A learning path automatic navigation method and system based on knowledge point association graph

Abstract

The present invention relates to the technical field of knowledge point association, and discloses a learning path automatic navigation method and system based on a knowledge point association map, the method comprising: collecting subject lecture text data and performing knowledge point extraction; extracting the association relationship between different knowledge points, constructing an initial subject knowledge point association map and performing knowledge point association completion; obtaining the student's learning goal and the mastery status of the initial knowledge point, and using the subject knowledge point association map to search for a learning path. The present invention extracts semantic vectors of knowledge points, represents the association relationship between knowledge points, constructs a subject knowledge point association map that characterizes the association direction and association weight of different knowledge points, extracts the optimal learning path from the subject knowledge point association map based on the number of learning knowledge points, the importance of the learned knowledge points, and the association degree between adjacent knowledge points, and performs learning process navigation, thereby realizing automatic navigation learning of knowledge points.

Description

Technical Field

[0001] The present invention relates to the field of knowledge point association, and in particular to a learning path automatic navigation method and system based on a knowledge point association graph. Background Art

[0002] With the rapid development of information technology, the field of education is gradually moving towards intelligence and personalization. The traditional classroom teaching model is limited by time, space and other factors, and it is difficult to meet the personalized learning needs of students. At the same time, with the popularization of the Internet and the rise of online learning platforms, students can access massive learning resources anytime and anywhere. However, how to find a learning path suitable for personal needs among numerous learning resources has become a problem to be solved urgently. In response to this problem, the present invention proposes a learning path automatic navigation method based on a knowledge point association map, which structures the knowledge points and intuitively displays the association and hierarchy between each knowledge point in the form of a map to help students understand the relationship between the knowledge points, and on this basis, plan personalized learning paths for students. By analyzing the students' existing knowledge level and setting learning goals, the best learning path is automatically generated, thereby helping students learn new knowledge more effectively. Summary of the invention

[0003] In view of this, the present invention provides a learning path automatic navigation method based on a knowledge point association map, the purpose of which is: 1) according to the position, frequency of occurrence and number of co-occurrences of different segmented word phrases in the teaching chapter, the knowledge information of the segmented word phrase in the corresponding teaching chapter is calculated. The greater the knowledge information, the longer the description of the segmented word phrase in the teaching chapter, and the knowledge relationship with other teaching chapters. Then the segmented word phrase is extracted as a knowledge point to realize the extraction of knowledge points in multiple education stages and multiple teaching chapters, and the semantic vectors of the knowledge points are extracted to perform the association relationship between the knowledge points. Representation, construct an initial subject knowledge point association map that characterizes the association directions and association weights of different knowledge points; 2) extract multi-scale weighted semantic vectors between different knowledge points, and calculate the probability of whether there is a semantic path, complete the semantic path from the multi-scale semantic association, obtain the completed subject knowledge point association map, and obtain the students' learning goals and the mastery status of the starting knowledge points. From the number of learning knowledge points in the learning path, the importance of the learned knowledge points, and the degree of association between adjacent knowledge points, select the optimal learning path to navigate the learning process, ensure the smoothness of knowledge point learning, and improve the efficiency of knowledge point learning.

[0004] To achieve the above object, the present invention provides a learning path automatic navigation method based on a knowledge point association graph, comprising the following steps:

[0005] S1: Collect subject lecture text data and extract knowledge points, among which the improved entity relationship extraction based on knowledge information is the implementation method of knowledge point extraction.

[0006] S2: extracting the association relationship between different knowledge points, and constructing an initial subject knowledge point association map, wherein the nodes in the initial subject knowledge point association map represent knowledge points, and the edges between the nodes represent the association relationship between the knowledge points.

[0007] S3: Perform knowledge point association completion on the initial subject knowledge point association graph to obtain a completed subject knowledge point association graph, wherein end-to-end knowledge point relationship prediction based on semantic analysis is an implementation method of the knowledge point association completion.

[0008] S4: Obtain the student's learning goal and the mastery status of the initial knowledge points, use the subject knowledge point association map to search the learning path, obtain the student's knowledge point learning path, and navigate the learning process according to the knowledge point learning path, wherein the implementation method of the learning path search is a comprehensive evaluation based on the multi-dimensional path of knowledge points.

[0009] As a further improvement method of the present invention:

[0010] Optionally, collecting subject lecture text data in step S1 includes:

[0011] Collect subject lecture text data, where the collection process of subject lecture text data is as follows: collect corresponding subject teaching text data in the order of education stages, divide the subject teaching text data into chapters in the order of chapters, and perform word segmentation on the teaching text data of each teaching chapter to form subject lecture text data.

[0012] The collected subject lecture text data is expressed in the form of:

[0013] .

[0014] in:

[0015] Represents the text data of subject lectures;

[0016] represents the sequence of segmented phrases of the mth teaching chapter in the nth education stage, N represents the number of education stages, and M represents the total number of teaching chapters of the subject teaching text data corresponding to each teaching stage;

[0017] Represents a sequence of participle phrases In participle phrases, Represents a sequence of participle phrases The number of participle phrases; Represents a sequence of word segments The i-th participle phrase in i ∈ [1, Num n m ] .

[0018] Knowledge points are extracted from the subject lecture text data to obtain knowledge point sets of different education stages. In the embodiment of the present invention, the education stages include primary school, junior high school, high school and university, and the knowledge points include basic arithmetic, basic geometry, algebra, geometry, number theory, calculus and the like.

[0019] Optionally, the extracting of knowledge points from the subject lecture text data includes:

[0020] Knowledge points are extracted from the subject lecture text data, where improved entity relationship extraction based on knowledge information is the implementation method of knowledge point extraction, and the knowledge point extraction process is as follows:

[0021] S11: Use the word vector model to encode the word vector of the segmented phrase to obtain the word vector corresponding to the segmented phrase; in the embodiment of the present invention, the selected word vector model is the BERT model.

[0022] S12: Calculate the knowledge information of different word segmentation phrases, where the word segmentation phrases The knowledge information is:

[0023] .

[0024] in:

[0025] Indicates participle phrase knowledge information;

[0026] represents an exponential function with a natural constant as base;

[0027] Represents a sequence of word segments Middle participle phrase Number of occurrences;

[0028] Indicates that in the subject teaching text data of the nth teaching stage, there are over-participle phrases Number of teaching chapters.

[0029] S13: taking a segmented word phrase whose knowledge information is higher than a preset threshold as a knowledge point, and obtaining a contextual neighboring word sequence of the segmented word phrase in the corresponding teaching chapter, wherein the contextual neighboring word sequence of the segmented word phrase is a word sequence centered on the segmented word phrase and having a length of Len words;

[0030] A collection of knowledge points that constitute the text data of subject lectures :

[0031] .

[0032] in:

[0033] Represents the sequence of knowledge points in the mth teaching chapter in the nth education stage;

[0034] Represents a sequence of knowledge points In Knowledge points, Represents a sequence of knowledge points The number of knowledge points in Represents a sequence of knowledge points The jth knowledge point in .

[0035] S14: Extract the semantic vector of the knowledge point, where the knowledge point The corresponding semantic vector is .

[0036] Optionally, extracting the association relationship between different knowledge points in step S2 includes:

[0037] Extract the association between different knowledge points, where knowledge points and The calculation process of the association relationship is as follows:

[0038] S21: Acquiring knowledge points and The corresponding semantic vector ,in n 1 , n 2 ∈ [1,N] , m 1 , m 2 ∈ [1,M] , j 1 ∈[1, J n 1 m 1 ] , j 2 ∈[1, J n 2 m 2 ] .

[0039] S22: Calculate the knowledge points and Semantic relevance:

[0040] .

[0041] in:

[0042] Representing knowledge points and semantic relevance;

[0043] represents the L2 norm.

[0044] S23: Calculate the knowledge points and The degree of knowledge contained between:

[0045] ;

[0046] .

[0047] in:

[0048] Representing knowledge points and The degree of knowledge inclusion between them;

[0049] Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence;

[0050] Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence;

[0051] Representing knowledge points and The positive and negative values ​​of the correlation between Indicates that positive and negative values ​​are empty.

[0052] S24: Calculate the knowledge points and The relationship between:

[0053] .

[0054] in:

[0055] Representing knowledge points and The correlation relationship between them.

[0056] Construct an initial subject knowledge point association map based on the association relationship between knowledge points.

[0057] Optionally, constructing an initial subject knowledge point association graph based on association relationships between knowledge points includes:

[0058] An initial subject knowledge point association map is constructed based on the association relationship between knowledge points. , is a collection of knowledge points in the subject lecture text data. is a set of association relationships between knowledge points, and the knowledge points are used as nodes. The association relationships between knowledge points are converted into directed edges between nodes. is positive, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points , is negative, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points , otherwise the direction of the corresponding directed edge is empty, the weight of the directed edge is the absolute value of the association relationship, and the weight is set below the preset weight threshold The weight of is empty.

[0059] Optionally, in step S3, the initial subject knowledge point association graph is supplemented with knowledge point associations, including:

[0060] The initial subject knowledge point association graph is used to complete the knowledge point association. The end-to-end knowledge point relationship prediction based on semantic analysis is the implementation method of the knowledge point association completion. The knowledge point association completion process is as follows:

[0061] S31: Get two knowledge points with empty direction or empty weight , and obtain knowledge points The corresponding semantic vector .

[0062] S32: Semantic vector Perform weighted processing of the multi-scale attention mechanism to form a semantic vector The corresponding multi-scale weighted semantic vectors .

[0063] S33: Calculate the multi-scale weighted semantic vector The probability that there is a semantic path between them:

[0064] ;

[0065] .

[0066] in:

[0067] Represents a multi-scale weighted semantic vector There is a probability of a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points ;

[0068] Represents a multi-scale weighted semantic vector There is a probability of a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points ;

[0069] Representing semantic vector The relationship between them;

[0070] Indicates the preset weight threshold;

[0071] represents the L1 norm.

[0072] S34: If the probability Above a preset probability threshold and , then the knowledge point There is a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points , and use the semantic path as the knowledge point The weight of the directed edge is the probability .

[0073] If the probability Above a preset probability threshold and , then the knowledge point There is a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points , and use the semantic path as the knowledge point The weight of the directed edge is the probability .

[0074] Otherwise, it means knowledge point There is no association between them, and the knowledge points are deleted The edge between.

[0075] Repeat steps S31 to S34 to obtain a subject knowledge point association graph with complete knowledge point associations.

[0076] Optionally, in step S32, the semantic vector Perform weighted processing of multi-scale attention mechanisms, including:

[0077] S321: Set the decomposition scale of the current semantic vector to h, where the initial value of h is 1 and the maximum value is H.

[0078] S322: Decomposition processing is performed on the semantic vector at the hth decomposition scale, where the semantic vector The decomposition formula is:

[0079] ;

[0080] .

[0081] in:

[0082] Representing semantic vector The decomposition result at the hth decomposition scale; in the embodiment of the present invention, ;

[0083] represents the scale convolution matrix;

[0084] Represents an activation function. In the embodiment of the present invention, the selected activation function is a ReLU function.

[0085] S323: Order , return to step S322 until the decomposition result of the Hth decomposition scale is obtained.

[0086] S324: Constructing semantic vector The corresponding multi-scale weighted semantic vector :

[0087] ;

[0088] .

[0089] in:

[0090] Represents the scale weight of the h-th decomposition scale.

[0091] Optionally, in step S4, obtaining the student's learning goal and the mastery status of the initial knowledge point, and searching for a learning path using a subject knowledge point association map include:

[0092] Obtain the student's learning goals and the mastery status of the starting knowledge points, and use the subject knowledge point association map to search for learning paths. The learning path search process is as follows:

[0093] S41: Obtain the student's learning target target, where the student's learning target is the knowledge point that the student currently needs to master, and obtain the student's initial knowledge point mastery status, where the student's initial knowledge point mastery status is the set of knowledge points that the student has mastered.

[0094] S42: Taking any knowledge point in the mastering state of the starting knowledge point as the starting point and the learning target target as the end point, traverse the subject knowledge point association graph to obtain U groups of learning candidate paths, wherein the traversed u-th group of learning candidate paths is:

[0095] .

[0096] in:

[0097] represents the u-th group of candidate learning paths traversed, u ∈ [1,U] ;

[0098] represents the u-th group of learning candidate paths In Knowledge points, represents the u-th group of learning candidate paths The total number of knowledge points in k ∈ [1, K u ] .

[0099] In the embodiment of the present invention, if there is a knowledge point in the subject knowledge point association map Point to knowledge points The directed edge of To knowledge point Traverse.

[0100] S43: Calculate and obtain the recommendation degree of the candidate learning paths, where the recommendation degree of the u-th group of candidate learning paths is:

[0101] .

[0102] in:

[0103] represents the u-th group of learning candidate paths The degree of recommendation;

[0104] Representing knowledge points Strength in the subject knowledge point association graph, where the knowledge point strength is the sum of the outgoing strength and the incoming strength, the outgoing strength is the sum of the weights of the directed edges sent out of the knowledge point, and the incoming strength is the sum of the weights of the directed edges pointing to the knowledge point; in the embodiment of the present invention, the higher the knowledge point strength, the more important the knowledge point;

[0105] Representing knowledge points The corresponding education stage, Indicates the student's current educational stage;

[0106] Representing knowledge points The weight of the directed edge between them.

[0107] S44: Select the candidate learning path with the highest recommendation degree as the learning path search result, and navigate the learning process according to the learning path. In the embodiment of the present invention, the navigation process of the learning path is to randomly select math problems that meet the knowledge points in the order of the learning path knowledge points for the students to push, until the student answers 5 consecutive math problems that meet the knowledge points with a correct answer rate higher than 60%, indicating that the student has mastered the knowledge point, and push the math problems of the next knowledge point according to the learning path.

[0108] Optionally, extracting the semantic vector of the knowledge point in step S14 includes:

[0109] S141: Obtain the contextual adjacent phrase sequence of the segmentation phrase corresponding to the knowledge point, where the knowledge point The contextual adjacent phrase sequence of the corresponding participle phrase is .

[0110] S142: Convert the segmented phrases in the context neighboring phrase sequence into word vectors to form a context neighboring word vector sequence.

[0111] in The corresponding contextual neighbor word vector sequence is , Represents a sequence of contextual neighbor word vectors The Len-th word vector in .

[0112] S143: Extract semantic vectors from the contextual neighboring word vector sequence, where The semantic vector extraction formula is:

[0113] ;

[0114] ;

[0115] .

[0116] in:

[0117] express The semantic vector of

[0118] represents the semantic extraction matrix, * represents the convolution operator;

[0119] Represents a sequence of contextual neighbor word vectors The weighted sequence of Represents a sequence of contextual neighbor word vectors The weight of the c-th word vector in, Represents a sequence of contextual neighbor word vectors The knowledge information of the word segmentation phrase corresponding to the c-th word vector in .

[0120] In order to solve the above problems, the present invention provides a learning path automatic navigation system based on a knowledge point association graph, characterized in that the system comprises:

[0121] The knowledge point association extraction module is used to collect subject lecture text data and extract knowledge points, and extract the association relationship between different knowledge points.

[0122] The association map construction module is used to construct an initial subject knowledge point association map, complete the knowledge point associations of the initial subject knowledge point association map, and obtain a completed subject knowledge point association map.

[0123] The learning path search device is used to obtain the student's learning objectives and the mastery status of the initial knowledge points, use the subject knowledge point association map to search the learning path, obtain the student's knowledge point learning path, and navigate the learning process according to the knowledge point learning path.

[0124] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0125] A memory storing at least one instruction;

[0126] Communication interface, enabling electronic equipment to communicate; and

[0127] The processor executes the instructions stored in the memory to implement the above-mentioned learning path automatic navigation method based on the knowledge point association map.

[0128] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned learning path automatic navigation method based on the knowledge point association map.

[0129] Compared with the prior art, the present invention proposes a learning path automatic navigation method based on a knowledge point association graph, which has the following advantages:

[0130] First, this scheme proposes a method for constructing an initial subject knowledge point association graph to extract knowledge points from subject lecture text data. The knowledge point extraction process is as follows: use the word vector model to encode the word vector of the segmentation phrase to obtain the word vector corresponding to the segmentation phrase; calculate the knowledge information of different segmentation phrases, where the segmentation phrase The knowledge information is:

[0131] .

[0132] in: Indicates participle phrase knowledge information; represents an exponential function with a natural constant as base; Represents a sequence of word segments Middle participle phrase Number of occurrences; Indicates that in the subject teaching text data of the nth teaching stage, there are over-participle phrases The number of teaching chapters; the segmented word phrases whose knowledge information is higher than the preset threshold are regarded as knowledge points, and the contextual neighboring word sequence of the segmented word phrase in the teaching chapter to which it belongs is obtained, wherein the contextual neighboring word sequence of the segmented word phrase is a word sequence with the segmented word phrase as the center and a length of Len words; the knowledge point set of the subject lecture text data is formed :

[0133] .

[0134] in: Represents the sequence of knowledge points in the mth teaching chapter in the nth education stage; Represents a sequence of knowledge points In Knowledge points, Represents a sequence of knowledge points The number of knowledge points in Represents a sequence of knowledge points The jth knowledge point in ; extract the semantic vector of the knowledge point, where the knowledge point The corresponding semantic vector is . Extract the associations between different knowledge points, where knowledge points and The calculation process of the association relationship is as follows: Get knowledge points and The corresponding semantic vector ,in n 1 , n 2 ∈ [1,N] , m 1 , m 2 ∈ [1,M] , j 1 ∈[1, J n 1 m 1 ] , j 2 ∈[1, J n 2 m 2 ] ; Calculate the knowledge points and Semantic relevance:

[0135] .

[0136] in: Representing knowledge points and semantic relevance; Represents the L2 norm; calculate the knowledge point and The degree of knowledge contained between:

[0137] ;

[0138] .

[0139] in: Representing knowledge points and The degree of knowledge inclusion between them; Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence; Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence; Representing knowledge points and The positive and negative values ​​of the correlation between Indicates that the positive and negative values ​​are empty; calculate to get the knowledge point and The relationship between:

[0140] .

[0141] in: Representing knowledge points and The association relationship between knowledge points; construct an initial subject knowledge point association map based on the association relationship between knowledge points. This solution calculates the knowledge information of different segmentation phrases in the corresponding teaching chapter according to their positions, occurrence frequencies, and co-occurrence times in other teaching chapters. The greater the knowledge information, the longer the description of the segmentation phrase in the teaching chapter, and the knowledge relationship with other teaching chapters. The segmentation phrase is then extracted as a knowledge point to achieve knowledge point extraction in multiple education stages and multiple teaching chapters, and the semantic vectors of the knowledge points are extracted to represent the association relationship between knowledge points, and an initial subject knowledge point association map is constructed to characterize the association direction and association weight of different knowledge points.

[0142] At the same time, this scheme proposes a learning path search method to obtain students' learning goals and starting knowledge point mastery status, and use the subject knowledge point association map to search for learning paths. The learning path search process is: obtain the student's learning goal target, where the student's learning goal is the knowledge point that the student currently needs to master, and obtain the student's starting knowledge point mastery status, where the student's starting knowledge point mastery status is the set of knowledge points that the student has mastered; take any knowledge point in the starting knowledge point mastery status as the starting point and the learning goal target as the end point, traverse the subject knowledge point association map to obtain U groups of learning candidate paths, where the u-th group of learning candidate paths traversed is:

[0143] .

[0144] in: represents the u-th group of candidate learning paths traversed, u ∈ [1,U] ; represents the u-th group of learning candidate paths In Knowledge points, represents the u-th group of learning candidate paths The total number of knowledge points in k ∈ [1, K u ] ; Calculate the recommendation degree of the candidate learning paths, where the recommendation degree of the u-th group of candidate learning paths is:

[0145] .

[0146] in: represents the u-th group of learning candidate paths The degree of recommendation; Representing knowledge points The strength in the subject knowledge point association graph, where the knowledge point strength is the sum of the outgoing strength and the incoming strength. The outgoing strength is the sum of the weights of the directed edges sent out from the knowledge point, and the incoming strength is the sum of the weights of the directed edges pointing to the knowledge point. Representing knowledge points The corresponding education stage, Indicates the student's current educational stage; Representing knowledge points The weights of the directed edges between them are calculated; the candidate learning path with the highest recommendation degree is selected as the learning path search result, and the learning process is navigated according to the learning path. This scheme extracts multi-scale weighted semantic vectors between different knowledge points, and calculates the probability of whether a semantic path exists. The semantic path is completed from the multi-scale semantic association to obtain the completed subject knowledge point association map, and the student's learning goals and the mastery status of the starting knowledge points are obtained. The optimal learning path is selected from the number of learning knowledge points in the learning path, the importance of the learned knowledge points, and the degree of association between adjacent knowledge points to navigate the learning process, ensure the smoothness of knowledge point learning, and improve the efficiency of knowledge point learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0147] Figure 1 A flowchart of a method for automatic navigation of a learning path based on a knowledge point association graph is provided in accordance with an embodiment of the present invention.

[0148] Figure 2 A functional module diagram of a learning path automatic navigation system based on a knowledge point association graph provided by an embodiment of the present invention.

[0149] Figure 2 In: 100 a learning path automatic navigation system based on a knowledge point association graph, 101 a knowledge point association extraction module, 102 an association graph construction module, 103 a learning path search device.

[0150] Figure 3 A knowledge point map of junior high school mathematics implemented by an automatic navigation method for a learning path based on a knowledge point association map provided in one embodiment of the present invention.

[0151] Figure 4 A schematic diagram of the structure of an electronic device for implementing a method for automatic navigation of a learning path based on a knowledge point association graph provided by an embodiment of the present invention.

[0152] Figure 4 In: 1 electronic device, 10 processor, 11 memory, 12 program, 13 communication interface.

[0153] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0154] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0155] The embodiment of the present application provides a method for automatic navigation of a learning path based on a knowledge point association graph. The execution subject of the method for automatic navigation of a learning path based on a knowledge point association graph includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for automatic navigation of a learning path based on a knowledge point association graph can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0156] Embodiment 1:

[0157] S1: Collect subject lecture text data and extract knowledge points.

[0158] The subject lecture text data collected in step S1 includes:

[0159] Collect subject lecture text data, where the collection process of subject lecture text data is as follows: collect corresponding subject teaching text data in the order of education stages, divide the subject teaching text data into chapters in the order of chapters, and perform word segmentation on the teaching text data of each teaching chapter to form subject lecture text data.

[0160] The collected subject lecture text data is expressed in the form of:

[0161] .

[0162] in:

[0163] Represents the text data of subject lectures;

[0164] represents the sequence of segmented phrases of the mth teaching chapter in the nth education stage, N represents the number of education stages, and M represents the total number of teaching chapters of the subject teaching text data corresponding to each teaching stage;

[0165] Represents a sequence of word segments In participle phrases, Represents a sequence of word segments The number of participle phrases; Represents a sequence of word segments The i-th participle phrase in i ∈ [1, Num n m ] .

[0166] Knowledge points are extracted from the subject lecture text data to obtain knowledge point sets of different education stages. In the embodiment of the present invention, the education stages include primary school, junior high school, high school and university, and the knowledge points include basic arithmetic, basic geometry, algebra, geometry, number theory, calculus and the like.

[0167] The extraction of knowledge points from the subject lecture text data includes:

[0168] Knowledge points are extracted from the subject lecture text data, where improved entity relationship extraction based on knowledge information is the implementation method of knowledge point extraction, and the knowledge point extraction process is as follows:

[0169] S11: Use the word vector model to encode the word vector of the segmented phrase to obtain the word vector corresponding to the segmented phrase; in the embodiment of the present invention, the selected word vector model is the BERT model.

[0170] S12: Calculate the knowledge information of different word segmentation phrases, where the word segmentation phrases The knowledge information is:

[0171] .

[0172] in:

[0173] Indicates participle phrase knowledge information;

[0174] represents an exponential function with a natural constant as base;

[0175] Represents a sequence of word segments Middle participle phrase Number of occurrences;

[0176] Indicates that in the subject teaching text data of the nth teaching stage, there are over-participle phrases Number of teaching chapters.

[0177] S13: Take the segmented word phrases whose knowledge information is higher than a preset threshold as knowledge points, and obtain the contextual neighboring phrase sequence of the segmented word phrase in the corresponding teaching chapter, wherein the contextual neighboring phrase sequence of the segmented word phrase is a phrase sequence centered on the segmented word phrase and having a length of Len phrases.

[0178] A collection of knowledge points that constitute the text data of subject lectures :

[0179] .

[0180] in:

[0181] Represents the sequence of knowledge points in the mth teaching chapter in the nth education stage;

[0182] Represents a sequence of knowledge points In Knowledge points, Represents a sequence of knowledge points The number of knowledge points in Represents a sequence of knowledge points The jth knowledge point in .

[0183] S14: Extract the semantic vector of the knowledge point, where the knowledge point The corresponding semantic vector is .

[0184] Extracting the semantic vector of the knowledge point in step S14 includes:

[0185] S141: Obtain the contextual adjacent phrase sequence of the segmentation phrase corresponding to the knowledge point, where the knowledge point The contextual adjacent phrase sequence of the corresponding participle phrase is .

[0186] S142: Convert the segmented phrases in the context neighboring phrase sequence into word vectors to form a context neighboring word vector sequence.

[0187] in The corresponding contextual neighbor word vector sequence is , Represents a sequence of contextual neighbor word vectors The Len-th word vector in .

[0188] S143: Extract semantic vectors from the contextual neighboring word vector sequence, where The semantic vector extraction formula is:

[0189] ;

[0190] ;

[0191] .

[0192] in:

[0193] express The semantic vector of

[0194] represents the semantic extraction matrix, * represents the convolution operator;

[0195] Represents a sequence of contextual neighbor word vectors The weighted sequence of Represents a sequence of contextual neighbor word vectors The weight of the c-th word vector in, Represents a sequence of contextual neighbor word vectors The knowledge information of the word segmentation phrase corresponding to the c-th word vector in .

[0196] S2: extracting the association relationship between different knowledge points, and constructing an initial subject knowledge point association map, wherein the nodes in the initial subject knowledge point association map represent knowledge points, and the edges between the nodes represent the association relationship between the knowledge points.

[0197] The S2 step extracts the association relationship between different knowledge points, including:

[0198] Extract the relationship between different knowledge points, where knowledge points and The calculation process of the association relationship is as follows:

[0199] S21: Acquiring knowledge points and The corresponding semantic vector ,in n 1 , n 2 ∈ [1,N] , m 1 , m 2 ∈ [1,M] , j 1 ∈[1, J n 1 m 1 ] , j 2 ∈[1, J n 2 m 2 ] .

[0200] S22: Calculate the knowledge points and Semantic relevance:

[0201] .

[0202] in:

[0203] Representing knowledge points and semantic relevance;

[0204] represents the L2 norm.

[0205] S23: Calculate the knowledge points and The degree of knowledge contained between:

[0206] ;

[0207] .

[0208] in:

[0209] Representing knowledge points and The degree of knowledge inclusion between them;

[0210] Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence;

[0211] Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence;

[0212] Representing knowledge points and The positive and negative values ​​of the correlation between Indicates that positive and negative values ​​are empty.

[0213] S24: Calculate the knowledge points and The relationship between:

[0214] .

[0215] in:

[0216] Representing knowledge points and The correlation relationship between them.

[0217] Construct an initial subject knowledge point association map based on the association relationship between knowledge points.

[0218] The initial subject knowledge point association graph is constructed based on the association relationship between knowledge points, including:

[0219] An initial subject knowledge point association map is constructed based on the association relationship between knowledge points. , is a collection of knowledge points in the subject lecture text data. is a set of association relationships between knowledge points, and the knowledge points are used as nodes. The association relationships between knowledge points are converted into directed edges between nodes. is positive, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points , is negative, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points , otherwise the direction of the corresponding directed edge is empty, the weight of the directed edge is the absolute value of the association relationship, and the weight is set below the preset weight threshold The weight of is empty.

[0220] S3: Complete the knowledge point association map of the initial subject knowledge point association map to obtain a completed subject knowledge point association map.

[0221] In the step S3, the initial subject knowledge point association graph is supplemented with knowledge point associations, including:

[0222] The initial subject knowledge point association graph is used to complete the knowledge point association. The end-to-end knowledge point relationship prediction based on semantic analysis is the implementation method of the knowledge point association completion. The knowledge point association completion process is as follows:

[0223] S31: Get two knowledge points with empty direction or empty weight , and obtain knowledge points The corresponding semantic vector .

[0224] S32: Semantic vector Perform weighted processing of the multi-scale attention mechanism to form a semantic vector The corresponding multi-scale weighted semantic vectors .

[0225] S33: Calculate the multi-scale weighted semantic vector The probability that there is a semantic path between them:

[0226] ;

[0227] .

[0228] in:

[0229] Represents a multi-scale weighted semantic vector There is a probability of a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points ;

[0230] Represents a multi-scale weighted semantic vector There is a probability of a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points ;

[0231] Representing semantic vector The relationship between them;

[0232] Indicates the preset weight threshold;

[0233] represents the L1 norm.

[0234] S34: If the probability Above a preset probability threshold and , then the knowledge point There is a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points , and use the semantic path as the knowledge point The weight of the directed edge is the probability .

[0235] If the probability Above a preset probability threshold and , then the knowledge point There is a semantic path between them, and the direction of the semantic path is the knowledge point Point to knowledge points , and use the semantic path as the knowledge point The weight of the directed edge is the probability .

[0236] Otherwise, it means knowledge point There is no association between them, and the knowledge points are deleted The edge between.

[0237] Repeat steps S31 to S34 to obtain a subject knowledge point association graph with complete knowledge point associations.

[0238] In the step S32, the semantic vector Perform weighted processing of multi-scale attention mechanisms, including:

[0239] S321: Set the decomposition scale of the current semantic vector to h, where the initial value of h is 1 and the maximum value is H.

[0240] S322: Decomposition processing is performed on the semantic vector at the hth decomposition scale, where the semantic vector The decomposition formula is:

[0241] ;

[0242] .

[0243] in:

[0244] Representing semantic vector The decomposition result at the hth decomposition scale; in the embodiment of the present invention, ;

[0245] represents the scale convolution matrix;

[0246] Represents an activation function. In the embodiment of the present invention, the selected activation function is a ReLU function.

[0247] S323: Order , return to step S322 until the decomposition result of the Hth decomposition scale is obtained.

[0248] S324: Constructing semantic vector The corresponding multi-scale weighted semantic vector :

[0249] ;

[0250] .

[0251] in:

[0252] Represents the scale weight of the h-th decomposition scale.

[0253] S4: Obtain students’ learning goals and initial knowledge point mastery status, use the subject knowledge point association map to search for learning paths, obtain students’ knowledge point learning paths, and navigate the learning process according to the knowledge point learning paths.

[0254] In the step S4, the learning goal of the student and the mastery status of the starting knowledge point are obtained, and the learning path search is performed using the subject knowledge point association map, including:

[0255] Obtain the student's learning goals and the mastery status of the starting knowledge points, and use the subject knowledge point association map to search for learning paths. The learning path search process is as follows:

[0256] S41: Obtain the student's learning target target, where the student's learning target is the knowledge point that the student currently needs to master, and obtain the student's initial knowledge point mastery status, where the student's initial knowledge point mastery status is the set of knowledge points that the student has mastered.

[0257] S42: Taking any knowledge point in the mastering state of the starting knowledge point as the starting point and the learning target target as the end point, traverse the subject knowledge point association graph to obtain U groups of learning candidate paths, wherein the traversed u-th group of learning candidate paths is:

[0258] .

[0259] in:

[0260] represents the u-th group of candidate learning paths traversed, u ∈ [1,U] ;

[0261] represents the u-th group of learning candidate paths In Knowledge points, represents the u-th group of learning candidate paths The total number of knowledge points in k ∈ [1, K u ] .

[0262] In the embodiment of the present invention, if there is a knowledge point in the subject knowledge point association map Point to knowledge points The directed edge of To knowledge point Traverse.

[0263] S43: Calculate and obtain the recommendation degree of the candidate learning paths, where the recommendation degree of the u-th group of candidate learning paths is:

[0264] .

[0265] in:

[0266] represents the u-th group of learning candidate paths The degree of recommendation;

[0267] Representing knowledge points The strength in the subject knowledge point association graph, where the knowledge point strength is the sum of the outgoing strength and the incoming strength. The outgoing strength is the sum of the weights of the directed edges sent out from the knowledge point, and the incoming strength is the sum of the weights of the directed edges pointing to the knowledge point.

[0268] Representing knowledge points The corresponding education stage, Indicates the student's current educational stage;

[0269] Representing knowledge points The weight of the directed edge between them.

[0270] S44: Select the candidate learning path with the highest recommendation degree as the learning path search result, and navigate the learning process according to the learning path.

[0271] In the embodiment of the present invention, a learning path of logarithmic operation knowledge points is obtained based on natural number knowledge points and a learning path is recommended.

[0272] The path obtained based on the constructed junior high school data knowledge point map is as follows:

[0273] 1. Natural numbers → integers → rational numbers → real numbers → rational number operations → logarithmic operations;

[0274] 2. Natural numbers → integers → rational numbers → real numbers → real number operations → logarithmic operations;

[0275] 3. Natural numbers → integers → rational numbers → real numbers → algebraic expressions → exponential operations → logarithmic operations;

[0276] 4. Natural numbers → integers → rational numbers → real numbers → algebraic expressions → fractions → logarithmic operations;

[0277] The calculation results of the learning path recommendation degree are shown in Table 1. It can be seen from the table that the learning path preferentially recommended by the present invention is path 2, which is in line with the current knowledge point learning progress in the actual teaching process.

[0278] Table 1

[0279] Path 1 Path 2 Path 3 Path 4 0.78 0.87 0.69 0.34

[0280] Based on the (point and line) knowledge points, the learning path to the volume knowledge point is obtained and the learning path is recommended. The path obtained based on the constructed junior high school data knowledge point map is as follows:

[0281] 1. Point → Line → Line Segment → Ray → Angle → Triangle → Quadrilateral → Circle → Cylinder → Volume;

[0282] 2. Point → Line → Line Segment → Ray → Angle → Triangle → Quadrilateral → Square → Rectangle → Cylinder → Volume;

[0283] 3. Point → Line → Line Segment → Ray → Angle → Triangle → Quadrilateral → Circle → Cone → Volume;

[0284] 4. Point → Line → Line Segment → Ray → Angle → Triangle → Quadrilateral → Cone → Volume;

[0285] The calculation results of the learning path recommendation degree are shown in Table 2. It can be seen from the table that the preferred learning path is path 2, which is in line with the current knowledge point learning progress in the actual teaching process.

[0286] Table 2

[0287] Path 1 Path 2 Path 3 Path 4 0.87 0.92 0.77 0.66 .

[0288] Embodiment 2:

[0289] like Figure 2 As shown, it is a functional module diagram of a learning path automatic navigation system based on a knowledge point association map provided by an embodiment of the present invention, which can implement the learning path automatic navigation method based on a knowledge point association map in Example 1.

[0290] The automatic learning path navigation system 100 based on the knowledge point association map of the present invention can be installed in an electronic device. According to the functions implemented, the automatic learning path navigation system based on the knowledge point association map can include a knowledge point association extraction module 101, an association map construction module 102 and a learning path search device 103. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0291] The knowledge point association extraction module 101 is used to collect subject lecture text data and perform knowledge point extraction, and extract the association relationship between different knowledge points.

[0292] The association map construction module 102 is used to construct an initial subject knowledge point association map, complete the knowledge point associations of the initial subject knowledge point association map, and obtain a completed subject knowledge point association map.

[0293] The learning path search device 103 is used to obtain the student's learning goal and the mastery status of the initial knowledge points, use the subject knowledge point association map to search the learning path, obtain the student's knowledge point learning path, and navigate the learning process according to the knowledge point learning path.

[0294] In detail, the modules in the learning path automatic navigation system 100 based on the knowledge point association map in the embodiment of the present invention are used in the same manner as above. Figure 1 The same technical means are used as the automatic navigation method of learning paths based on the knowledge point association map described in , and can produce the same technical effects, so they will not be repeated here.

[0295] Embodiment 3:

[0296] like Figure 4 , is a schematic diagram of the structure of an electronic device for implementing a method for automatic navigation of a learning path based on a knowledge point association graph provided by an embodiment of the present invention.

[0297] The electronic device 1 may include a processor 10 , a memory 11 , a communication interface 13 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10 , such as a program 12 .

[0298] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 may not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the program 12, etc., but also be used to temporarily store data that has been output or is to be output.

[0299] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (such as a program 12 for realizing automatic navigation of a learning path based on a knowledge point association map), and calls data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

[0300] The communication interface 13 may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices, and to achieve connection and communication between internal components of the electronic device.

[0301] The bus may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0302] Figure 4 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 4 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0303] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0304] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0305] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0306] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0307] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0308] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A learning path automatic navigation method based on a knowledge point association graph, characterized in that: The method comprises: S1: Collect subject lecture text data and extract knowledge points; S2: extracting the association relationship between different knowledge points and constructing an initial subject knowledge point association map, wherein the nodes in the initial subject knowledge point association map represent knowledge points, and the edges between the nodes represent the association relationship between the knowledge points; S3: completing the knowledge point association map of the initial subject knowledge point association map to obtain a completed subject knowledge point association map; S4: Obtain the student's learning goal and the mastery status of the starting knowledge points, use the subject knowledge point association map to search for the learning path, obtain the student's knowledge point learning path, and navigate the learning process according to the knowledge point learning path; The subject lecture text data collected in step S1 includes: Collecting subject lecture text data, wherein the collection process of subject lecture text data is as follows: collecting corresponding subject teaching text data in the order of education stages, dividing the subject teaching text data into chapters in the order of chapters, and performing word segmentation processing on the teaching text data of each teaching chapter to form subject lecture text data; The collected subject lecture text data is expressed in the form of: in: X represents the subject lecture text data; represents the sequence of segmented phrases of the mth teaching chapter in the nth education stage, N represents the number of education stages, and M represents the total number of teaching chapters of the subject teaching text data corresponding to each teaching stage; Represents a sequence of participle phrases In participle phrases, Represents a sequence of participle phrases The number of participle phrases; Represents a sequence of participle phrases The i-th participle phrase in Extract knowledge points from subject lecture text data to obtain knowledge point sets at different educational stages; The extraction of knowledge points from the subject lecture text data includes: Knowledge points are extracted from the subject lecture text data, and the knowledge point extraction process is as follows: S11: Use the word vector model to encode the word vector of the segmented word group to obtain the word vector corresponding to the segmented word group; S12: Calculate the knowledge information of different word segmentation phrases, where the word segmentation phrases The knowledge information is: in: Indicates participle phrase knowledge information; exp(·) represents an exponential function with a natural constant as the base; Represents a sequence of word segments Middle participle phrase Number of occurrences; Indicates that in the subject teaching text data of the nth teaching stage, there are over-participle phrases Number of teaching chapters; S13: taking a segmented word phrase whose knowledge information is higher than a preset threshold as a knowledge point, and obtaining a contextual neighboring word sequence of the segmented word phrase in the corresponding teaching chapter, wherein the contextual neighboring word sequence of the segmented word phrase is a word sequence centered on the segmented word phrase and having a length of Len words; The knowledge point set x that constitutes the subject lecture text data: in: Represents the sequence of knowledge points in the mth teaching chapter in the nth education stage; Represents a sequence of knowledge points In Knowledge points, Represents a sequence of knowledge points The number of knowledge points in Represents a sequence of knowledge points The jth knowledge point in ; S14: Extract the semantic vector of the knowledge point, where the knowledge point The corresponding semantic vector is The S2 step extracts the association relationship between different knowledge points, including: Extract the relationship between different knowledge points, where knowledge points and The calculation process of the association relationship is as follows: S21: Acquiring knowledge points and The corresponding semantic vector where n1,n2∈[1,N],m1,m2∈[1,M], S22: Calculate the knowledge points and Semantic relevance: in: Representing knowledge points and semantic relevance; ||·||2 represents the L2 norm; S23: Calculate the knowledge points and The degree of knowledge contained between: in: Representing knowledge points and The degree of knowledge inclusion between them; Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence; Representing knowledge points The corresponding participle phrase sequence contains knowledge points The number of included phrases of any participle phrase in the corresponding participle phrase sequence; sgn(n1-n2) represents knowledge points and The positive and negative values ​​of the relationship between them, / means the positive and negative values ​​are empty; S24: Calculate the knowledge points and The relationship between: in: Representing knowledge points and The relationship between them; Construct an initial subject knowledge point association graph based on the association relationship between knowledge points; The initial subject knowledge point association graph is constructed based on the association relationship between knowledge points, including: Based on the association relationship between knowledge points, an initial subject knowledge point association graph is constructed, wherein the initial subject knowledge point association graph G0=(x,Q), where x is the knowledge point set of subject lecture text data, and Q is the association relationship set between knowledge points. The knowledge points are used as nodes, and the association relationship between knowledge points is converted into directed edges between nodes. is positive, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points is negative, then the direction of the corresponding directed edge is from the knowledge point Point to knowledge points Otherwise, the direction of the corresponding directed edge is empty, and the weight of the directed edge is the absolute value of the association relationship, and the weight is set to be lower than the preset weight threshold. The weight of is empty; In the step S3, the initial subject knowledge point association graph is supplemented with knowledge point associations, including: The initial subject knowledge point association graph is used to complete the knowledge point association, and the knowledge point association completion process is as follows: S31: Obtain two knowledge points x(1) and x(2) whose directions are empty or whose weights are empty, and obtain the semantic vectors z(1) and z(2) corresponding to the knowledge points x(1) and x(2); S32: Perform weighted processing of the semantic vectors z(1) and z(2) using a multi-scale attention mechanism to form multi-scale weighted semantic vectors B(1) and B(2) corresponding to the semantic vectors z(1) and z(2) respectively; S33: Calculate the probability that a semantic path exists between the multi-scale weighted semantic vectors B(1) and B(2): in: P(B(1), B(2)) represents the probability that there is a semantic path between the multi-scale weighted semantic vectors B(1) and B(2), and the direction of the semantic path is from knowledge point x(1) to knowledge point x(2); P(B(2),B(1)) represents the probability that there is a semantic path between the multi-scale weighted semantic vectors B(1) and B(2), and the direction of the semantic path is from knowledge point x(2) to knowledge point x(1); Q(z(1),z(2)) represents the association relationship between semantic vectors z(1) and z(2); Indicates the preset weight threshold; ||·|| represents the L1 norm; S34: If the probability P(B(1), B(2)) is higher than the preset probability threshold and P(B(2), B(1)), it is considered that there is a semantic path between the knowledge points x(1) and x(2), and the direction of the semantic path is from the knowledge point x(1) to the knowledge point x(2), and the semantic path is used as a directed edge between the knowledge points x(1) and x(2), and the weight of the directed edge is the probability P(B(1), B(2)); If the probability P(B(2), B(1)) is higher than the preset probability threshold and P(B(1), B(2)), it is considered that there is a semantic path between knowledge points x(1) and x(2), and the direction of the semantic path is from knowledge point x(2) to knowledge point x(1). The semantic path is taken as a directed edge between knowledge points x(1) and x(2), and the weight of the directed edge is the probability P(B(2), B(1)). Otherwise, it means that there is no association between knowledge points x(1) and x(2), and the edge between knowledge points x(1) and x(2) is deleted; Repeat steps S31 to S34 to obtain a subject knowledge point association graph with complete knowledge point associations; In the step S32, the semantic vectors z(1) and z(2) are subjected to weighted processing using a multi-scale attention mechanism, including: S321: setting the decomposition scale of the current semantic vector to h, where the initial value of h is 1 and the maximum value is H; S322: Decomposition processing is performed on the semantic vector at the hth decomposition scale, wherein the decomposition processing formula of the semantic vectors z(1) and z(2) is: With h (1)=δ(W2*z h-1 (1))+with h-1 (1); With h (2)=δ(W2*z h-1 (2))+with h-1 (2); in: z h (1),z h (2) represents the decomposition results of semantic vectors z(1), z(2) at the hth decomposition scale; W2 represents the scale convolution matrix; δ(·) represents the activation function; S323: Let h=h+1, and return to step S322 until the decomposition result of the Hth decomposition scale is obtained; S324: Construct multi-scale weighted semantic vectors B(1), B(2) corresponding to the semantic vectors z(1), z(2): in: represents the scale weight of the h-th decomposition scale; In the step S4, the learning goal of the student and the mastery status of the starting knowledge point are obtained, and the learning path search is performed using the subject knowledge point association map, including: Obtain the student's learning goals and the mastery status of the starting knowledge points, and use the subject knowledge point association map to search for learning paths. The learning path search process is as follows: S41: obtaining the student's learning target target, where the student's learning target is the knowledge point that the student currently needs to master, and obtaining the student's initial knowledge point mastery status, where the student's initial knowledge point mastery status is the set of knowledge points that the student has mastered; S42: Taking any knowledge point in the mastering state of the starting knowledge point as the starting point and the learning target target as the end point, traverse the subject knowledge point association graph to obtain U groups of learning candidate paths, wherein the traversed u-th group of learning candidate paths is: in: L u represents the u-th group of candidate learning paths traversed, u∈[1,U]; target represents the u-th group of learning candidate paths L u K u +1 knowledge point, K u +1 indicates the u-th group of learning candidate paths L u The total number of knowledge points in , k∈[1,K u ]; S43: Calculate and obtain the recommendation degree of the candidate learning paths, where the recommendation degree of the u-th group of candidate learning paths is: in: V u Denotes the u-th group of learning candidate paths L u The degree of recommendation; Representing knowledge points The strength in the subject knowledge point association graph, where the knowledge point strength is the sum of the outgoing strength and the incoming strength. The outgoing strength is the sum of the weights of the directed edges sent out from the knowledge point, and the incoming strength is the sum of the weights of the directed edges pointing to the knowledge point. Representing knowledge points Corresponding to the education stage, Level indicates the student’s current education stage; Representing knowledge points The weights of the directed edges between them; S44: selecting the candidate learning path with the highest recommendation degree as the learning path search result, and navigating the learning process according to the learning path; Extracting the semantic vector of the knowledge point in step S14 includes: S141: Obtain the contextual adjacent phrase sequence of the segmentation phrase corresponding to the knowledge point, where the knowledge point The contextual adjacent phrase sequence of the corresponding participle phrase is S142: Convert the segmented word groups in the context neighboring word group sequence into word vectors to form a context neighboring word vector sequence; in The corresponding contextual neighbor word vector sequence is Represents a sequence of contextual neighbor word vectors The Lenth word vector in ; S143: Extract semantic vectors from the contextual neighboring word vector sequence, where The semantic vector extraction formula is: in: express The semantic vector of W1 represents the semantic extraction matrix, * represents the convolution operator; Represents a sequence of contextual neighbor word vectors The weighted sequence of Represents a sequence of contextual neighbor word vectors The weight of the c-th word vector in, Represents a sequence of contextual neighbor word vectors The knowledge information of the word segmentation phrase corresponding to the c-th word vector in .

2. A learning path automatic navigation system based on a knowledge point association graph, characterized in that: The system comprises: The knowledge point association extraction module is used to collect subject lecture text data and extract knowledge points, and extract the association relationship between different knowledge points; The association map construction module is used to construct an initial subject knowledge point association map, and to complete the knowledge point associations of the initial subject knowledge point association map to obtain a completed subject knowledge point association map; A learning path search device is used to obtain students' learning goals and starting knowledge point mastery status, use subject knowledge point association maps to search for learning paths, obtain students' knowledge point learning paths, and navigate the learning process according to the knowledge point learning paths, so as to realize the learning path automatic navigation method based on the knowledge point association map as described in claim 1.

Images