding technology in natural language processing to convert knowledge points and related texts into vectors; then similarity calculation is performed to determine whether the extracted different expressions point to the same knowledge point by calculating the cosine similarity or Euclidean distance between the text embedding vectors. The cosine similarity is used to measure the angle between the vectors to determine the similarity between the two.

[0032] S133): semantic disambiguation based on context: to ensure the accuracy of the disambiguation process, the sliding window technology is used to take context information into consideration, so that in the process of identifying knowledge points, not only the current sentence but also the previous and next sentences are considered, and the knowledge point relationship is extracted from the pre-processed knowledge points;

[0033] S134): Iterative verification and feedback optimization: After disambiguation, the knowledge points and relationships extracted in the previous step will be mapped to the knowledge point entity standard set, and the two will be matched to find possible corresponding relationships. Further verification will be performed through the extracted relationships. If there is no match, the entity will be deleted;

[0034] S14): Knowledge graph construction: first establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges. Then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph. In the later use process, use the AI ​​model regularly to re-analyze the text data and update the knowledge graph to ensure that it reflects the latest knowledge points and their changes in a timely manner.

[0035] In the personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific method of data cleaning of text materials in unstructured data in S11) is as follows:

[0036] S111): Text cleaning: First, clean the text data in the unstructured data, use the deep learning model to classify different text areas in the text, distinguish the irrelevant content from the main text through the context and semantic features of the text, automatically identify the main text and irrelevant content, and remove the irrelevant content through regular expressions; eliminate the noise in the text data, delete the extra spaces, special characters, irrelevant HTML tags, and repeated text;

[0037] S112): Sentence processing: In order to provide a clear context for subsequent knowledge point extraction, the cleaned text needs to be sentence-separated;

[0038] S113): Extracting repeated content, extracting repeated content in the text data by detecting repeated sentences and repeated paragraphs, specifically:

[0039] Detect repeated sentences and use text similarity detection algorithms to identify and merge similar or repeated sentences; Detect repeated paragraphs: Detect and merge repeated paragraphs for longer texts through text comparison and clustering algorithms;

[0040] S114): Context preservation: In order to preserve the original context of the text data, the following measures are taken:

[0041] Maintain paragraph structure: While dividing sentences, maintain the paragraph structure of the original text so that the complete paragraph information can be referenced when extracting knowledge points;

[0042] Use sliding windows: In order to capture a wider range of contextual information, a sliding window technique can be used when extracting knowledge points, which allows the model to consider the information of a sentence and its surrounding sentences at the same time.

[0043] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific process of sentence processing in S112) is as follows:

[0044] S1121): Based on punctuation: using periods, question marks, exclamation marks and other punctuation marks as sentence separators;

[0045] S1122): Sentence segmentation using AI models or natural language processing libraries based on natural language processing tools that can more accurately identify sentence boundaries even in the absence of obvious punctuation;

[0046] S1123): Dependency-based syntactic analysis: By analyzing the dependency relationship in a sentence, the main and subordinate structures of the sentence are determined, so as to divide the sentence more accurately; its core idea is to decompose the sentence into a "dependency relationship" graph between words, so as to determine the predicate verb, subject, object elements in the sentence and their mutual relationship.

[0047] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific analysis process of dependency syntax analysis in S1123) is as follows:

[0048] S11231): Input preprocessing, which includes word segmentation, denoising, and normalization of the input text; the details are as follows:

[0049] Text cleaning: First, the input text data needs to be cleaned and preprocessed, specifically removing noise, symbols, irrelevant HTML tags, and normalizing complex text, such as converting it to lowercase and correcting typos;

[0050] Word segmentation: Before analyzing the syntax, use word segmentation tools to split the sentence into individual words or phrases;

[0051] S11232): Part-of-speech tagging

[0052] First, use natural language processing tools to mark the part of speech for each word. That is, use the part-of-speech tagger to tag each word or phrase after word segmentation and determine the grammatical role of each word or phrase in the sentence, such as noun, verb, or adjective. Part-of-speech tagging provides basic information for subsequent dependency analysis.

[0053] Then simplify the parts of speech: simplify the parts of speech according to the specific application scenarios of words or phrases, such as simplifying detailed verb forms into a single "verb" category;

[0054] S11233): Dependency relationship parsing, using the dependency syntax analysis model to generate a dependency relationship tree, as follows:

[0055] First, sentence component analysis is performed: dependency syntactic analysis is used to analyze the dependency relationship between words in a sentence and obtain a dependency tree. The goal of dependency syntactic analysis is to find the subordinate relationship between the "core word" and other words in a sentence.

[0056] Then determine the core components of the sentence: Analyze the predicate verb in the sentence and the related subject, object, and adverbial components. For example, `subject` depends on `verb`, and `object` also depends on `verb`. These components constitute the basic structure of the sentence.

[0057] Model selection: Select the appropriate dependency syntax model according to the actual application and build dependency relationships;

[0058] S11234): Extraction of main structure, i.e. analyzing the subject, predicate, object, attributive, adverbial and complement in the sentences in the text to identify the main components of the sentences;

[0059] S11235): Dependency tree construction, using visualization tools to display dependency relationships, so as to analyze sentence structure more intuitively, as follows:

[0060] Dependency tree structured representation: A dependency tree is generated for each sentence. The root node of the tree is the predicate verb of the sentence. Other words are attached to the trunk of the dependency tree in sequence according to the dependency relationship. The dependency tree makes it easy to see the master-slave relationship of the components in the sentence, thereby achieving grammatical analysis and sentence reorganization.

[0061] Dependency tree visualization: Use spaCy's displaCy tool or other dependency tree visualization tools to visualize the parsed dependency relationships, which helps to understand the sentence structure more intuitively.

[0062] S11236): Syntax rule verification and optimization

[0063] Dependency tree trunk and subordinate structure identification: By analyzing the dependency tree, the trunk structure of the sentence can be determined, that is, the predicate verb and its main modifying components. The subject and object directly depend on the predicate verb, while the attributive and adverbial subordinate components depend on the core component.

[0064] Optimization of syntactic rules: According to the needs of the field, some specific syntactic rules are set to correct dependency relationships. For example, for complex sentence patterns such as compound sentences and parallel sentences, conjunctions can be processed through dependency syntax to clarify the position and function of conjunctions in the sentence structure and ensure the accuracy of semantic dependency relationships.

[0065] In the personalized learning recommendation method based on personalized knowledge graph described in the present invention, the context embedding model is used in the knowledge point and relationship disambiguation in S13), and the extracted knowledge points and their contexts are converted into high-dimensional vectors, so as to achieve more accurate semantic matching. Compared with traditional word vectors, context embedding can better capture the different meanings of the same word in different contexts and improve the robustness of disambiguation;

[0066] The details are as follows:

[0067] For the input text t, use the contextual embedding model BERT to convert the input text into a high-dimensional vector representation v;

[0068] v t =BERT(t)

[0069] Among them, v t is the context embedding vector of text t,

[0070] Let cosine similarity calculate the angle between two vectors, defined as:

[0071]

[0072] Among them, v1 and v2 are the contextual embedding vector representations of the two texts;

[0073] In dynamic weighted cosine similarity, a weight w is assigned to each dimension, considering that different dimensions contribute differently to the similarity. i , so that the similarity calculation can more flexibly reflect the important information in the text:

[0074]

[0075] Where d is the dimension, usually d is the hidden layer size of the BERT model;

[0076] Among them, w i is the weight used for weighted cosine similarity, which is a weight dynamically calculated based on context or category, and is used to measure the contribution of different dimensions to similarity. If two knowledge points are consistent in a certain category, the weight w of the dimension related to the category can be increased. i ;

[0077] The w i Based on the following factors

[0078] Term frequency (TF): the frequency with which a word appears in a text, reflecting its importance;

[0079] Inverse Document Frequency (IDF): how rare a word is in the entire corpus;

[0080] w i =TFi·IDFi

[0081] It can also be calculated by semantic level similarity, as follows:

[0082] Assume two sentences s1 and s2, their semantic vector representations are and

[0083] SBERT uses cosine similarity to measure the similarity between two sentences s1 and s2:

[0084]

[0085] Compared with traditional sentence representation methods, SBERT can capture more fine-grained semantic information, making the similarity calculation between two sentences at the semantic level more accurate;

[0086] Multi-level similarity:

[0087] In order to match semantics more precisely, the similarity of knowledge points at multiple levels can be calculated; the details are as follows:

[0088] First, the similarity based on the word level is calculated, and then the similarity at the sentence level or paragraph level is calculated. The final similarity can be the weighted average of the similarities at each level:

[0089] total_similarity(x1, x2)

[0090] =α·word_similarity(x1,x2)+β·sentence_similarity(x1,x2)+γ

[0091] document_similarity(1,x2)

[0092] Among them, α, β, and γ are the weights of similarity at different levels, which are adjusted according to the specific application scenario;

[0093] In the processing process, if there are multiple representations of the same entity, we fine-tune the large model to determine whether they are the same knowledge point, and then perform mapping recognition to obtain the final knowledge points and relationships;

[0094] Mapped to the knowledge point entity standard set through similarity calculation,

[0095] It also includes a mapping method, and introduces context-aware semantic matching in the matching process by capturing the context of knowledge points in the text; context analysis combines the context information of surrounding words and sentences to enhance the ability to distinguish synonyms or related knowledge points.

[0096] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific construction process of the knowledge point extraction fine-tuning model in S12) is as follows:

[0097] (1) Data preparation, that is, preparing enough domain-related text data, including annotations of knowledge points. For different expressions of knowledge points, it is necessary to prepare a standardized set of knowledge points to facilitate subsequent mapping and verification. The data source can be manually annotated text or use existing knowledge bases, such as Wiki or expert-annotated domain data sets;

[0098] (2) Model selection

[0099] Choose a pre-trained large language model (such as BERT, GPT, etc.) as the pre-training model. This model has been pre-trained with massive general data and has good language understanding capabilities. For the task of knowledge point extraction, you can choose a model based on the Transformer architecture. These models perform well in natural language processing tasks.

[0100] (3) Label and clean the text

[0101] Preprocess the data, including text cleaning (removing noise and irrelevant content) and sentence segmentation. Then, use annotation tools or write rules to mark entities and relationships in the text. This step can use tools such as SpaCy or NLTK for automatic sentence segmentation and part-of-speech tagging.

[0102] (4) Fine-tune the pre-trained model to obtain a fine-tuned knowledge point extraction fine-tuning model.

[0103] For the knowledge point extraction task, the pre-trained model such as BERT or GPT is fine-tuned to obtain the fine-tuned model. The training process of the pre-trained model is completed through the following stages:

[0104] Entity recognition: Use pre-trained models to automatically identify entities in text, such as concepts, facts, and principles;

[0105] Relationship extraction: extract the relationships between text entities automatically identified by the pre-trained model, such as inclusion, subordination, and predecessor;

[0106] Disambiguation processing: For different descriptions of the same knowledge point, vectorization technology is used to calculate similarity and map similar expressions to the same entity;

[0107] (5) Model verification and optimization

[0108] The fine-tuned model is tested on the validation set to evaluate its accuracy in knowledge point recognition and relationship extraction tasks. The similarity algorithm is used to verify the accuracy and consistency of the extracted entities and relationships. If the accuracy and consistency meet the requirements, it means that the model can meet the needs.

[0109] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the learner portrait model construction process in S2) is as follows:

[0110] S21): Data collection, collecting various data of learners, including learning behavior data, learning resource interaction data, and test scores, and using association rule algorithms and data mining technology to mine users' learning process data;

[0111] S22): Preprocess the collected data, including cleaning outliers and missing values ​​to ensure data quality. Through attribute matching and feature extraction, combined with learners’ personal attributes and learning styles, various learning portrait modules are integrated using statistical methods to obtain learners’ learning ability, cognitive level, and learning goal portraits.

[0112] S23): Extract the learner's personalized learning features from the three dimensions of learning ability, cognitive level, and learning goals. Use AprioriAll to quantify the feature labels of learning ability, cognitive level, and learning goals to form the learner's learning feature matrix. Each row represents a learner, and each column corresponds to a feature dimension.

[0113] S24): Then, based on the learning feature matrix, a machine learning algorithm is used to group or classify the learners to construct a learner portrait model.

[0114] In the personalized learning recommendation method based on personalized knowledge graph described in the present invention, the reinforcement learning DQN network in the personalized learning recommendation engine makes a decision through the interaction between the agent and the environment, and the decision needs to bring as many rewards as possible; the agent here is the personalized learning recommendation engine, the environment is the user's state and knowledge graph, and the reward is the income of the learning effect;

[0115] The specific working process is as follows:

[0116] S61): First, the data is processed to obtain user portrait features.

[0117] The user portrait feature is composed of a learning ability vector c, a cognitive level vector l, and a learning goal vector g. The learning ability vector c, the cognitive level vector l, and the learning goal vector g are calculated respectively, and then the user comprehensive feature vector is processed. The specific process is as follows:

[0118] The following normalization formulas are involved:

[0119]

[0120] Where X min is the minimum value, X max is the maximum value, the purpose is to map the range of data to [0,1];

[0121] The learning ability includes memory, logical reasoning ability, learning speed, and learning concentration:

[0122] The memory dimension is calculated by using the memory-type question score and the total score to calculate the score rate M as an indicator;

[0123]

[0124] Normalization:

[0125]

[0126] The logical reasoning ability dimension uses the score rate L of reasoning questions as an indicator;

[0127]

[0128] Normalization:

[0129]

[0130] Learning speed dimension: the ability of users to effectively complete learning tasks within a unit of time, reflecting the speed at which students accept and master new knowledge; including average learning time, learning concentration

[0131] Average study time:

[0132]

[0133] The learning speed is inversely proportional to the average learning time L kp

[0134]

[0135] Normalization:

[0136]

[0137] Study concentration:

[0138] The learning concentration refers to the ability of students to maintain concentration and avoid distraction during the learning process, reflecting the degree of learning input;

[0139] The main indicators of learning concentration are as follows:

[0140] Continuous learning time period, number of learning interruptions, and other abnormal operations;

[0141] Learning concentration index

[0142]

[0143] Average study time

[0144]

[0145] Learning concentration final score formula

[0146]

[0147] A min and A max are the minimum and maximum values ​​of the average continuous learning time of users.

[0148] D min and D max are the minimum and maximum values ​​of distraction indicators of all users, respectively;

[0149] Among them, ω1 and ω2 are weights, with a value of 0.5;

[0150] The learning ability matrix is ​​represented by:

[0151] C=[M norm L norm L kp_norm A score ]

[0152] The cognitive levels in the cognitive level vector l are as follows:

[0153]

[0154] Among them, A total The total score of all tests in the subject for the user; M total is the sum of the full scores of all tests in the subject; l is the ratio of the total score of all tests in the subject to the sum of the full scores of all tests in the subject;

[0155] Normalization:

[0156]

[0157] The learning goal in the learning goal vector g is:

[0158] According to the user's target school and target major, knowledge points are selected from the knowledge graph, and then the proficiency of each knowledge point is calculated based on the target score.

[0159] Assume that the knowledge points obtained by screening according to the target colleges and majors are as follows:

[0160] k=[k1 k2 ... k n ]

[0161] The weight of each knowledge point is

[0162] ω(k i ), where i is 1....n;

[0163] Normalize the weight of each knowledge point

[0164]

[0165] According to the normalized weight, the target total score S goal Assign to each knowledge point and adjust the base value to ensure that the weight is greater than 1

[0166] S target (k i )=ω′(k i )×(S goal )+1

[0167] The matrix of the user learning target vector g is as follows:

[0168] g=[S target (k1) S target (k2) ... S target (k i )]

[0169] Combine the above features to form the user's comprehensive feature vector

[0170] F student =[c, l, g]

[0171] Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector;

[0172] Using the comprehensive feature vector F student As clustering features, students are divided into K groups and each student is assigned a cluster label G j (j=1, 2, ..., K), represents the group to which the student belongs, and the cluster label G j As features, incorporated into the state representation of subsequent models;

[0173] S62): Knowledge graph embedding, that is, collecting all knowledge points and constructing a knowledge point set K = k1, k2, ..., k N ;

[0174] The Node2Vec algorithm is used to embed the knowledge points in the knowledge graph into the low-dimensional vector space to obtain the embedding vector e of the i-th knowledge point i ;

[0175] Then add the user's comprehensive feature vector, the embedding vector of the knowledge point, and the student's clustering label to obtain the corresponding user state vector, as follows:

[0176] The user state vector S with the user's comprehensive feature vector added is expressed as follows:

[0177] s=[s1,s2,...,s N ,c,l,g]

[0178] Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector;

[0179] Add the student's cluster label G j and the embedding vector e of the knowledge point i ;

[0180] The knowledge point vector is calculated as follows:

[0181]

[0182] in,

[0183] is the mastery level: that is, the students’ i The degree of mastery, range [0,1];

[0184] is the learning frequency: that is, the number of times a student learns knowledge point k i The number of times;

[0185] is the test performance: that is, the student's performance on knowledge point k i Average test scores on

[0186] is the most recent learning time interval: that is, since the last learning of knowledge point k i the time interval since;

[0187] e i Embedding vector of knowledge points;

[0188] f i is a frequent pattern feature;

[0189] Knowledge point k i The support is:

[0190]

[0191] Add the student's cluster label G j The user state vector S after is expressed as follows

[0192] s=[s1,s2,...,s N ,c,l,g,G j ]

[0193] If the user does not fully grasp the knowledge point, the optional action space is as follows:

[0194] A′(s)={k i ∈A(s)|g i >0}

[0195] A(s) is the set of knowledge points that satisfy the prerequisite relationship but are not fully mastered, g i >0 indicates knowledge point k i It is the student’s learning goal;

[0196] Reward learners for completing their learning goals. The specific reward function is as follows

[0197] r t =w(k i )×Δs i ×f(c,l)×G(k i )×PM(k i )×AR(k i )×LF(k i )×FC(k i )

[0198] Among them, w(k i ) is the knowledge point k i Importance weight

[0199] Δs i For students to understand the knowledge point k i Improved mastery

[0200] f(c, l) is the adjustment function of students’ learning ability and cognitive level, as follows:

[0201] f(c,l)=δ1·c+δ2·l, where δ1 and δ2 are weight parameters, with values ​​of 0.7 and 0.3 respectively;

[0202] G(k i ) is the learning objective weighting function

[0203]

[0204] Among them, g i is the i-th element of the student's learning goal vector, indicating the student's understanding of knowledge point k iThe degree of learning needs; γ is a constant greater than 1, which is the bonus coefficient for the learning target knowledge points;

[0205] When i >0(Knowledge point k i is the student's learning goal), G(k i ) = γ, the reward is amplified by γ times (usually 1.2-2.0), and the expected reward for selecting this knowledge point is submitted. The personalized learning recommendation engine is more inclined to recommend this knowledge point;

[0206] When i =0, G(k i )=1, the reward remains unchanged, and the personalized learning recommendation engine will not be particularly biased towards recommending this knowledge point;

[0207] PM(k i ) is a frequent pattern weighting function, which adjusts the reward according to the importance of the knowledge point in the frequent item set, reflecting the importance of the knowledge point in the learning group, as follows:

[0208] PM(k i )=1+λ PM ×Support(k i )

[0209] Among them, λ PM To adjust the parameter, the degree of influence on the reward, usually a small positive number such as 0.1 or 0.2;

[0210] Support(k i ) is the knowledge point k i The support degree of k i Frequency of occurrence in frequent itemsets;

[0211] By considering the frequency of occurrence of knowledge points in frequent learning patterns, the personalized learning recommendation engine is guided to recommend knowledge points that are commonly learned in the group;

[0212] In the association rule algorithm, AR(k i )The association rule weighting function is:

[0213] AR(k i )=1+λ AR ×Conf(k i )

[0214] Among them, λ AR To adjust the parameters, control the influence of association rules on rewards, and usually take a positive value;

[0215] Conf(k i )Knowledge point k iThe highest confidence in the association rule, denoted by k i The strength of the connection with the knowledge points already learned;

[0216] Using the knowledge learned: By considering k i The association with the knowledge points that students have already learned guides the model to recommend knowledge points that are closely related to what has been learned, thus promoting the coherent learning of knowledge;

[0217] LF(k i )The learning frequency adjustment function is

[0218]

[0219] Among them, λ LF Take w(k i ), the more important the knowledge point is, the smaller the reduced reward is;

[0220] As the learning frequency increases, the function value will gradually decrease, reducing the contribution to the reward and preventing students from excessively repeating familiar knowledge points;

[0221] FC(k i ) Forgetting curve adjustment function

[0222]

[0223] Among them, λ FC Take the proportional function of the knowledge point weight, with a value range of 0.5-1; when Increase, and the reward decreases, prompting students to review this knowledge point;

[0224] S63): Reinforce the learning DQN network to update the network parameters, as follows:

[0225] Input layer: user state vector s

[0226] Output layer: the Q value of each optional action (the Q value is the mathematical expectation of the sum of all future state returns);

[0227] The DQN algorithm network training steps are as follows:

[0228] A. Status acquisition: Get the current status s t ;

[0229] B. Action selection: According to the ∈ greedy strategy, select an action from the set of optional actions;

[0230] C. Execution action: Students learn knowledge points, update their mastery level, and obtain a new state s t+1 ;

[0231] D. Reward calculation: Calculate the immediate reward based on the reward function;

[0232] E. Experience storage: t , a t , r t ,s t+1 ) is stored in the experience replay pool;

[0233] F. Network update: Randomly extract small batches of samples (s j , a j , r j ,s j+1 )

[0234] Conduct training and update network parameters;

[0235] The real-time recommendation process through the recommendation engine is as follows:

[0236] Status acquisition: Get the student's current user status S;

[0237] Action selection: Select the best knowledge point from A′(s) according to the ∈ greedy strategy;

[0238] Knowledge point recommendation: recommend knowledge points to students and provide corresponding learning resources

[0239] S64): Provide feedback and update on learners through user feedback collection, as follows:

[0240] Learning feedback collection: record students’ learning results of recommended knowledge points, update their mastery level, learning frequency, and recent learning time;

[0241] User status update: Update the student's user status S based on learning feedback

[0242] Reinforcement learning DQN network update: Regularly train the reinforcement learning DQN network, incorporate new learning data, and improve model performance;

[0243] Add learner feedback to the experience replay pool

[0244] In reinforcement learning, experience samples are usually represented in the form of four tuples:

[0245] (s t , a t , r t ,s t+1 )

[0246] In order to incorporate students' feedback into experience samples, the structure of experience samples needs to be expanded to include feedback information:

[0247] (s t , a t , r t′ ,s t+1 , ft )

[0248] r t′ Adjusted instant rewards, taking into account student feedback

[0249] f t The student performs action a at time step t t Feedback

[0250] If the student's feedback is positive, increase the reward immediately

[0251] r t′ =r t +δr

[0252] If negative, reduce the reward

[0253] r t′ =r t -δr.

[0254] The personalized learning recommendation method based on personalized knowledge graph described in the present invention, the specific process of constructing the learning path recommendation model in S7) is as follows:

[0255] S601): The learning feature representation is based on the subject knowledge graph, and the subject knowledge elements and related learning resources are extracted using the association rule algorithm to form the learning element of the knowledge graph, which is represented by a triple L = (Ki, Kj, r);

[0256] Among them, r represents the logical relationship between knowledge element Ki and learning resource Kj;

[0257] Then, using knowledge graph technology, we serialize and annotate the attributes of subject knowledge elements and related learning resources, characterizing and describing the basic characteristics of learning elements.

[0258] Combined with the learning needs of learners, the learning elements of the learning activity process are formed into a learning element sequence (s1, s2, ..., s N );

[0259] At this time, the user status S is expressed as follows:

[0260] s[=s1 s2 ... s N ]

[0261] S602): Recommend learning paths, based on the user's current learning state matrix and the prior relationship of knowledge points in the knowledge graph, recommend the next knowledge point with the maximum reward and the corresponding learning materials;

[0262] S603): Learning path evaluation and optimization: During the learning process, learners complete tasks based on the learning paths and resources recommended by personalized learning, and submit feedback on their learning experience and results. The personalized learning recommendation engine collects and analyzes user feedback, combines the learners' goal completion status, compares the expert path with the recommended path, and evaluates the accuracy and practicality of the recommendation;

[0263] Based on evaluation results and user feedback, we continuously optimize the personalized learning recommendation engine to continuously improve the effectiveness of personalized recommendations and user satisfaction.

[0264] It can be seen from the above technical solution that the present invention has the following beneficial effects:

[0265] 1. The personalized learning recommendation method based on personalized knowledge graph described in the present invention builds a learner portrait model based on the user's current actual learning situation and set goals, combined with learning ability, learning level, learning interest, and learning style, records the learning process data through the learning process database, and conducts timely evaluation through the learning evaluation module, and adjusts the learning content in a timely manner according to the evaluation results.

[0266] Analyze the learning data generated by students during the learning process on the platform to clarify their basic personal information and build a learner portrait model. By combining subject knowledge graphs, using knowledge graph technology and learning recommendation algorithms, the personalized learning features are adapted to the learning knowledge element sequence, search and optimize the sequence combination of subject knowledge elements, and build a learning path recommendation model to provide different personalized learning recommendations to different students, helping them improve their learning outcomes.

[0267] 2. The knowledge graph construction in the present invention obtains processed data from data preprocessing, and then automatically extracts knowledge points to determine entities and relationship edges by utilizing the knowledge point extraction fine-tuning model. The extracted knowledge points are further identified and optimized through a variety of disambiguation methods, and the complex relationships between knowledge points are identified and constructed, which effectively improves the efficiency and accuracy of data processing. Finally, the disambiguated data is stored in the graph database model to construct a knowledge graph, realize the automatic construction and update of the knowledge graph, and effectively solve the problems of large data processing workload and time-consuming.

[0268] 3. The knowledge graph of the present invention provides a visual and structured way to present knowledge and decision-making processes. By embedding the knowledge graph into the framework of deep reinforcement learning, the decision-making interpretability of the system can be enhanced, so that learners and system managers can understand the logic and basis behind the recommended learning path and strategy selection. The knowledge graph can also provide a global explanation for the overall strategy learning process, that is, the system can display the relationship between all knowledge points through the knowledge graph and explain the recommended strategy for the entire learning path.

[0269] 4. In the data preprocessing process, the accuracy of data processing is effectively improved, invalid data is effectively removed, and the efficiency of data processing is improved through text cleaning, sentence processing, repeated content extraction, and context retention.

[0270] 5. Extract knowledge points and relationships in the data source by building a special extraction fine-tuning model to accurately identify the relationship between entities, which effectively solves the problem of difficulty in identifying the relationship between entities in current data processing.

[0271] 6. The present invention uses a large AI model combined with vector embedding, semantic similarity calculation, and symbolic reasoning to efficiently identify different representations of the same knowledge point and disambiguate it. The accuracy and flexibility of knowledge point disambiguation are further improved by dynamically updating the standard set and combining domain knowledge. This method is particularly effective in building and maintaining large-scale knowledge graphs, and can ensure the high accuracy and robustness of knowledge graphs.

[0272] 7. The present invention can dynamically adjust the difficulty and progress of learning materials according to the learner's performance, promote the improvement of learning efficiency, and ensure that learners can master the necessary knowledge and skills.

[0273] 8. The machine learning in the present invention can automatically adjust the recommendation strategy through algorithms, and continuously improve the quality of recommendations over time; it supports the modeling of nonlinear relationships and can process complex learning behavior data; it can also automatically identify which factors have the greatest impact on learning outcomes, and optimize recommendations accordingly to better meet user needs.

[0274] 9. The data processing in the present invention supports the storage and rapid query of massive data, ensuring the response speed of the recommendation system; by analyzing large-scale data sets, potential learning trends and patterns are discovered. BRIEF DESCRIPTION OF THE DRAWINGS

[0275] Figure 1 A schematic diagram of the steps of the personalized learning recommendation method based on the personalized knowledge graph described in the present invention;

[0276] Figure 2 A framework diagram for constructing a knowledge graph in the present invention;

[0277] Figure 3 A flowchart for knowledge point and relationship disambiguation in the present invention;

[0278] Figure 4 It is a schematic diagram of the learner portrait model in the present invention;

[0279] Figure 5 This is a framework diagram of the personalized learning recommendation platform based on knowledge graph in the present invention;

[0280] Figure 6 This is a schematic diagram of the structure of the personalized learning recommendation platform based on knowledge graph described in the present invention;

[0281] Figure 7 This is an algorithm flow chart of the reinforcement learning DQN network in the present invention. DETAILED DESCRIPTION

[0282] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments.

[0283] Example

[0284] like Figure 1 A personalized learning recommendation method based on a personalized knowledge graph is shown, comprising:

[0285] S1): To build a knowledge graph, first pre-process the data source, then automatically identify entities in the text through a dedicated knowledge point extraction fine-tuning model, and extract the relationship between entities to obtain pre-processed knowledge points and relationships; then disambiguate the knowledge points and relationships; establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges, and then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph;

[0286] S2): When learners first enter the personalized learning platform through the learner interaction interface, they must first provide basic personal information and complete registration;

[0287] Afterwards, learners enter the learning style determination module and learning goal setting module respectively under the guidance of the personalized learning platform wizard to complete their own learning style assessment and learning goal setting, including selectively completing the learner's learning style scale questionnaire, filling in a self-description of the learning style and the expected content of the learning goal; and taking the entrance test module to conduct an initial assessment of the learner's current learning situation;

[0288] S3): Initial learner portrait model establishment, that is, the personalized learning recommendation platform builds a preliminary learner portrait model through the learner portrait model establishment module according to the collected user information; the learner portrait model establishment module processes the collected information and stores it in the user model database;

[0289] S4): The learner enters the database to study, and the learning process recording module records the data of the learner's entire learning process, records the sequence of the learner's accessed pages, learning materials, access time information in real time, and instantly updates the user learning record database;

[0290] S5): The learning assessment module conducts periodic assessments on the learners’ learning status. That is, after each unit of knowledge is learned, the learning assessment module assesses the learners’ learning status recorded by the learning process recording module, and sends the assessment results to the learner portrait model and the personalized learning recommendation engine;

[0291] S6): updating the learner portrait model, that is, updating the learner portrait model according to the evaluation results of the learning evaluation module, and storing the updated learner model in the learner model database;

[0292] S7): Based on the updated learner portrait model and learning process records, the personalized learning recommendation engine calculates the learner's mastery of knowledge elements in real time, generates a learner portrait based on this, and calculates the user's comprehensive feature vector. Then, the knowledge graph technology and the reinforcement learning DQN network are used to adaptively and self-organize the learning path and learning resources that are suitable for the learner based on the calculated user comprehensive feature vector, providing users with accurate personalized learning guidance services;

[0293] S8): Update the learning path database, and store the learning path suitable for the learner newly generated in S7) into the learning path database;

[0294] S9): Personalized learning content recommendation recommends new learning paths to learners from the learning path database;

[0295] S10): Then the learner repeats step 4) to step 9) until all the knowledge points to be learned are completed.

[0296] In the personalized learning recommendation method based on the personalized knowledge graph described in this embodiment, the specific method for constructing the knowledge graph in S1) is as follows:

[0297] S11): data preprocessing, i.e. processing the data source of the knowledge points to be extracted, wherein the data source of the knowledge points to be extracted includes structured data and unstructured data, the structured data is directly input into the mapper, and the text data in the unstructured data is cleaned;

[0298] S12): extracting knowledge points and relationships from the data source, first building a dedicated knowledge point extraction fine-tuning model, inputting the data after data preprocessing into the knowledge point extraction fine-tuning model, automatically identifying entities in the text through the knowledge point extraction fine-tuning model, and extracting the relationships between the entities to obtain preprocessed knowledge points and relationships;

[0299] S13): Knowledge point and relationship disambiguation, that is, identification and mapping of different descriptions of the same knowledge point; the specific disambiguation process is as follows:

[0300] S131): Establishing a standard knowledge point library, that is, first constructing a standard set of knowledge point entities, and using the established standard set of knowledge point entities as a mapping target set to disambiguate knowledge points extracted from the text;

[0301] S132): Similarity calculation and mapping, i.e., using text embedding technology in natural language processing to convert knowledge points and related texts into vectors; then similarity calculation is performed to determine whether the extracted different expressions point to the same knowledge point by calculating the cosine similarity or Euclidean distance between the text embedding vectors. The cosine similarity is used to measure the angle between the vectors to determine the similarity between the two.

[0302] S133): semantic disambiguation based on context: to ensure the accuracy of the disambiguation process, the sliding window technology is used to take context information into consideration, so that in the process of identifying knowledge points, not only the current sentence but also the previous and next sentences are considered, and the knowledge point relationship is extracted from the pre-processed knowledge points;

[0303] S134): Iterative verification and feedback optimization: After disambiguation, the knowledge points and relationships extracted in the previous step will be mapped to the knowledge point entity standard set, and the two will be matched to find possible corresponding relationships. Further verification will be performed through the extracted relationships. If there is no match, the entity will be deleted;

[0304] S14): Knowledge graph construction: first establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges. Then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph. In the later use process, use the AI ​​model regularly to re-analyze the text data and update the knowledge graph to ensure that it reflects the latest knowledge points and their changes in a timely manner.

[0305] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the specific method of data cleaning of text materials in unstructured data in S11) is as follows:

[0306] S111): Text cleaning: First, clean the text data in the unstructured data, use the deep learning model to classify different text areas in the text, distinguish the irrelevant content from the main text through the context and semantic features of the text, automatically identify the main text and irrelevant content, and remove the irrelevant content through regular expressions; eliminate the noise in the text data, delete the extra spaces, special characters, irrelevant HTML tags, and repeated text;

[0307] S112): Sentence processing: In order to provide a clear context for subsequent knowledge point extraction, the cleaned text needs to be sentence-separated;

[0308] S113): Extracting repeated content, extracting repeated content in the text data by detecting repeated sentences and repeated paragraphs, specifically:

[0309] Detect repeated sentences and use text similarity detection algorithms to identify and merge similar or repeated sentences; Detect repeated paragraphs: Detect and merge repeated paragraphs for longer texts through text comparison and clustering algorithms;

[0310] S114): Context preservation: In order to preserve the original context of the text data, the following measures are taken:

[0311] Maintain paragraph structure: While dividing sentences, maintain the paragraph structure of the original text so that the complete paragraph information can be referenced when extracting knowledge points;

[0312] Use sliding windows: In order to capture a wider range of contextual information, a sliding window technique can be used when extracting knowledge points, which allows the model to consider the information of a sentence and its surrounding sentences at the same time.

[0313] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the specific process of sentence processing in S112) is as follows:

[0314] S1121): Based on punctuation: using periods, question marks, exclamation marks and other punctuation marks as sentence separators;

[0315] S1122): Sentence segmentation using AI models or natural language processing libraries based on natural language processing tools that can more accurately identify sentence boundaries even in the absence of obvious punctuation;

[0316] S1123): Dependency-based syntactic analysis: By analyzing the dependency relationship in a sentence, the main and subordinate structures of the sentence are determined, so as to divide the sentence more accurately; its core idea is to decompose the sentence into a "dependency relationship" graph between words, so as to determine the predicate verb, subject, object elements in the sentence and their mutual relationship.

[0317] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the specific analysis process of dependency syntax analysis in S1123) is as follows:

[0318] S11231): Input preprocessing, which includes word segmentation, denoising, and normalization of the input text; the details are as follows:

[0319] Text cleaning: First, the input text data needs to be cleaned and preprocessed, specifically removing noise, symbols, irrelevant HTML tags, and normalizing complex text, such as converting it to lowercase and correcting typos;

[0320] Word segmentation: Before analyzing the syntax, use word segmentation tools to split the sentence into individual words or phrases;

[0321] S11232): Part-of-speech tagging

[0322] First, use natural language processing tools to mark the part of speech for each word. That is, use the part-of-speech tagger to tag each word or phrase after word segmentation and determine the grammatical role of each word or phrase in the sentence, such as noun, verb, or adjective. Part-of-speech tagging provides basic information for subsequent dependency analysis.

[0323] Then simplify the parts of speech: simplify the parts of speech according to the specific application scenarios of words or phrases, such as simplifying detailed verb forms into a single "verb" category;

[0324] S11233): Dependency relationship parsing, using the dependency syntax analysis model to generate a dependency relationship tree, as follows:

[0325] First, sentence component analysis is performed: dependency syntactic analysis is used to analyze the dependency relationship between words in a sentence and obtain a dependency tree. The goal of dependency syntactic analysis is to find the subordinate relationship between the "core word" and other words in a sentence.

[0326] Then determine the core components of the sentence: Analyze the predicate verb in the sentence and the related subject, object, and adverbial components. For example, `subject` depends on `verb`, and `object` also depends on `verb`. These components constitute the basic structure of the sentence.

[0327] Model selection: Select the appropriate dependency syntax model according to the actual application and build dependency relationships;

[0328] S11234): Extraction of main structure, i.e. analyzing the subject, predicate, object, attributive, adverbial and complement in the sentences in the text to identify the main components of the sentences;

[0329] S11235): Dependency tree construction, using visualization tools to display dependency relationships, so as to analyze sentence structure more intuitively, as follows:

[0330] Dependency tree structured representation: A dependency tree is generated for each sentence. The root node of the tree is the predicate verb of the sentence. Other words are attached to the trunk of the dependency tree in sequence according to the dependency relationship. The dependency tree makes it easy to see the master-slave relationship of the components in the sentence, thereby achieving grammatical analysis and sentence reorganization.

[0331] Dependency tree visualization: Use spaCy's displaCy tool or other dependency tree visualization tools to visualize the parsed dependency relationships, which helps to understand the sentence structure more intuitively.

[0332] S11236): Syntax rule verification and optimization

[0333] Dependency tree trunk and subordinate structure identification: By analyzing the dependency tree, the trunk structure of the sentence can be determined, that is, the predicate verb and its main modifying components. The subject and object directly depend on the predicate verb, while the attributive and adverbial subordinate components depend on the core component.

[0334] Optimization of syntactic rules: According to the needs of the field, some specific syntactic rules are set to correct dependency relationships. For example, for complex sentence patterns such as compound sentences and parallel sentences, conjunctions can be processed through dependency syntax to clarify the position and function of conjunctions in the sentence structure and ensure the accuracy of semantic dependency relationships.

[0335] It should be noted that the subordinate relationship between core words and other words mainly follows the following principles:

[0336] 1. Take the predicate verb as the core: The predicate verb is the root node of the dependency tree, and other words establish dependency relationships around it.

[0337] 2. Dependency of the main components: The subject and object directly depend on the predicate verb, while subordinate components such as attributives and adverbials depend on the core components.

[0338] 3. Hierarchical structure: Through the form of a dependency tree, the master-slave relationship of each component in the sentence is clearly displayed, which facilitates grammatical analysis and sentence reorganization.

[0339] 4. Grammatical rules: Specific syntactic rules will be set according to the needs of specific fields, especially when dealing with complex sentence patterns such as compound sentences and parallel sentences, to ensure the accuracy of semantic dependencies.

[0340] The words in S1235) are linked based on grammatical relations and hierarchical structures, and the linking process follows the following principles:

[0341] -First, use the predicate verb as the root node and trunk of the dependency tree.

[0342] -The main grammatical components (such as subject and object) are directly dependent on the predicate verb, while other modifying components (such as attributives and adverbials) are dependent on the core components they modify.

[0343] -The whole forms a hierarchical structure, clearly showing the master-slave relationship of each component in the sentence.

[0344] In the specific process of linking, the main structure of the sentence is identified by analyzing the subject, predicate, object, attributive, adverbial, complement and other components of the sentence, following specific grammatical rules, especially when dealing with complex sentence patterns such as compound sentences and parallel sentences, to ensure the accuracy of semantic dependencies.

[0345] It should be noted that the predicate verb is the core, the root node of the dependency tree, and an important part of the main structure of the entire sentence;

[0346] The core components and their subordinate relationships are as follows:

[0347] The subject and object directly depend on the predicate verb, while the attributive, adverbial, etc. are subordinate components that depend on the core component.

[0348] 3. The concept of core ingredients:

[0349] - "Core word" refers specifically to the word that is the dominant word in a sentence, usually a predicate verb

[0350] - "Core components of a sentence" is a broader concept, including the predicate verb and the subject, object, adverbial and other sentence components directly related to it;

[0351] This hierarchical structure clearly displays the master-slave relationship of each component in the sentence in the form of a dependency tree.

[0352] Application Scenario

[0353] Machine translation: Dependency syntactic analysis can help machine translation systems better understand the structure of complex sentences and ensure the accuracy of translation.

[0354] -Text Summarization: Dependency parsing can be used to generate concise text summaries by extracting the backbone structure of sentences.

[0355] - Question-Answering System: Dependency parsing can help question-answering systems understand the core components of user questions and generate more accurate answers.

[0356] As an effective method to parse the internal structure of a sentence, dependency parsing has been widely used in text analysis, information extraction and other fields.

[0357] In the process of extracting knowledge points and relationships in the data source (S12) described in this embodiment, an AI big model is used to identify entities and concepts in the text as preprocessing knowledge points. For the application of the big model, knowledge points are mainly extracted through Prompt and mounting a limited knowledge set. It should be noted that the AI ​​big model uses an existing pre-trained model, Ali qwen-max, and it can also select other models that can meet the requirements according to actual needs. Mounting a limited knowledge set is to clarify the limited knowledge point set within the vertical range. This part of the content comes from an external knowledge base, such as Wikipedia or annotations by domain experts within the company, such as a computer teacher defining 408 subject test knowledge points and a custom rule set: according to specific needs, define a set of your own limited knowledge sets as constraints for entity and relationship extraction.

[0358] The following is a prompt text:

[0359] You are a knowledge graph expert in the field of online education, specializing in entity extraction and relationship extraction. You can extract entities from text and the relationships between entities.

[0360] The entity types are limited to the following:

[0361] Facts: Information describing observed facts, knowledge of specific things, such as capitals, dates, formulas, etc.

[0362] Composition: refers to the understanding of the components of things, such as human body structure, engine parts, computer hardware, etc.

[0363] Concept: refers to the abstract and general definition of things, such as democracy, evolution, force, etc.

[0364] Principle: refers to the understanding of how things work, such as Newton's laws of motion, circuit principles, programming languages, etc., including subcategories such as rules and formulas.

[0365] Procedure: refers to the steps and methods of how to complete a task, such as solving math problems, doing experiments, writing essays, etc.

[0366] Core courses: In an education system, basic and key courses are set up to achieve specific educational goals. Core courses usually cover a wide range of basic knowledge and skills. For example: "Advanced Mathematics: A required basic course for university science and engineering students."

[0368] Entity relationships are limited to the following:

[0369] Contains: One entity contains another entity, that is, a larger entity contains a smaller part or subset. For example: "The mathematics course contains algebra, geometry, and calculus."

[0370] Belong to: knowledge point B is part of knowledge point A. For example: "Hardware is part of the computer system."

[0371] Precedence: One entity is the previous state or stage of another entity, usually an earlier stage in a process. For example: "Algebra is a prerequisite course for learning calculus."

[0372] Synonymy: Two entities are identical or very similar in meaning and can be used interchangeably. For example: "Differential and derivative are synonymous in some contexts."

[0373] Reference: reference, related relationship, there is some kind of association or connection between two entities, such as "the instruction word length depends on the length of the opcode and the length and number of the operand addresses."

[0374] Application: apply to, knowledge point A is a kind of ability, method and approach to apply knowledge point B. For example: "Through learning and practice, theoretical knowledge can be transformed into practical skills."

[0375] Correlation: There is some association or connection between two entities, but it is not necessarily a causal relationship. For example: "There is a correlation between study time and test scores."

[0376] Follow these steps:

[0377] Step 1: High-precision entity recognition and entity extraction

[0378] - Deeply understand the entity context and accurately identify the entity attributes and meaning.

[0379] -Entity types are limited to the provided types, and new categories cannot be generated arbitrarily.

[0380] Step 2: Complex relationship parsing and extracting relationships between entities

[0381] - Ensure that the parsing logic is rigorous and complies with the construction principles and standards of the knowledge graph.

[0382] -Entity relationships are limited to the pre-provided relationship range, and new relationships cannot be generated arbitrarily.

[0383] Step 3: The final output data format is as follows:

[0384] "entities":[

[0385] {

[0386] "name":"Entity name",

[0387] "type":"Entity type"

[0388] }

[0389] ],

[0390] "relations":[

[0391] {

[0392] "head":"Relationship head entity name",

[0393] "tail": "Relationship tail entity name",

[0394] "type":"Relationship type"

[0395] } ]

[0397] }

[0398] It should be noted that the data format of json type contains array keys such as "entities".

[0399] Example:

[0400] Content: Recursion and iteration are two common methods for solving data structure problems. Recursion is breaking down a problem into smaller sub-problems, while iteration is repeatedly executing a set of instructions until a condition is met.

[0401] result:

[0402] {

[0403] "entities":[

[0404] {

[0405] "name":"Semiconductor Random Access Memory",

[0406] "type":"Concept"

[0407] },

[0408] {

[0409] "name":"Computer performance indicators",

[0410] "type":"Concept"

[0411] },

[0412] {

[0413] "name":"Basic principles of multiplication / division operations",

[0414] "type":"Principle"

[0415] },

[0416] {

[0417] "name":"Calculation method",

[0418] "type":"Principle"

[0419] },

[0420] {

[0421] "name":"Command",

[0422] "type":"Concept"

[0423] },

[0424] {

[0425] "name":"Basic Principles of Cache",

[0426] "type":"Principle"

[0427] },

[0428] {

[0429] "name":"Replacement algorithm for main memory blocks in cache",

[0430] "type":"Principle"

[0431] },

[0432] {

[0433] "name":"Instruction format",

[0434] "type":"Concept"

[0435] },

[0436] {

[0437] "name":"Data Structure",

[0438] "type":"Core Courses"

[0439] }

[0440] ],

[0441] "relations":[

[0442] {

[0443] "head":"Calculation method",

[0444] "tail":"Basic principles of multiplication / division operations",

[0445] "type":"include"

[0446] },

[0447] {

[0448] "head":"Instructions",

[0449] "tail":"Instruction format",

[0450] "type":"include"

[0451] },

[0452] {

[0453] "head":"Replacement algorithm of main memory blocks in cache",

[0454] "tail":"Basic Principles of Cache",

[0455] "type":"Related"

[0456] } ]

[0458] }

[0459] Here is an example scenario:

[0460] enter:

[0461] Advanced mathematics is a basic course for university science and engineering students, including calculus, linear algebra and probability theory. Calculus is a prelude to further study of physics, while linear algebra is closely related to engineering mathematics. Probability theory is mainly used in the field of data analysis.

[0462] Output:

[0463]

[0464]

[0465]

[0466] The knowledge point and relationship disambiguation uses cosine similarity to measure the angle between two vectors to determine whether they represent the same concept. The calculation formula of cosine similarity is:

[0467] Among them, a and b are two vectors, ||a|| and ||b|| are their moduli.

[0468] In this way, similar knowledge points can be mapped to the same entity, thereby building accurate entities and relationships in the knowledge graph. This method can improve the accuracy and robustness of the knowledge graph, allowing the knowledge graph to better reflect the knowledge and information in the text material.

[0469] The knowledge point entity standard set is the basis of knowledge disambiguation, which usually comes from authoritative public resources (such as Wikipedia, DBpedia), specialized knowledge bases annotated by domain experts, or knowledge bases within enterprises. In this process, it is crucial to ensure the accuracy and coverage of the standard set.

[0470] Authoritative public resources: such as structured data provided by WikiData, which can cover basic concepts and entities in multiple fields.

[0471] Expert-labeled specialized data sources: Combined with knowledge in specific fields, enterprises can build customized sets of standard knowledge points based on their business needs, which are particularly suitable for vertical scenario applications (such as mathematics, education, etc.).

[0472] Dynamic update mechanism: During use, the knowledge point entity standard set is regularly updated from the latest literature and databases to continuously introduce new knowledge points, thereby ensuring the real-time nature of the system's knowledge.

[0473] Standardization and naming conventions:

[0474] Name and standardize each knowledge point to ensure uniform naming. For example, a course can be unified as "Linear Algebra" instead of "Linear Algebra", "Linear Algebra Class" and other variants. At the same time, these standardized knowledge points need to contain unique identifiers (such as knowledge point IDs) for subsequent mapping and query.

[0475] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the context embedding model is used in the knowledge point and relationship disambiguation in S13), and the extracted knowledge points and their contexts are converted into high-dimensional vectors, so as to achieve more accurate semantic matching. Compared with traditional word vectors, context embedding can better capture the different meanings of the same word in different contexts and improve the robustness of disambiguation;

[0476] The details are as follows:

[0477] For the input text t, the context embedding model BERT is used to transform the input text into a high-dimensional

[0478] The quantity represents v;

[0479] v t =BERT(t)

[0480] Among them, v t is the context embedding vector of text t,

[0481] Let cosine similarity calculate the angle between two vectors, defined as:

[0482]

[0483] Among them, v1 and v2 are the contextual embedding vector representations of the two texts;

[0484] In dynamic weighted cosine similarity, a weight w is assigned to each dimension, considering that different dimensions contribute differently to the similarity. i , so that the similarity calculation can more flexibly reflect the important information in the text:

[0485]

[0486] Where d is the dimension, usually d is the hidden layer size of the BERT model;

[0487] Among them, w i is the weight used for weighted cosine similarity, which is a weight dynamically calculated based on context or category, and is used to measure the contribution of different dimensions to similarity. If two knowledge points are consistent in a certain category, the weight w of the dimension related to the category can be increased. i ;

[0488] The w i Based on the following factors

[0489] Term frequency (TF): the frequency with which a word appears in a text, reflecting its importance;

[0490] Inverse Document Frequency (IDF): how rare a word is in the entire corpus;

[0491] w i =TFi·IDFi

[0492] It can also be calculated by semantic level similarity, as follows:

[0493] Assume two sentences s1 and s2, their semantic vector representations are and

[0494] SBERT uses cosine similarity to measure the similarity between two sentences s1 and s2:

[0495]

[0496] Compared with traditional sentence representation methods, SBERT can capture more fine-grained semantic information, making the similarity calculation between two sentences at the semantic level more accurate;

[0497] Multi-level similarity:

[0498] In order to match semantics more precisely, the similarity of knowledge points at multiple levels can be calculated; the details are as follows:

[0499] First, the similarity based on the word level is calculated, and then the similarity at the sentence level or paragraph level is calculated. The final similarity can be the weighted average of the similarities at each level:

[0500] total_similarity(x1,x2)

[0501] =α·word_similarity(x1,x2)+β·sentence_similarity(x1,x2)+γ

[0502] ·document_similarity(x1,x2)

[0503] Among them, α, β, and γ are weights of similarity at different levels, which are adjusted according to specific application scenarios. In the processing process, if multiple representations are used for the same entity, the large model is fine-tuned to determine whether they are the same knowledge point, and then mapping and recognition are performed to obtain the final knowledge points and relationships.

[0504] Mapped to the knowledge point entity standard set through similarity calculation,

[0505] It also includes mapping methods, and context-aware semantic matching is introduced in the matching process, by capturing the context of knowledge points in the text; context analysis combines the context information of surrounding words and sentences to enhance the ability to distinguish synonyms or related knowledge points. For example, if "quantum mechanics" is mentioned in the context of physics, it can be inferred that the statement is consistent with the relevant knowledge points in this field.

[0506] It should be noted that dynamic weighted cosine similarity is suitable for scenarios where the importance of different dimensions needs to be considered. By assigning weights w_i to different dimensions, the calculation results can more flexibly reflect important information. In particular, when two knowledge points are consistent in a certain category, the weight of the relevant dimension of the category can be increased;

[0507] Semantic Hierarchical Similarity (SBERT): Suitable for scenarios that require more fine-grained semantic analysis. Compared with traditional sentence representation methods, SBERT can capture more fine-grained semantic information, making the similarity calculation between sentences more accurate. Multi-level Similarity: Suitable for scenarios that require more comprehensive semantic matching; this method calculates similarities at multiple levels simultaneously.

[0508] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the specific construction process of the knowledge point extraction fine-tuning model in S12) is as follows:

[0509] (1) Data preparation, that is, preparing enough domain-related text data, including annotations of knowledge points. For different expressions of knowledge points, it is necessary to prepare a standardized set of knowledge points to facilitate subsequent mapping and verification. The data source can be manually annotated text or use existing knowledge bases, such as Wiki or expert-annotated domain data sets;

[0510] (2) Model selection

[0511] Choose a pre-trained large language model (such as BERT, GPT, etc.) as the pre-training model. This model has been pre-trained with massive general data and has good language understanding capabilities. For the task of knowledge point extraction, you can choose a model based on the Transformer architecture. These models perform well in natural language processing tasks.

[0512] (3) Label and clean the text

[0513] Preprocess the data, including text cleaning (removing noise and irrelevant content) and sentence segmentation. Then, use annotation tools or write rules to mark entities and relationships in the text. This step can use tools such as SpaCy or NLTK for automatic sentence segmentation and part-of-speech tagging.

[0514] (4) Fine-tune the pre-trained model to obtain a fine-tuned knowledge point extraction fine-tuning model.

[0515] For the knowledge point extraction task, the pre-trained model such as BERT or GPT is fine-tuned to obtain the fine-tuned model. The training process of the pre-trained model is completed through the following stages:

[0516] Entity recognition: Use pre-trained models to automatically identify entities in text, such as concepts, facts, and principles;

[0517] Relationship extraction: extract the relationships between text entities automatically identified by the pre-trained model, such as inclusion, subordination, and predecessor;

[0518] Disambiguation processing: For different descriptions of the same knowledge point, vectorization technology is used to calculate similarity and map similar expressions to the same entity;

[0519] (5) Model verification and optimization

[0520] The fine-tuned model is tested on the validation set to evaluate its accuracy in knowledge point recognition and relationship extraction tasks. The similarity algorithm is used to verify the accuracy and consistency of the extracted entities and relationships. If the accuracy and consistency meet the requirements, it means that the model can meet the needs.

[0521] The specific process of fine-tuning the pre-trained model to obtain the fine-tuned model is as follows:

[0522] (401) First, determine the task objectives: that is, determine the task objectives of entity recognition, relationship extraction, and disambiguation processing respectively;

[0523] (402) Prepare the data set: label the entities and relations (BI0 labeling, entity pairs and relation categories);

[0524] (403) Model selection: that is, according to actual needs, select a suitable model from “BERT-type models (suitable for classification and sequence labeling), GPT-type models (suitable for generation tasks), and graph neural networks (suitable for complex relationship modeling)”;

[0525] (404) Design fine-tuning tasks: that is, setting fine-tuning tasks for the selected model, specifically, performing sequence labeling (BI0) for entity recognition and fine-tuning using a linear layer; Relationship extraction: extracting relationships from entities and contexts, extracting and outputting relationships through a classifier; for disambiguation processing, performing entity vectorization description, and training a similarity classifier or clustering model for fine-tuning;

[0526] (405): Train and optimize the fine-tuned model using the AdamW optimizer and set the learning rate to 1e -5 To 5e -5 ;

[0527] (406): Evaluate the fine-tuned training model, specifically: evaluate the entity recognition extraction through the F1 score performance indicator; evaluate the accuracy of relationship extraction by judging the text classification accuracy and F1 score performance indicator during the relationship extraction process; evaluate the effect of disambiguation processing through Top-k accuracy and similarity evaluation.

[0528] The personalized learning recommendation method based on the personalized knowledge graph described in this embodiment, the specific method of model verification and optimization is as follows:

[0529] Adaptive threshold adjustment: During the similarity calculation process, the similarity threshold is dynamically adjusted by continuously learning the accuracy of historical data;

[0530] The specific process is as follows:

[0531] Data initialization: Collect the initial historical data of similarity calculation, including the results of each similarity calculation and the correctness of actual verification;

[0532] Initial threshold setting: Set the similarity threshold based on common empirical values, such as the default threshold of cosine similarity is 0.8;

[0533] Similarity calculation: For each knowledge point to be processed, a preliminary similarity score is obtained based on cosine similarity, SBERT or other similarity algorithms;

[0534] Feedback collection: Based on the results of previous processing, the system collects feedback data from users or experts and marks whether each resolution is correct;

[0535] Adaptive learning: The system uses machine learning algorithms to dynamically adjust the similarity thresholds based on accumulated feedback data and errors in disambiguation results. The adjustment range is usually between 0.05 and 0.10. The thresholds can be adaptively changed based on different fields, text length, and semantic complexity.

[0536] Apply dynamic threshold: In a new round of similarity calculation, a dynamically adjusted threshold is used to determine whether two knowledge points are considered the same.

[0537] For example, when the system disambiguates “quantum physics” and “quantum mechanics,” based on similar data previously processed, the system can adaptively adjust the threshold to ensure that these statements are considered the same knowledge point.

[0538] In this embodiment, the extracted knowledge points and relationships are stored in the graph database. The specific process of forming the knowledge graph is as follows: 1. Insert the knowledge points and relationships into the graph database

[0539] Graph databases (such as Neo4j, JanusGraph, ArangoDB, etc.) store data in the form of nodes and edges. Nodes represent entities (knowledge points) and edges represent the relationship between entities. The process of building a knowledge graph mainly involves the following steps:

[0540] (1) Design of graph data model

[0541] Entity Nodes: Each extracted knowledge point is inserted into the graph database as a node. Nodes usually have multiple attributes, such as entity name, type, unique ID, etc.

[0542] Edges: The relationship between each two entities is represented as an edge. The attributes of the edge include the relationship type (such as "contains", "related", "predecessor", etc.), and the direction of the relationship.

[0543] (2) Insertion process using Neo4j as an example

[0544] Connect to the graph database: Use the Neo4j Python client (such as `neo4j-driver`) to connect to the Neo4j database:

[0545] ```Python

[0546] from neo4j import GraphDatabase

[0547] uri="bolt: / / localhost:7687"

[0548] driver=GraphDatabase.driver(uri,auth=("username","password"))

[0549] Create nodes and relationships:

[0550] Create entity nodes: Each extracted knowledge point is inserted into the graph database as a node.

[0551] ```cypher

[0552] CREATE(n:Entity{name:'Advanced Mathematics',type:'Course'})

[0553] ```

[0554] You can use the Python API to insert nodes in batches:

[0555]

[0556] 2. Create relationship edges: Create relationships between entity nodes.

[0557] ```cypher

[0558] MATCH(a:Entity{name:'Advanced Mathematics'}),(b:Entity{name:'Calculus'})

[0559] CREATE (a)-[:contains]->(b)

[0560] ```

[0561] In Python, you can create relationships in batches:

[0562]

[0563] (3) Insertion process using JanusGraph

[0564] For distributed graph databases such as JanusGraph, Gremlin is often used as the query language, similar to Neo4j's Cypher. To batch insert nodes and relationships through the Gremlin interface:

[0565] ```gremlin

[0566] g.addV('Entity').property('name','Advanced Mathematics').property('type','Course')

[0567] gV().has('name','Advanced Mathematics').addE('includes').to(gV().has('name','Calculus')).

[0568] The personalized learning recommendation method based on the personalized knowledge graph described in this embodiment can be combined with symbolic reasoning in the process of disambiguation. Symbolic reasoning can further correct the disambiguation results through known rules and logic. The specific process is as follows:

[0569] (1) Knowledge point extraction: Extract potential similar knowledge points in the original data from the literature and obtain preliminary semantic similarity through the deep learning model;

[0570] (2): Semantic disambiguation: Use deep learning methods to perform preliminary similarity calculations. Those with similarities lower than the target value will be deleted to disambiguate some basic knowledge points;

[0571] (3): Application of symbolic reasoning rules: For knowledge points that cannot be disambiguated or have insufficient similarity, symbolic reasoning is combined with defined rule sets or logical reasoning to further correct the results based on known knowledge in the field;

[0572] (4): Reasoning optimization: The result of symbolic reasoning in the previous step is fed back to the deep learning model to further adjust the weights or thresholds of the deep learning model to reduce errors;

[0573] (5): Output the final result: Combine the final results of symbolic reasoning and deep learning disambiguation to determine whether two knowledge points are the same.

[0574] The known rules include logical rules based on logical expressions (mutual exclusivity, transitivity), ontology rules for defining the hierarchy, attributes and relationships of concepts, data consistency rules for verifying the consistency of data, semantic rules for reasoning through context or defined semantic relationships, mathematical or physical laws and priority rules.

[0575] The deep learning model (such as BERT or SBERT) obtains preliminary semantic similarity and generates fixed word vectors, which can be used to calculate semantic similarity as follows:

[0576] Word2Vec: Generates word vectors based on a context window and uses cosine similarity of the vectors to compare semantic similarity;

[0577] GloVe: Generates word embeddings based on global statistical information, suitable for semantic similarity calculation.

[0578] FastText: Extends Word2Vec to support subword representation, which can better handle word form changes. The rule set defined above, where the rule set is the core of symbolic reasoning, in specific implementations, its definition usually needs to be combined with domain knowledge. The following are some general principles and examples for defining rule sets:

[0579] The ruleset is structured as follows:

[0580] (1) Static rules, which are deterministic rules based on known facts and domain constraints. For example:

[0581] Classification rules: If a knowledge point belongs to a certain category, it may have attributes common to that category.

[0582] Association rule: Two knowledge points are equivalent or related under certain conditions.

[0583] (2) Dynamic rules, which involve data-driven patterns and can usually be combined with statistical analysis or frequent itemset discovery:

[0584] Conditional probability rule: if knowledge points A and B appear at the same time with high frequency, it can be inferred that they may be the same or related.

[0585] Context-sensitive rules, that is, the meaning of knowledge points changes with the context.

[0586] The specific rule set examples are as follows:

[0587] (1) Field: Mathematical knowledge points

[0588] Equivalence rule: If the description of a knowledge point contains "equivalent to" or "also known as", they may be the same concept.

[0589] TextContains(x,"aka")→Equivalent(x,y)

[0590] If two formulas have the same transformed form, they may be equivalent, such as

[0591] FormulaSimilar(A,B)∧Transform(A,B)→Equivalent(A,B)

[0592] Derivation rule: If a formula can be transformed into another formula through simple algebraic transformation, they may be the same knowledge point;

[0593] AlgebraicTransform(Equation1,Equation2)→SamePoint(Equation1,Equation2)

[0594] (2) Field: Biological terminology

[0595] Synonym rule: If two terms differ in spelling only in the suffix (such as British and American spellings), they are probably synonyms.

[0596] Hyponym and hyponym rules: If one term is a hyponym of another term and certain conditions are met, they may be partially related:

[0597] (3) Cross-domain rules

[0598] Temporal correlation: If knowledge points A and B are frequently cited in the same time period, they may be related.

[0599] 4. The rule set is constructed as follows:

[0600] Based on expert knowledge: Domain experts summarize the relationships between concepts and manually define rules.

[0601] Based on data mining: extracting potential pattern generation rules from the corpus.

[0602] Use tools such as the Apriori algorithm to generate frequent itemsets.

[0603] Using knowledge graphs to mine term relationships.

[0604] Automatic generation and optimization: Combine deep learning and reinforcement learning to generate new rules and optimize them through verification feedback.

[0605] 5. Rule verification and optimization

[0606] Verification: Test the defined rules and verify the correctness of the reasoning using known data.

[0607] Optimization: Adjust rule weights based on feedback from actual applications, or design meta-rules to prioritize conflicting rules.

[0608] By combining these rule sets, symbolic reasoning can effectively correct ambiguous semantics during the disambiguation process and further improve the accuracy and interpretability of the final results.

[0609] The specific logic of reasoning, specifically, the reasoning logic can be divided into the following key steps:

[0610] 1. Formalization of input data. In order for symbolic reasoning to handle problems, the input data (knowledge points, semantic information) must first be formalized into a logical language. Common forms include:

[0611] Predicate logic: Describe the relationship between knowledge points through logical predicates.

[0612] Term("Optical diffraction") → Related("Light wave interference")

[0613] Graph structure: Knowledge points can be represented by a knowledge graph, where nodes are knowledge points and edges are relationships.

[0614] 2. Application of knowledge point matching rules, using predefined rule sets to perform pattern matching on knowledge points and their contextual information to extract potential logical relationships.

[0615] The rule matching method is as follows:

[0616] 2.1 Static matching: directly matching the conditions in the rules and comparing them with known knowledge points or context;

[0617] For example, the rule:

[0618] Synonym(a,b)→Equivalent(a,b)

[0619] If Synonym("Optical Diffraction","Diffraction") holds, then it follows that:

[0620] Equivalent("Optical Diffraction","Diffraction")

[0621] 2.2 Dynamic conditional deduction: for uncertain knowledge points, conditional judgment is performed based on the logical rules of the context;

[0622] Context("optics")∧Related("light wave","interference")→Related("optics","diffraction")

[0623] 2.3 Fuzzy rule processing: For knowledge points with insufficient semantic similarity, fuzzy logic is used to define the credibility threshold:

[0624] For example: Rule:

[0625] Similarity(a,b)>0.7→PossibleEquivalent(a,b)

[0626] If Similarity("Optical Diffraction", "Diffraction") = 0.75, then it is inferred that they are likely equivalent.

[0627] 3. Application of the inference engine. Using the inference engine to execute the rule set and combine the logic to deduce the relationship between knowledge points, the recommendation engine can achieve the following reasoning:

[0628] 3.1 Forward reasoning, that is, starting from known conditions, triggering rules in sequence to derive new knowledge point relationships;

[0629] Known:

[0630] Synonym("Optical Diffraction","Diffraction")

[0631] rule:

[0632] Equivalent(a,b)→Related(a,b)

[0633] Derivation:

[0634] Related("Optical diffraction","Diffraction").

[0635] 3.2 Backward reasoning: starting from the goal, looking backwards for the conditions that meet the goal;

[0636] 3.3 Fuzzy logic reasoning, which introduces confidence between conditions and conclusions.

[0637] 4. Conflict resolution and priority processing. In the symbolic reasoning process, multiple rule conflicts may occur, and the priority needs to be clarified:

[0638] 4.1. Priority based on weight: Define weights for rules, and rules with higher weights will be executed first.

[0639] 4.2. Meta-rule based conflict resolution: Conflict resolution strategies are defined through meta-rules.

[0640] During the reasoning process, the rule Synonym(a,b) is used to match the text, check whether the context supports it, and finally merge the reasoning results to draw a conclusion.

[0641] If RuleA∧RuleB conflict→Prefer(RuleA)

[0642] 5. Example Process

[0643] Question: Are "optical diffraction" and "interference" equivalent knowledge points?

[0644] Known:

[0645] rule:

[0646] Synonym(a,b)→Equivalent(a,b)

[0647] Context("optics")∧Related(a,b)→Equivalent(a,b)

[0648] data:

[0649] Synonym("Optical Diffraction","Diffraction")

[0650] Context("Optics")

[0651] Related("light wave","interference")

[0652] Reasoning process:

[0653] 1. Use the rule Synonym(a,b) to match:

[0654] Synonym("Optical Diffraction","Diffraction")→Equivalent("Optical Diffraction","Diffraction")

[0655] 2. Check if the context supports:

[0656] Context("optics")∧Related("light wave","interference")→Equivalent("optical diffraction","interference")

[0657] 3. Combine the inference results:

[0658] Equivalent("Optical Diffraction","Interference")∧Equivalent("Optical Diffraction","Diffraction")→True

[0659] Finally, it is concluded that "optical diffraction" is equivalent to "interference".

[0660] 5. Combined with the feedback of the deep learning model, that is, after the symbolic reasoning is completed, the results can be used as feedback to further adjust the similarity calculation or knowledge representation of the deep learning model. For example:

[0661] Update the semantic similarity model: If the symbolic reasoning result is True, adjust the similarity weight to make it closer to the corrected standard.

[0662] Retrain the semantic embedding model: add new disambiguation results to the corpus and update word vectors or relationship graphs.

[0663] In this way, symbolic reasoning and deep learning can optimize each other and improve the accuracy and robustness of the overall disambiguation.

[0664] The personalized learning recommendation method based on the personalized knowledge graph described in this embodiment, in the process of disambiguation, the disambiguation mechanism based on domain knowledge can be combined with expert knowledge or rules in a specific field to perform disambiguation in addition to general semantic matching and similarity calculation in a specific field;

[0665] Implementation process:

[0666] A. Knowledge point semantic matching: Use deep learning models to perform general semantic matching on the extracted knowledge points;

[0667] In the disambiguation task, the general semantic matching of Part A mainly relies on the deep learning model to compare knowledge points at the semantic level. This process involves specific semantic embedding representation, similarity calculation and matching rules. The following are the specific implementation methods and corresponding matching rules:

[0668] B. Introduction of domain expert knowledge: Introduce expert rules, definitions, vocabularies or knowledge graphs in specific fields to further verify and correct the semantic matching results through these domain knowledge;

[0669] C. Domain rule disambiguation: more finely classify or exclude matching results based on a domain-specific rule set;

[0670] D. Output disambiguation results: Combine the domain knowledge disambiguation results with the general semantic matching results, and output the final disambiguated knowledge point matching.

[0671] In knowledge point disambiguation, the combination method of part D is not a simple direct superposition, but through weight distribution, multi-layer decision logic or collaborative optimization, the results of domain knowledge disambiguation and general semantic matching are integrated to improve the accuracy and robustness of the final result. The following is an analysis of the specific combination method, as well as the purpose and function of the combination:

[0672] The core process of knowledge point semantic matching is as follows:

[0673] (1) The semantic representation of knowledge points transforms knowledge points into comparable semantic vectors through a deep learning model.

[0674] There are the following ways to embed word vectors:

[0675] Use pre-trained word vector models (such as Word2Vec, GloVe) to convert knowledge points into vectors.

[0676] Use language models (such as BERT, RoBERTa) to generate contextual vector representations.

[0677] If the knowledge point contains multimodal information (such as text, formula, image), a joint embedding representation can be generated through a multimodal model (such as CLIP).

[0678] (2) Semantic similarity calculation, that is, calculating the similarity between the vector representations of two knowledge points.

[0679] (3) Setting the matching threshold: Based on the similarity score, define a matching threshold:

[0680] Exact match: if similarity > 0.9

[0681] Possible match: 0.7≤similarity≤0.9

[0682] No match: Similarity < 0.7

[0683] It should be noted that the setting of the threshold can be adjusted according to the actual field.

[0684] 3. Tools and models for implementing general semantic matching can use the following models and tools:

[0685] (1) Pre-trained language model: Use a model pre-trained on large-scale corpus to generate semantic embeddings, for example:

[0686] BERT, RoBERTa: suitable for text semantic matching; SciBERT: optimized for scientific literature; Sentence-BERT: generates sentence-level semantic vectors, suitable for calculating similarity.

[0687] (2) Knowledge base and dictionary: use general knowledge base (such as WordNet, ConceptNet) to expand semantic information and use domain-related dictionaries to supplement terminology.

[0688] (3) Semantic similarity calculation library, including Python tools, scikit-learn: provides cosine similarity and Euclidean distance calculation; sentence-transformers directly calculates embedding vector similarity.

[0689] Example: The complete process of general semantic matching

[0690] Question: Are “optical diffraction” and “wave interference” the same knowledge point?

[0691] 1 Generate embeddings using BERT:

[0692] "Optical Diffraction" → [0.32, 0.45, 0.76]

[0693] "Interference of waves" → [0.33, 0.44, 0.78]

[0694] 2. Calculate cosine similarity: CosineSimilarity = 0.98

[0695] 3. Matching rule verification: Verify by checking spelling variants:

[0696] If "optical diffraction", "interference of waves"

[0697] EditDistance("Optical Diffraction","Wave Interference")>3→NoMatch

[0698] EditDistance(a,b): represents the edit distance between strings a and b. Edit distance (LevenshteinDistance) is a common metric for measuring the similarity between two strings. It represents the minimum number of edit operations required to transform one string into another. These edit operations usually include:

[0699] Insert a character;

[0700] Delete a character;

[0701] Replaces a character.

[0702] >3: Indicates that the edit distance is greater than 3; that is, the number of edit operations between strings a and b is greater than 3.

[0703] NoMatch: If the condition is met, that is, the edit distance is greater than 3, a and b are considered to be unmatched (NoMatch).

[0704] Check context overlap: If the context overlap is greater than or equal to 0.7, then it is a possible match;

[0705] 4. Matching threshold judgment: judge as "possible match" based on similarity and rule results;

[0706] In the process of combining the domain knowledge disambiguation results with the general semantic matching results, weighted fusion is used to combine them.

[0707] Different weights are assigned to the domain knowledge disambiguation results and the general semantic matching results, and the two are weighted according to the importance of the actual task;

[0708] The details are as follows:

[0709] General semantic similarity score: S1 = 0.8

[0710] Domain rule disambiguation score: S2 = 0.9

[0711] Weighted formula: S final =w1·S1+w2·S2

[0712] Among them, w1 and w2 are weight coefficients, which can be adjusted through experiments;

[0713] Assume the weights are w1=0.4, w2=0.6

[0714] S final =0.4·0.8+0.6·0.9=0.86

[0715] Just determine whether it matches based on the set threshold;

[0716] The combination of weighted fusion can effectively overcome the limitations of a single method and make the final disambiguation result more comprehensive and accurate. The details are as follows:

[0717] (1) Make up for the shortcomings of a single method

[0718] Disadvantages of general semantic matching: General semantic matching is weak in processing domain-specific terms or rare knowledge points, which may lead to misjudgment or omission. For example, a general model may mistake "optical diffraction" and "wave interference" as unrelated.

[0719] Disadvantages of domain rule disambiguation: Domain rules are highly dependent on known expert knowledge and may be powerless for unknown or uncovered situations;

[0720] By combining the broad coverage of general semantic matching and the high precision of domain rules, the disambiguation results are made more reliable.

[0721] (2) Handling special situations,

[0722] Some matching pairs may have low scores in general semantic matching, but domain knowledge shows that the two are closely related;

[0723] For example: General matching result: Similarity = 0.6

[0724] Domain rules: rules indicate that there is a causal relationship between the two;

[0725] After combination: determined as a match;

[0726] (3) Improve confidence and explainability. The introduction of domain rules provides a clear logical basis, making the matching results more explainable, while the general semantic matching results provide quantitative support.

[0727] (4) Optimize model performance,

[0728] Combined with the feedback from domain rules, the parameters of the semantic model can be dynamically adjusted to gradually adapt it to domain requirements.

[0729] Although the domain rule disambiguation in Part C has classified and excluded the common semantic matching results, the following reasons make the combination more necessary:

[0730] (1) Improve matching coverage,

[0731] The coverage of domain rules is limited and cannot cover all knowledge points. Combined with general semantic matching, it can handle parts that are not defined or covered by the rules;

[0732] (2) Handling low confidence matches

[0733] In semantic matching and rule disambiguation, some knowledge points may be in a boundary state at the same time:

[0734] For example, the semantic similarity score is close to the threshold or the domain rules are contradictory or incomplete;

[0735] By combining the results from both, these edge cases can be treated more comprehensively;

[0736] (3) Verification and calibration results, combined with the results of general semantic matching, can verify the judgment of domain rules, thereby reducing errors;

[0737] (4) Improve scalability. Combined with semantic matching, the system can still output valid results even when domain rules are insufficiently covered, making the model more general and flexible.

[0738] The following are some examples:

[0739] Determine whether "interference of waves" and "optical diffraction" are the same

[0740] 1. General semantic matching:

[0741] Semantic similarity score: 0.6 (below the matching threshold);

[0742] Initial judgment: mismatch;

[0743] 2. Domain rule disambiguation:

[0744] Rule 1: If the knowledge point contains "wave" and "diffraction", the correlation is high;

[0745] Rule application results: relevance score 0.9;

[0746] 3. Weighted fusion:

[0747] Weight distribution: w1=0.3, w2=0.7;

[0748] Final score:

[0749] S final =0.3·0.6+0.7·0.9=0.78

[0750] From the above results, it can be seen that the match threshold is exceeded and it is determined to be a match.

[0751] 4. Output result: Knowledge point matching: "Wave interference" = "Optical diffraction"

[0752] Therefore, combining general semantic matching and domain rule disambiguation not only improves coverage and confidence, but also provides the model with the ability to dynamically adjust and expand. This combined strategy can output more accurate and reliable disambiguation results when a single method is insufficient.

[0753] For example, although "myocardial infarction" and "heart attack" in the medical field are similar in general semantics, they have more subtle differences in some literature, so they need to be strictly distinguished in combination with domain rules.

[0754] The personalized learning recommendation method based on the personalized knowledge graph described in this embodiment uses the following multiple means to ensure the accuracy of the inserted knowledge points and relationships during the knowledge graph construction process, as follows:

[0755] (1) Multi-level verification

[0756] Rule-based verification: Different verification rules are defined for different fields to detect knowledge points and relationships. If data that does not meet the verification rules is matched, potential errors will be marked, and the knowledge points and relationships marked as not wrong will be deleted;

[0757] Context consistency verification: By analyzing the context in which the knowledge point is located, ensure that the extracted entities and relationships are consistent with the logic in the text;

[0758] Domain expert review: For knowledge in a specific field, a knowledge base annotated by experts can be introduced for comparison to ensure that the extracted knowledge points are consistent with the information in the existing knowledge base;

[0759] (2) Semantic similarity matching

[0760] Use semantic similarity calculation to verify the extracted knowledge points. For example, use language models such as BERT and SBERT to vectorize the extracted knowledge points and standard knowledge points, and then calculate the similarity to determine whether the knowledge points are accurately extracted. If the similarity is lower than the set threshold, it is marked as a potential error and deleted.

[0761] (3) Relational logic verification

[0762] Relationship type restrictions: ensure that the relationship type conforms to the logic within the domain, and perform logical verification on the relationship type through the domain rule library; for example, in the education field, the relationship between "course" and "chapter" should be "included" instead of "preceded";

[0763] Loop detection: Use graph algorithms to detect whether there are unreasonable loops in the knowledge graph to avoid unreasonable knowledge graph structure caused by incorrect relationships.

[0764] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the means and process of verifying the accuracy of extracting knowledge points and relationships in the data source in S2) are as follows:

[0765] Before storing the knowledge points and relationships in the graph database, use the following steps to verify the accuracy of the extracted knowledge points and relationships:

[0766] (1): Automated test dataset

[0767] Build a standard annotated data set and input it into the AI ​​model as a benchmark. Then compare the knowledge points extracted by the knowledge point extraction and fine-tuning model with the standard answers. If the extracted knowledge points and relationships are consistent with those in the annotated data set, the verification is passed; otherwise, the knowledge point extraction and fine-tuning model needs to be further debugged or the algorithm needs to be adjusted.

[0768] Accuracy: number of correctly extracted knowledge points / total number of extracted knowledge points;

[0769] Recall rate: the number of correctly extracted knowledge points / the number of knowledge points in the standard set;

[0770] (2): Manual review and expert feedback

[0771] Knowledge points in specific fields are manually reviewed by experts to ensure that the extracted content meets industry standards; for example, in the medical field, doctors can review the extracted terms and relationships;

[0772] (3): Context verification

[0773] After knowledge points are extracted, secondary verification is performed in combination with their context to ensure that the extracted knowledge points are consistent with their meanings in the context. For example, "Java" as a knowledge point of programming language may also represent other concepts in different contexts. This error can be reduced through context verification.

[0774] (4): Model adaptive optimization

[0775] By continuously collecting error cases extracted by the knowledge point extraction fine-tuning model, analyzing and providing feedback, adding this feedback to the training data, and continuously optimizing the performance of the model, this circular feedback mechanism can effectively improve the accuracy of extraction.

[0776] In the process of building the knowledge graph, complex knowledge graphs can be built by inserting the extracted knowledge points and relationships into the graph database using tools such as Neo4j and JanusGraph. At the same time, the accuracy and logic of the inserted data can be ensured through context consistency checks, rule-based validation, semantic similarity matching, and loop detection. In order to further verify the accuracy of knowledge points and relationships, various methods such as automated data set testing, expert review, and context verification can be used to ensure the high quality and accuracy of the knowledge graph.

[0777] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the learner portrait model construction process in S2) is as follows:

[0778] S21): Data collection, collecting various data of learners, including learning behavior data, learning resource interaction data, and test scores, and using association rule algorithms and data mining technology to mine users' learning process data;

[0779] S22): Preprocess the collected data, including cleaning outliers and missing values ​​to ensure data quality. Through attribute matching and feature extraction, combined with learners’ personal attributes and learning styles, various learning portrait modules are integrated using statistical methods to obtain learners’ learning ability, cognitive level, and learning goal portraits.

[0780] S23): Extract the learner's personalized learning features from the three dimensions of learning ability, cognitive level, and learning goals. Use AprioriAll to quantify the feature labels of learning ability, cognitive level, and learning goals to form the learner's learning feature matrix. Each row represents a learner, and each column corresponds to a feature dimension.

[0781] S24): Then, based on the learning feature matrix, a machine learning algorithm is used to group or classify the learners to construct a learner portrait model.

[0782] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the reinforcement learning DQN network in the personalized learning recommendation engine makes decisions through the interaction between the agent and the environment, and the decision needs to bring as many rewards as possible; the agent here is the personalized learning recommendation engine, the environment is the user's state and knowledge graph, and the reward is the income of the learning effect;

[0783] The specific working process is as follows:

[0784] S61): First, the data is processed to obtain user portrait features.

[0785] The user portrait feature is composed of a learning ability vector c, a cognitive level vector l, and a learning goal vector g. The learning ability vector c, the cognitive level vector l, and the learning goal vector g are calculated respectively, and then the user comprehensive feature vector is processed. The specific process is as follows:

[0786] The following normalization formulas are involved:

[0787]

[0788] Where X min is the minimum value, X max is the maximum value, the purpose is to map the range of data to [0,1];

[0789] The learning ability includes memory, logical reasoning ability, learning speed, and learning concentration:

[0790] The memory dimension is calculated by using the memory-type question score and the total score to calculate the score rate M as an indicator;

[0791]

[0792] Normalization:

[0793]

[0794] The logical reasoning ability dimension uses the score rate L of reasoning questions as an indicator;

[0795]

[0796] Normalization:

[0797]

[0798] Learning speed dimension: the ability of users to effectively complete learning tasks within a unit of time, reflecting the speed at which students accept and master new knowledge; including average learning time, learning concentration

[0799] Average study time:

[0800]

[0801] The learning speed is inversely proportional to the average learning time L kp

[0802]

[0803] Normalization:

[0804]

[0805] Study concentration:

[0806] The learning concentration refers to the ability of students to maintain concentration and avoid distraction during the learning process, reflecting the degree of learning input;

[0807] The main indicators of learning concentration are as follows:

[0808] Continuous learning time period, number of learning interruptions, and other abnormal operations;

[0809] Learning concentration index

[0810]

[0811] Average study time

[0812]

[0813] Learning concentration final score formula

[0814]

[0815] A min and A max are the minimum and maximum values ​​of the average continuous learning time of users.

[0816] D min and D max are the minimum and maximum values ​​of distraction indicators of all users, respectively;

[0817] Among them, ω1 and ω2 are weights, with a value of 0.5;

[0818] The learning ability matrix is ​​represented by:

[0819] C=[M norm L norm L kp_norm A score ]

[0820] The cognitive levels in the cognitive level vector l are as follows:

[0821]

[0822] Among them, A total is the total score of all tests in the subject for the user; M total is the sum of the full scores of all tests in the subject; l is the ratio of the total score of all tests in the subject to the sum of the full scores of all tests in the subject;

[0823] Normalization:

[0824]

[0825] The learning goal in the learning goal vector g is:

[0826] According to the user's target school and target major, knowledge points are selected from the knowledge graph, and then the proficiency of each knowledge point is calculated based on the target score.

[0827] Assume that the knowledge points obtained by screening according to the target colleges and majors are as follows:

[0828] k=[k1 k2 ... k n ]

[0829] The weight of each knowledge point is

[0830] ω(k i), where i is 1....n;

[0831] Normalize the weight of each knowledge point

[0832]

[0833] According to the normalized weight, the target total score S goal Assign to each knowledge point and adjust the base value to ensure that the weight is greater than 1

[0834] S target (k i )=ω′(k i )×(S goal )+1

[0835] The matrix of the user learning target vector g is as follows:

[0836] g=[S target (k1) S target (k2) ... S target (ki)] Combine the above features to form the user's comprehensive feature vector

[0837] F student =[c, l, g]

[0838] Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector;

[0839] Using the comprehensive feature vector F student As clustering features, students are divided into K groups and each student is assigned a cluster label G j (j=1, 2, ..., K), represents the group to which the student belongs, and the cluster label G j As features, incorporated into the state representation of subsequent models;

[0840] S62): Knowledge graph embedding, that is, collecting all knowledge points and constructing a knowledge point set K = k1, k2, ..., k N ;

[0841] The Node2Vec algorithm is used to embed the knowledge points in the knowledge graph into the low-dimensional vector space to obtain the embedding vector e of the i-th knowledge point i ;

[0842] Then add the user's comprehensive feature vector, the embedding vector of the knowledge point, and the student's clustering label to obtain the corresponding user state vector, as follows:

[0843] The user state vector S with the user's comprehensive feature vector added is expressed as follows:

[0844] s=[s1,s2,...,sN ,c,l,g]

[0845] Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector;

[0846] Add the student's cluster label G j and the embedding vector e of the knowledge point i ;

[0847] The knowledge point vector is calculated as follows:

[0848]

[0849] in,

[0850] is the mastery level: that is, the students’ i The degree of mastery, range [0,1];

[0851] is the learning frequency: that is, the number of times a student learns knowledge point k i The number of times;

[0852] is the test performance: that is, the student's performance on knowledge point k i Average test scores on

[0853] is the most recent learning time interval: that is, since the last learning of knowledge point k i the time interval since;

[0854] e i Embedding vector of knowledge points;

[0855] f i is a frequent pattern feature;

[0856] Knowledge point k i The support is:

[0857]

[0858] Add the student's cluster label G j The user state vector S after is expressed as follows

[0859] s=[s1,s2,...,s N ,c,l,g,G j ]

[0860] If the user does not fully grasp the knowledge point, the optional action space is as follows:

[0861] A′(s)={k i ∈A(s)|g i >0}

[0862] A(s) is the set of knowledge points that satisfy the prerequisite relationship but are not fully mastered, g i >0 indicates knowledge point k i It is the student’s learning goal;

[0863] Reward learners for completing their learning goals. The specific reward function is as follows

[0864] r t =w(k i )×Δs i ×f(c,l)×G(k i )×PM(k i )×AE(k i )×LF(k i )×FC(k i )

[0865] Among them, w(k i ) is the knowledge point k i Importance weight

[0866] Δs i For students to understand the knowledge point k i Improved mastery

[0867] f(c, l) is the adjustment function of students’ learning ability and cognitive level, as follows:

[0868] f(c,l)=δ1·c+δ2·l, where δ1 and δ2 are weight parameters, with values ​​of 0.7 and 0.3 respectively;

[0869] G(k i ) is the learning objective weighting function

[0870]

[0871] Among them, g i is the i-th element of the student's learning goal vector, indicating the student's understanding of knowledge point k i The degree of learning needs; γ is a constant greater than 1, which is the bonus coefficient for the learning target knowledge points;

[0872] When i >0(Knowledge point k i is the student's learning goal), G(k i ) = γ, the reward is amplified by γ times (usually 1.2-2.0), and the expected reward for selecting this knowledge point is submitted. The personalized learning recommendation engine is more inclined to recommend this knowledge point;

[0873] When i =0, G(k i)=1, the reward remains unchanged, and the personalized learning recommendation engine will not be particularly biased towards recommending this knowledge point;

[0874] PM(k i ) is a frequent pattern weighting function, which adjusts the reward according to the importance of the knowledge point in the frequent item set, reflecting the importance of the knowledge point in the learning group, as follows:

[0875] PM(k i )=1+λ PM ×Support(k i )

[0876] Among them, λ PM To adjust the parameter, the degree of influence on the reward, usually a small positive number such as 0.1 or 0.2;

[0877] Support(k i ) is the knowledge point k i The support degree of k i Frequency of occurrence in frequent itemsets;

[0878] By considering the frequency of occurrence of knowledge points in frequent learning patterns, the personalized learning recommendation engine is guided to recommend knowledge points that are commonly learned in the group;

[0879] In the association rule algorithm, AR(k i )The association rule weighting function is:

[0880] AR(k i )=1+λ AR ×Conf(k i )

[0881] Among them, λ AR To adjust the parameters, control the influence of association rules on rewards, usually a positive number; Conf(k i )Knowledge point k i The highest confidence in the association rule, denoted by k i The strength of the connection with the knowledge points already learned;

[0882] Using the knowledge learned: By considering k i The association with the knowledge points that students have already learned guides the model to recommend knowledge points that are closely related to what has been learned, thus promoting the coherent learning of knowledge;

[0883] LF(k i )The learning frequency adjustment function is

[0884]

[0885] Among them, λ LF Take w(ki ), the more important the knowledge point is, the smaller the reduced reward is;

[0886] As the learning frequency increases, the function value will gradually decrease, reducing the contribution to the reward and preventing students from excessively repeating familiar knowledge points;

[0887] FC(k i ) Forgetting curve adjustment function

[0888]

[0889] Among them, λ FC Take the proportional function of the knowledge point weight, with a value range of 0.5-1; when Increase, and the reward decreases, prompting students to review this knowledge point;

[0890] S63): Reinforce the learning DQN network to update the network parameters, as follows:

[0891] Input layer: user state vector s

[0892] Output layer: the Q value of each optional action (the Q value is the mathematical expectation of the sum of all future state returns);

[0893] The DQN algorithm network training steps are as follows:

[0894] A. Status acquisition: Get the current status s t ;

[0895] B. Action selection: According to the ∈ greedy strategy, select an action from the set of optional actions;

[0896] C. Execution action: Students learn knowledge points, update their mastery level, and obtain a new state s t+1 ;

[0897] D. Reward calculation: Calculate the immediate reward based on the reward function;

[0898] E. Experience storage: t , a t , r t ,s t+1 ) is stored in the experience replay pool;

[0899] F. Network update: Randomly extract small batches of samples (s j , a j , r j ,s j+1 )

[0900] Conduct training and update network parameters;

[0901] The real-time recommendation process through the recommendation engine is as follows:

[0902] Status acquisition: Get the student's current user status S;

[0903] Action selection: Select the best knowledge point from A′(s) according to the ∈ greedy strategy;

[0904] Knowledge point recommendation: recommend knowledge points to students and provide corresponding learning resources

[0905] S64): Provide feedback and update on learners through user feedback collection, as follows:

[0906] Learning feedback collection: record students’ learning results of recommended knowledge points, update their mastery level, learning frequency, and recent learning time;

[0907] User status update: Update the student's user status S based on learning feedback

[0908] Reinforcement learning DQN network update: Regularly train the reinforcement learning DQN network, incorporate new learning data, and improve model performance;

[0909] Add learner feedback to the experience replay pool

[0910] In reinforcement learning, experience samples are usually represented in the form of four tuples:

[0911] (s t , a t , r t ,s t+1 )

[0912] In order to incorporate students' feedback into experience samples, the structure of experience samples needs to be expanded to include feedback information:

[0913] (s t , a t , r t′ ,s t+1 , f t )

[0914] r t′ Adjusted instant rewards, taking into account student feedback

[0915] f t The student performs action a at time step t t Feedback

[0916] If the student's feedback is positive, increase the reward immediately

[0917] r t′ =r t +δr

[0918] If negative, reduce the reward

[0919] r t′ =r t -δr.

[0920] In the personalized learning recommendation method based on the personalized knowledge graph described in this embodiment, the specific process of constructing the learning path recommendation model in S7) is as follows:

[0921] S601): The learning feature representation is based on the subject knowledge graph, and the subject knowledge elements and related learning resources are extracted using the association rule algorithm to form the learning element of the knowledge graph, which is represented by a triple L = (Ki, Kj, r);

[0922] Among them, r represents the logical relationship between knowledge element Ki and learning resource Kj;

[0923] Then, using knowledge graph technology, we serialize and annotate the attributes of subject knowledge elements and related learning resources, characterizing and describing the basic characteristics of learning elements.

[0924] Combined with the learning needs of learners, the learning elements of the learning activity process are formed into a learning element sequence (s1, s2, ..., s N );

[0925] At this time, the user status S is expressed as follows:

[0926] s=[s1 s2 ... s N ]

[0927] S602): Recommend learning paths, based on the user's current learning state matrix and the prior relationship of knowledge points in the knowledge graph, recommend the next knowledge point with the maximum reward and the corresponding learning materials;

[0928] S603): Learning path evaluation and optimization: During the learning process, learners complete tasks based on the learning paths and resources recommended by personalized learning, and submit feedback on their learning experience and results. The personalized learning recommendation engine collects and analyzes user feedback, combines the learners' goal completion status, compares the expert path with the recommended path, and evaluates the accuracy and practicality of the recommendation;

[0929] Based on evaluation results and user feedback, we continuously optimize the personalized learning recommendation engine to continuously improve the effectiveness of personalized recommendations and user satisfaction.

[0930] In the personalized learning recommendation method based on personalized knowledge graph described in this embodiment, the knowledge extraction using knowledge graph technology in S3) follows the following rules, which are as follows:

[0931] S31): Template-based rules such as regular expression rules, template matching;

[0932] S32): Syntactic and lexical analysis rules: HanLP (Chinese processing library, supports part-of-speech tagging and word segmentation), spaCy (provides fast part-of-speech tagging, Stanford NLP (part-of-speech tagging, supports multiple languages)

[0933] S33): context-dependent rules;

[0934] S34): Statistical learning rules;

[0935] The specific method of establishing the logical association between subject knowledge elements and related learning resources is as follows: first classify and annotate learning resources, which can be divided into textbooks, exercises, videos, cases, experiments and extracurricular materials. Each resource has clear metadata to describe its association with knowledge elements;

[0936] Create a labeling system for knowledge elements and learning resources, ensure the accuracy of logical associations through manual and automatic labeling (natural language processing or large models, analyze text content, extract knowledge elements or label entities), and perform manual verification.

[0937] According to the knowledge system and curriculum standards of each subject area, knowledge graph technology is used to acquire and extract knowledge, and Python code or large models are used to align entities, attributes, relationships, and resolve conflicts to complete knowledge fusion. The knowledge fusion process is as follows:

[0938] Entity alignment: Different data sources may use different names to describe the same entity, and entity alignment is needed to identify them as the same object; for example, "Apple" and "Apple Inc." are actually the same entity.

[0939] Attribute alignment: One data source uses "Date of Birth" and another uses "Birthdate", but they are actually the same attribute.

[0940] Relationship alignment: Align relationship types to uniformly describe relationships in different data sources.

[0941] “is employed by” and “works for” can be aligned to the same relation type.

[0942] Conflict resolution: Different data sources may provide conflicting attributes about the same entity. For example, one data source says that a person's date of birth is 1980, while another says it is 1981. Conflict resolution techniques are needed to select the most credible version.

[0943] For example, the reward function not only provides feedback based on the learner's task completion or test score, but can also be combined with the knowledge mastery in the knowledge graph for multi-dimensional explanation. By looking at the association of knowledge points in the graph, the system can explain why the learner is rewarded or punished, such as "After completing the recommendation task, the learner has mastered multiple key knowledge points in the knowledge graph, so he received a higher reward."

[0944] The associated paths in the knowledge graph can help the system explain which knowledge points play a key role in the overall progress, which makes the reward mechanism more transparent.

[0945] It also includes a personalized learning recommendation platform based on knowledge graph, which is characterized by:

[0946] Learner interaction interface, which is used as a login port for users to log in to the personalized learning recommendation platform;

[0947] Backstage management interface, used by managers or teachers to log in to the personalized learning recommendation platform port;

[0948] Learning style determination module, used for learning style testing of learners;

[0949] A learning goal setting module is used for learners to set learning goals. The learner interaction interface is connected with the learning style determination module and the learning goal setting module;

[0950] Admission test module, used to test learners’ current learning status;

[0951] Learning process recording module, a module used to record the learner's learning process;

[0952] A learning evaluation module is used to evaluate the collected learning process data of learners, analyze and assess the learning results. The learning evaluation module is bidirectionally connected with the learning process recording module;

[0953] The learner portrait model module is used to establish the learner portrait model and update the learner portrait model according to the learner's later learning situation. The learning style determination module, learning goal setting module, learning process recording module and learning evaluation module are all connected to the learner portrait model module;

[0954] A personalized learning recommendation engine processes the collected user learning data and learning evaluation results to generate a personalized learning recommendation engine for the user, and the personalized learning recommendation engine is connected to the learning evaluation module and the learning process recording module;

[0955] A personalized learning recommendation module, used to provide personalized learning content recommendations and services to users, wherein the personalized learning recommendation content and service module is connected to a personalized learning recommendation engine;

[0956] A learning database is used to provide a data repository for learning-related content. The learning database is connected to the learner portrait model building module, the learning process recording module, the personalized learning recommendation content and service module, and the learning auxiliary tool module, and the personalized learning recommendation engine is bidirectionally connected to the learning database;

[0957] The course content background is a platform for managing all learning content of learners. The course content background is connected with the database and the learning content database in the learning database;

[0958] The knowledge point platform is used to provide the background of knowledge points to the course content background and the knowledge graph in the learning database; it is connected to the knowledge graph in the learning database and the course content background respectively;

[0959] When working, the learner interaction interface, background management interface, learning style determination module, learning goal setting module, entrance test module, learning process recording module, learning assessment module, learning database, course content background and knowledge point platform are the data basis. Every time the learner takes a test in the learning assessment module, the learner portrait model in the personalized learning recommendation platform will be updated in real time. The personalized learning recommendation engine adjusts the recommended learning path and learning content according to the adjusted user portrait model.

[0960] The learning database described in this embodiment includes a learner model database, a learning process database, a learning course database, a learning path database, a learning content database and a knowledge graph. The learner model database, the learning process database, the learning course database, the learning path database and the learning content database are bidirectionally connected to each other. The learner portrait model establishment module is connected to the learner model database, the learning process recording module is connected to the learning process database, the personalized learning recommendation engine is connected to the learning path database, the learning path database is connected to the personalized learning content recommendation module, the learning content database is connected to the course content background, and the knowledge graph is connected to the knowledge point platform.

[0961] The above description is only a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements without departing from the principle of the present invention, and these improvements should also be regarded as within the protection scope of the present invention.

Claims

1. A personalized learning recommendation method based on personalized knowledge graph, characterized by: include: S1): Construct a knowledge graph by first preprocessing the data source, then automatically identifying entities in the text through a dedicated knowledge point extraction fine-tuning model, and extracting the relationships between entities to obtain preprocessed knowledge points and relationships; then disambiguating the knowledge points and relationships; Establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges. Then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph. S2): When learners first enter the personalized learning platform through the learner interaction interface, they must first provide basic personal information and complete registration; Afterwards, learners enter the learning style determination module and learning goal setting module respectively under the guidance of the personalized learning platform wizard to complete their own learning style assessment and learning goal setting, including selectively completing the learner's learning style scale questionnaire, filling in the self-description of learning style and expected content of learning goals; And conduct an initial assessment of the learners’ current learning situation through the entrance test module; S3): Initial learner portrait model establishment, that is, the personalized learning recommendation platform builds a preliminary learner portrait model through the learner portrait model establishment module based on the collected user information; The learner portrait model building module processes the collected information and stores it in the user model database; S4): The learner enters the database to study, and the learning process recording module records the data of the learner's entire learning process, records the sequence of the learner's accessed pages, learning materials, access time information in real time, and instantly updates the user learning record database; S5): The learning assessment module conducts periodic assessments on the learners’ learning status. That is, after each unit of knowledge is learned, the learning assessment module assesses the learners’ learning status recorded by the learning process recording module, and sends the assessment results to the learner portrait model and the personalized learning recommendation engine; S6): updating the learner portrait model, that is, updating the learner portrait model according to the evaluation results of the learning evaluation module, and storing the updated learner model in the learner model database; S7): Based on the updated learner portrait model and learning process records, the personalized learning recommendation engine calculates the learner's mastery of knowledge elements in real time, generates a learner portrait based on this, and calculates the user's comprehensive feature vector. Then, the knowledge graph technology and the reinforcement learning DQN network are used to adaptively and self-organize the learning path and learning resources that are suitable for the learner based on the calculated user comprehensive feature vector, providing users with accurate personalized learning guidance services; S8): updating the learning path database, storing the learning path suitable for the learner newly generated in S7) into the learning path database; S9): Personalized learning content recommendation recommends new learning paths to learners from the learning path database; S10): Then the learner repeats step 4) to step 9) until all the knowledge points to be learned are completed.

2. The personalized learning recommendation method based on personalized knowledge graph according to claim 1, characterized in that: The specific construction method of the knowledge graph in S1) is as follows: S11): data preprocessing, i.e. processing the data source of the knowledge points to be extracted, wherein the data source of the knowledge points to be extracted includes structured data and unstructured data, the structured data is directly input into the mapper, and the text data in the unstructured data is cleaned; S12): extracting knowledge points and relationships from the data source, first building a dedicated knowledge point extraction fine-tuning model, inputting the data after data preprocessing into the knowledge point extraction fine-tuning model, automatically identifying entities in the text through the knowledge point extraction fine-tuning model, and extracting the relationships between the entities to obtain preprocessed knowledge points and relationships; S13): Knowledge point and relationship disambiguation, that is, identification and mapping of different descriptions of the same knowledge point; the specific disambiguation process is as follows: S131): Establishing a standard knowledge point library, that is, first constructing a standard set of knowledge point entities, and using the established standard set of knowledge point entities as a mapping target set to disambiguate knowledge points extracted from the text; S132): Similarity calculation and mapping, i.e., using text embedding technology in natural language processing to convert knowledge points and related texts into vectors; then similarity calculation is performed to determine whether the extracted different expressions point to the same knowledge point by calculating the cosine similarity or Euclidean distance between the text embedding vectors. The cosine similarity is used to measure the angle between the vectors to determine the similarity between the two. S133): semantic disambiguation based on context: to ensure the accuracy of the disambiguation process, the sliding window technology is used to take context information into consideration, so that in the process of identifying knowledge points, not only the current sentence but also the previous and next sentences are considered, and the knowledge point relationship is extracted from the pre-processed knowledge points; S134): Iterative verification and feedback optimization: After disambiguation, the knowledge points and relationships extracted in the previous step will be mapped to the knowledge point entity standard set, and the two will be matched to find possible corresponding relationships. Further verification will be performed through the extracted relationships. If there is no match, the entity will be deleted; S14): Knowledge graph construction: first establish a graph database model, innovate entity nodes, and create relationships between entity nodes to form relationship edges. Then store the disambiguated entity nodes and relationship edges in the graph database model to form a dynamic and scalable knowledge graph. In the later use process, use the AI ​​model regularly to re-analyze the text data and update the knowledge graph to ensure that it reflects the latest knowledge points and their changes in a timely manner.

3. The personalized learning recommendation method based on personalized knowledge graph according to claim 2, characterized in that: The specific method for performing data cleaning on text data in unstructured data in S11) is as follows: S111): Text cleaning: First, clean the text data in the unstructured data, use the deep learning model to classify different text areas in the text, distinguish the irrelevant content from the main text through the context and semantic features of the text, automatically identify the main text and irrelevant content, and remove the irrelevant content through regular expressions; eliminate the noise in the text data, delete the extra spaces, special characters, irrelevant HTML tags, and repeated text; S112): Sentence processing: In order to provide a clear context for subsequent knowledge point extraction, the cleaned text needs to be sentence-separated; S113): Extracting repeated content, extracting repeated content in the text data by detecting repeated sentences and repeated paragraphs, specifically: Detect repeated sentences and use text similarity detection algorithms to identify and merge similar or repeated sentences; Detect repeated paragraphs: Detect and merge repeated paragraphs for longer texts through text comparison and clustering algorithms; S114): Context preservation: In order to preserve the original context of the text data, the following measures are taken: Maintain paragraph structure: While dividing sentences, maintain the paragraph structure of the original text so that the complete paragraph information can be referenced when extracting knowledge points; Use sliding windows: In order to capture a wider range of contextual information, a sliding window technique can be used when extracting knowledge points, which allows the model to consider the information of a sentence and its surrounding sentences at the same time.

4. The personalized learning recommendation method based on personalized knowledge graph according to claim 2, characterized in that: The specific process of sentence processing in S112) is as follows: S1121): Based on punctuation: using periods, question marks, exclamation marks and other punctuation marks as sentence separators; S1122): Sentence segmentation using AI models or natural language processing libraries based on natural language processing tools that can more accurately identify sentence boundaries even in the absence of obvious punctuation; S1123): Dependency-based syntactic analysis: By analyzing the dependency relationship in the sentence, the main and subordinate structures of the sentence are determined, so as to divide the sentence more accurately; The core idea is to decompose a sentence into a "dependency" graph between words, so as to determine the predicate verb, subject, object elements in the sentence and their relationships.

5. The personalized learning recommendation method based on personalized knowledge graph according to claim 4 is characterized in that: The specific analysis process of the dependency syntax analysis in S1123) is as follows: S11231): Input preprocessing, which includes word segmentation, denoising, and normalization of the input text; the details are as follows: Text cleaning: First, the input text data needs to be cleaned and preprocessed, specifically removing noise, symbols, irrelevant HTML tags, and normalizing complex text, such as converting it to lowercase and correcting typos; Word segmentation: Before analyzing the syntax, use word segmentation tools to split the sentence into individual words or phrases; S11232): Part-of-speech tagging First, use natural language processing tools to mark the part of speech for each word. That is, use the part-of-speech tagger to tag each word or phrase after word segmentation and determine the grammatical role of each word or phrase in the sentence, such as noun, verb, or adjective. Part-of-speech tagging provides basic information for subsequent dependency analysis. Simplify the parts of speech: Simplify the parts of speech according to the specific application scenarios of words or phrases, such as simplifying detailed verb forms into a single "verb" category; S11233): Dependency relationship parsing, using the dependency syntax analysis model to generate a dependency relationship tree, as follows: First, sentence component analysis is performed: through dependency syntactic analysis, the dependency relationship between words in the sentence is analyzed to obtain a dependency tree. The goal of dependency syntactic analysis is to find the subordinate relationship between the "core word" and other words in the sentence; Then determine the core components of the sentence: Analyze the predicate verb in the sentence and the related subject, object, and adverbial components. For example, `subject` depends on `verb`, and `object` also depends on `verb`. These components constitute the basic structure of the sentence. Model selection: Select the appropriate dependency syntax model according to the actual application and build dependency relationships; S11234): Extraction of main structure, i.e. analyzing the subject, predicate, object, attributive, adverbial and complement in the sentences in the text to identify the main components of the sentences; S11235): Dependency tree construction, using visualization tools to display dependency relationships, so as to analyze sentence structure more intuitively, as follows: Dependency tree structured representation: A dependency tree is generated for each sentence. The root node of the tree is the predicate verb of the sentence. Other words are attached to the trunk of the dependency tree in sequence according to the dependency relationship. The dependency tree makes it easy to see the master-slave relationship of the components in the sentence, thereby achieving grammatical analysis and sentence reorganization. Dependency tree visualization: Use spaCy's displaCy tool or other dependency tree visualization tools to visualize the parsed dependency relationships, which helps to understand the sentence structure more intuitively. S11236): Syntax rule verification and optimization Dependency tree trunk and subordinate structure identification: By analyzing the dependency tree, the trunk structure of the sentence can be determined, that is, the predicate verb and its main modifying components. The subject and object directly depend on the predicate verb, while the attributive and adverbial subordinate components depend on the core component. Optimization of syntactic rules: According to the needs of the field, some specific syntactic rules are set to correct dependency relationships. For example, for complex sentence patterns such as compound sentences and parallel sentences, conjunctions can be processed through dependency syntax to clarify the position and function of conjunctions in the sentence structure and ensure the accuracy of semantic dependency relationships.

6. The personalized learning recommendation method based on personalized knowledge graph according to claim 1, characterized in that: In the S13), a context embedding model is used in the knowledge point and relationship disambiguation to convert the extracted knowledge points and their contexts into high-dimensional vectors, thereby achieving more accurate semantic matching. Compared with traditional word vectors, context embedding can better capture the different meanings of the same word in different contexts and improve the robustness of disambiguation. The details are as follows: For the input text t, use the contextual embedding model BERT to convert the input text into a high-dimensional vector representation v; υ t =BERT(t) Among them, t is the context embedding vector of text t, Let cosine similarity calculate the angle between two vectors, defined as: Among them, υ1 and υ2 are the contextual embedding vector representations of the two texts; In dynamic weighted cosine similarity, a weight w is assigned to each dimension, considering that different dimensions contribute differently to the similarity. i , so that the similarity calculation can more flexibly reflect the important information in the text: Where d is the dimension, usually d is the hidden layer size of the BERT model; Among them, w i is the weight used for weighted cosine similarity, which is a weight dynamically calculated based on context or category, and is used to measure the contribution of different dimensions to similarity. If two knowledge points are consistent in a certain category, the weight w of the dimension related to the category can be increased. i ; The w i Based on the following factors Term frequency (TF): the frequency with which a word appears in a text, reflecting its importance; Inverse Document Frequency (IDF): how rare a word is in the entire corpus; <h2 style=";text-align:left;direction:ltr">w<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> =TFi·IDFi It can also be calculated by semantic level similarity, as follows: Assume two sentences s1 and s2, their semantic vector representations are and SBERT uses cosine similarity to measure the similarity between two sentences s1 and s2: Compared with traditional sentence representation methods, SBERT can capture more fine-grained semantic information, making the similarity calculation between two sentences at the semantic level more accurate; Multi-level similarity: In order to match semantics more precisely, the similarity of knowledge points at multiple levels can be calculated; the details are as follows: First, the similarity based on the word level is calculated, and then the similarity at the sentence level or paragraph level is calculated. The final similarity can be the weighted average of the similarities at each level: total_similarity(x1, x2) =α·word_similarity(x1,x2)+β·sentence_similarity(x1,x2)+γ·document_similarity(x1,x2) Among them, α, β, and γ are weights of similarity at different levels, which are adjusted according to specific application scenarios. In the processing process, if multiple representations are used for the same entity, the large model is fine-tuned to determine whether they are the same knowledge point, and then mapping and recognition are performed to obtain the final knowledge points and relationships. Mapped to the knowledge point entity standard set through similarity calculation, It also includes a mapping method, and introduces context-aware semantic matching in the matching process by capturing the context of knowledge points in the text; context analysis combines the context information of surrounding words and sentences to enhance the ability to distinguish synonyms or related knowledge points.

7. The personalized learning recommendation method based on personalized knowledge graph according to claim 2, characterized in that: The specific construction process of the knowledge point extraction fine-tuning model in S12) is as follows: (1) Data preparation, that is, preparing enough domain-related text data, including annotations of knowledge points. For different expressions of knowledge points, a standardized set of knowledge points needs to be prepared to facilitate subsequent mapping and verification. The data source can be manually annotated text or use existing knowledge bases such as Wiki or expert Annotated domain datasets; (2) Model selection Choose a pre-trained large language model (such as BERT, GPT, etc.) as the pre-training model. This model has been pre-trained with massive general data and has good language understanding capabilities. For the task of knowledge point extraction, you can choose a model based on the Transformer architecture. These models perform well in natural language processing tasks. (3) Label and clean the text Preprocess the data, including text cleaning (removing noise and irrelevant content) and sentence segmentation. Then, use annotation tools or write rules to mark entities and relationships in the text. This step can use tools such as SpaCy or NLTK for automatic sentence segmentation and part-of-speech tagging. (4) Fine-tune the pre-trained model to obtain a fine-tuned knowledge point extraction fine-tuning model. For the knowledge point extraction task, the pre-trained model such as BERT or GPT is fine-tuned to obtain the fine-tuned model. The training process of the pre-trained model is completed through the following stages: Entity recognition: Use pre-trained models to automatically identify entities in text, such as concepts, facts, and principles; Relationship extraction: extract the relationships between text entities automatically identified by the pre-trained model, such as inclusion, subordination, and predecessor; Disambiguation processing: For different descriptions of the same knowledge point, vectorization technology is used to calculate similarity and map similar expressions to the same entity; (5) Model verification and optimization The fine-tuned model is tested on the validation set to evaluate its accuracy in knowledge point recognition and relationship extraction tasks. The similarity algorithm is used to verify the accuracy and consistency of the extracted entities and relationships. If the accuracy and consistency meet the requirements, it means that the model can meet the needs.

8. The personalized learning recommendation method based on personalized knowledge graph according to claim 1, characterized in that: The learner portrait model construction process in S2) is as follows: S21): Data collection, collecting various data of learners, including learning behavior data, learning resource interaction data, and test scores, and using association rule algorithms and data mining technology to mine users' learning process data; S22): Preprocess the collected data, including cleaning outliers and missing values ​​to ensure data quality. Through attribute matching and feature extraction, combined with learners’ personal attributes and learning styles, various learning portrait modules are integrated using statistical methods to obtain learners’ learning ability, cognitive level, and learning goal portraits. S23): Extract the learner's personalized learning features from the three dimensions of learning ability, cognitive level, and learning goals. Use AprioriAll to quantify the feature labels of learning ability, cognitive level, and learning goals to form the learner's learning feature matrix. Each row represents a learner, and each column corresponds to a feature dimension. S24): Then, based on the learning feature matrix, a machine learning algorithm is used to group or classify the learners to construct a learner portrait model.

9. The personalized learning recommendation method based on personalized knowledge graph according to claim 1, characterized in that: The reinforcement learning DQN network in the personalized learning recommendation engine makes decisions through the interaction between the agent and the environment, and the decision needs to bring as many rewards as possible; the agent here is the personalized learning recommendation engine, the environment is the user's state and knowledge graph, and the reward is the benefit of the learning effect; The specific working process is as follows: S61): First, the data is processed to obtain user portrait features. The user portrait feature is composed of a learning ability vector c, a cognitive level vector l, and a learning goal vector g. The learning ability vector c, the cognitive level vector l, and the learning goal vector g are calculated respectively, and then the user comprehensive feature vector is processed. The specific process is as follows: The following normalization formulas are involved: Where X min is the minimum value, X max is the maximum value, the purpose is to map the range of data to [0, 1]; The learning ability includes memory, logical reasoning ability, learning speed, and learning concentration: The memory dimension is calculated by using the memory-type question score and the total score to calculate the score rate M as an indicator; Normalization: The logical reasoning ability dimension uses the score rate L of reasoning questions as an indicator; Normalization: Learning speed dimension: the ability of users to effectively complete learning tasks within a unit of time, reflecting the speed at which students accept and master new knowledge; including average learning time, learning concentration Average study time: The learning speed is inversely proportional to the average learning time L kp Normalization: Study concentration: The learning concentration refers to the ability of students to maintain concentration and avoid distraction during the learning process, reflecting the degree of learning input; The main indicators of learning concentration are as follows: Continuous learning time period, number of learning interruptions, and other abnormal operations; Learning concentration index Average study time Learning concentration final score formula A min and A max are the minimum and maximum values ​​of the average continuous learning time of users, respectively. min and D max are the minimum and maximum values ​​of distraction indicators of all users, respectively; Among them, ω1 and ω2 are weights, with a value of 0.5; The learning ability matrix is ​​expressed as: C=[M norm L norm L kp_norm A score ] The cognitive levels in the cognitive level vector l are as follows: Among them, A total The total score of all tests in the subject for the user; M total is the sum of the full scores of all tests in the subject; l is the ratio of the total score of all tests in the subject to the sum of the full scores of all tests in the subject; Normalization: The learning goal in the learning goal vector g is: According to the user's target school and target major, knowledge points are selected from the knowledge graph, and then the proficiency of each knowledge point is calculated based on the target score. Assume that the knowledge points obtained by screening according to the target colleges and majors are as follows: k=[k1 k2 … k n ] The weight of each knowledge point is ω(k i ), where i is 1....n; Normalize the weight of each knowledge point According to the normalized weight, the target total score S goal Assign to each knowledge point and adjust the base value to ensure that the weight is greater than 1 S target (k i )=ω′(k i )×(S goal )+1 The matrix of the user learning target vector g is as follows: g=[S target (k1) S target (k2) … S target (ki)] Combine the above features to form the user's comprehensive feature vector F student =[c,l,g] Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector; Using the comprehensive feature vector F student As clustering features, students are divided into K groups and each student is assigned a cluster label G j (j=1, 2, ..., K), indicating the group to which the student belongs, and the clustering label G j As features, incorporated into the state representation of subsequent models; S62): Knowledge graph embedding, that is, collecting all knowledge points and constructing a knowledge point set K = k1, k2, ..., k N ; The Node2Vec algorithm is used to embed the knowledge points in the knowledge graph into the low-dimensional vector space to obtain the embedding vector ei of the i-th knowledge point; Then add the user's comprehensive feature vector, the embedding vector of the knowledge point, and the student's clustering label to obtain the corresponding user state vector, as follows: The user state vector S with the user's comprehensive feature vector added is expressed as follows: Among them, c is the learning ability vector, l is the cognitive level vector, and g is the learning goal vector; the clustering label G of the students is added j and the embedding vector e of the knowledge point i ; The knowledge point vector is calculated as follows: in, is the mastery level: that is, the students’ i The degree of mastery, range [0, 1]; is the learning frequency: that is, the number of times a student learns knowledge point k i The number of times; is the test performance: that is, the student's performance on knowledge point k i Average test scores on is the most recent learning time interval: that is, since the last learning of knowledge point k i the time interval since; e i Embedding vector of knowledge points; f i is a frequent pattern feature; Knowledge point k i The support is: Add the student's cluster label G j The user state vector after It is expressed as follows If the user does not fully grasp the knowledge point, the optional action space is as follows: A(s) is the set of knowledge points that satisfy the prerequisite relationship but are not fully mastered, g i >0 indicates knowledge point k i It is the student’s learning goal; Reward learners for completing their learning goals. The specific reward function is as follows r t =w(k i )×Δs i ×f(c,l)×G(k i )×PM(k i )×AR(k i )×LF(k i )×FC(k i ) Among them, w(k i ) is the knowledge point k i The importance weight Δs i For students to understand the knowledge point k i Improved mastery f(c, l) is the adjustment function of students’ learning ability and cognitive level, as follows: f(c,l)=δ1·c+δ2·l, where δ1 and δ2 are weight parameters, with values ​​of 0.7 and 0.3 respectively; G(k i ) is the learning objective weighting function Among them, g i is the i-th element of the student's learning goal vector, indicating the student's understanding of knowledge point k i The degree of learning needs; γ is a constant greater than 1, which is the bonus coefficient for the learning target knowledge points; When i >0(Knowledge point k i is the student's learning goal), G(k i ) = γ, the reward is amplified by γ times (usually 1.2-2.0), and the expected reward for selecting this knowledge point is submitted. The personalized learning recommendation engine is more inclined to recommend this knowledge point; When i =0, G(k i )=1, the reward remains unchanged, and the personalized learning recommendation engine will not be particularly biased towards recommending this knowledge point; PM(k i ) is a frequent pattern weighting function, which adjusts the reward according to the importance of the knowledge point in the frequent item set, reflecting the importance of the knowledge point in the learning group, as follows: PM(k i )=1+A PM ×Support(k i ) Among them, λ PM To adjust the parameter, the degree of influence on the reward, usually a small positive number such as 0.1 or 0.2; Support(k i ) is the knowledge point k i The support degree of k i Frequency of occurrence in frequent itemsets; By considering the frequency of occurrence of knowledge points in frequent learning patterns, the personalized learning recommendation engine is guided to recommend knowledge points that are commonly learned in the group; In the association rule algorithm, AR(k i )The association rule weighting function is: OR(k i )=1+λ AR ×Conf(k i ) Among them, λ AR To adjust the parameters, control the influence of association rules on rewards, and usually take a positive value; Conf(k i )Knowledge point k i The highest confidence in the association rule, denoted by k i The strength of the connection with the knowledge points already learned; Using the knowledge learned: By considering k i The association with the knowledge points that students have already learned guides the model to recommend knowledge points that are closely related to what has been learned, thus promoting the coherent learning of knowledge; LF(k i )The learning frequency adjustment function is Among them, λ LF Take w(k i ), the more important the knowledge point is, the smaller the reduced reward is; As the learning frequency increases, the function value will gradually decrease, reducing the contribution to the reward and preventing students from excessively repeating familiar knowledge points; FC(k i ) Forgetting curve adjustment function Among them, λ FC Take the proportional function of the knowledge point weight, with a value range of 0.5-1; when Increase, and the reward decreases, prompting students to review this knowledge point; S63): Reinforce the learning DQN network to update the network parameters, as follows: Input layer: user state vector s Output layer: the Q value of each optional action (the Q value is the mathematical expectation of the sum of all future state returns); The DQN algorithm network training steps are as follows: A. Status acquisition: Get the current status St; B. Action selection: According to the ∈ greedy strategy, select an action from the set of optional actions; C. Execution action: Students learn knowledge points, update their mastery level, and obtain a new state s t+1 ; D. Reward calculation: Calculate the immediate reward based on the reward function; E. Experience storage: t , a t , r t ,s t+1 ) is stored in the experience replay pool; F. Network update: Randomly extract small batches of samples (s j , a j , r j ,s j+1 ) to train and update network parameters; The real-time recommendation process through the recommendation engine is as follows: Status acquisition: Get the student's current user status S; Action selection: Select the best knowledge point from A′(s) according to the ∈ greedy strategy; Knowledge point recommendation: recommend knowledge points to students and provide corresponding learning resources S64): Provide feedback and update on learners through user feedback collection, as follows: Learning feedback collection: record students’ learning results of recommended knowledge points, update their mastery level, learning frequency, and recent learning time; User status update: Update the student's user status S based on learning feedback Reinforcement learning DQN network update: Regularly train the reinforcement learning DQN network, incorporate new learning data, and improve model performance; Add learner feedback to the experience replay pool In reinforcement learning, experience samples are usually represented in the form of a quadruple: (s t ,a t ,r t ,s t+1 ) In order to incorporate students' feedback into experience samples, the structure of experience samples needs to be expanded to include feedback information: (s t ,a t ,r t′ ,s t+1 ,f t ) r t′ Adjusted instant rewards, taking into account student feedback f t The student performs action a at time step t t Feedback If the student's feedback is positive, increase the reward immediately r t′ =r t +δr If negative, reduce the reward r t′ =r t -δr。 10. The personalized learning recommendation method based on personalized knowledge graph according to claim 1, characterized in that: The specific process of constructing the learning path recommendation model in S7) is as follows: S601): The learning feature representation is based on the subject knowledge graph, and the subject knowledge elements and related learning resources are extracted using the association rule algorithm to form the learning element of the knowledge graph, which is represented by a triple L = (Ki, Kj, r); Among them, r represents the logical relationship between knowledge element Ki and learning resource N; Then, using knowledge graph technology, we serialize and annotate the attributes of subject knowledge elements and related learning resources, characterizing and describing the basic characteristics of learning elements. Combined with the learning needs of learners, the learning elements of the learning activity process are formed into a learning element sequence (S1, S2, ..., S N ); At this time, the user status S is expressed as follows: s=[s1 s2 … s N ] S602): Recommend learning paths, based on the user's current learning state matrix and the prior relationship of knowledge points in the knowledge graph, recommend the next knowledge point with the maximum reward and the corresponding learning materials; S603): Learning path evaluation and optimization: During the learning process, learners complete tasks based on the learning paths and resources recommended by personalized learning, and submit feedback on their learning experience and results. The personalized learning recommendation engine collects and analyzes user feedback, combines the learners' goal completion status, compares the expert path with the recommended path, and evaluates the accuracy and practicality of the recommendation; Based on evaluation results and user feedback, we continuously optimize the personalized learning recommendation engine to continuously improve the effectiveness of personalized recommendations and user satisfaction.

Images