A knowledge point location method driven by big data

Abstract

The present invention relates to the technical field of knowledge point positioning, and discloses a knowledge point positioning method driven by big data, the method comprising: evaluating and quantifying knowledge points in combination with a multidimensional knowledge point evaluation index system; interactively fusing knowledge point index vectors according to the learner's answering process to obtain a learner's knowledge fusion ability vector; and using a knowledge point mastery degree evaluation model to predict the mastery degree of knowledge points and locate weak knowledge points. The present invention combines the frequency of occurrence of knowledge points and the joint frequency of occurrence to calculate the importance of knowledge points and the correlation with other knowledge points, constructs a knowledge point index vector that characterizes the multidimensional information of knowledge points, extracts the learner's ability value and the effort characteristics that characterize the learner's ranking, obtains the learner's mastery weight coefficients of different knowledge points through fusion calculation, performs fusion calculation and mapping processing on the two, obtains the learner's weak knowledge points, and realizes knowledge point positioning.

Description

Technical Field

[0001] The present invention relates to the technical field of knowledge point positioning, and in particular to a big data driven knowledge point positioning method. Background Art

[0002] With the rapid development of information technology and the popularization of the Internet, big data technology has gradually become a key supporting technology for various industries, leading a new round of technological revolution. In the field of education, the application of educational big data is increasingly valued, and its potential is believed to be able to profoundly change traditional teaching methods and improve teaching quality and efficiency. Knowledge point positioning is a crucial part of teaching. Teachers and educational institutions need to accurately locate students' knowledge points in order to provide more targeted teaching content and methods. In the process of realizing big data-driven knowledge point positioning, there are challenges such as data diversity, massiveness and real-time nature, and there are many difficulties in the quantification process of students' abilities and knowledge point mastery. To address this problem, a new knowledge point positioning method can be studied. By analyzing students' learning data and behavior patterns, the knowledge point mastery can be mined from students' learning behaviors, test scores and other data, which can help teachers better understand students' learning status, weaknesses and strengths, and provide support for personalized teaching and precise tutoring. Summary of the invention

[0003] In view of this, the present invention provides a big data driven knowledge point positioning method, which uses machine learning algorithms to analyze massive student learning data and automatically generates a learning model to help teachers better grasp students' difficult knowledge points and improve the targeted teaching.

[0004] To achieve the above object, the present invention provides a big data driven knowledge point location method, comprising the following steps:

[0005] S1: Construct a multi-dimensional knowledge point evaluation index system, combine the multi-dimensional knowledge point evaluation index system to evaluate and quantify the knowledge points, and obtain the knowledge point index vector;

[0006] S2: According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector;

[0007] S3: construct the knowledge point mastery degree assessment model structure, and optimize the model parameters to obtain the optimal model parameters, and construct the knowledge point mastery degree assessment model based on the knowledge point mastery degree assessment model structure and the optimal model parameters;

[0008] The knowledge point mastery degree assessment model structure includes an input layer, a knowledge encoding layer, a reliability feature extraction layer, an information fusion layer and a knowledge point assessment layer;

[0009] The input layer is used to receive the learner's knowledge fusion ability vector;

[0010] The knowledge encoding layer is used to encode the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector;

[0011] The reliability feature extraction layer is used to extract the reliability features of knowledge points from the encoded mastery weight coefficients;

[0012] The information fusion layer is used to fuse the credibility features and the encoded mastery weight coefficients to obtain the mastery information features of the knowledge points;

[0013] The knowledge point evaluation layer is used to map the mastery information features of the knowledge points to obtain the predicted results of the mastery degree of the knowledge points, and regard the knowledge points whose predicted results of the mastery degree of the knowledge points are lower than the specified threshold as weak knowledge points;

[0014] S4: Use the knowledge point mastery evaluation model to receive the learner's knowledge fusion ability vector and predict the knowledge point mastery. The knowledge points whose knowledge point mastery prediction results are lower than the specified threshold are regarded as weak knowledge points, and the knowledge points are located.

[0015] As a further improvement method of the present invention:

[0016] Optionally, the S1 step includes:

[0017] Construct a multi-dimensional knowledge point evaluation index system and a question bank containing all knowledge points, wherein the multi-dimensional knowledge point evaluation index system includes knowledge point relevance, knowledge point difficulty, knowledge point distinction, knowledge point importance and knowledge point guessing factor. There are several questions in the question bank, each question contains at least one knowledge point, and the set of the knowledge points is :

[0018] ;

[0019] in:

[0020] Indicates the first Knowledge points, Represents the total number of knowledge points in the set;

[0021] Combined with the multi-dimensional knowledge point evaluation index system, the knowledge points are evaluated and quantified to obtain the knowledge point index vector, where the knowledge point The corresponding knowledge point index vector is , knowledge points The evaluation and quantification process is as follows:

[0022] S11: Calculate the knowledge points The correlation between other knowledge points in the knowledge point set is constructed to obtain the knowledge point The relevance of knowledge points :

[0023] ;

[0024] in:

[0025] Representing knowledge points And knowledge points The degree of correlation between

[0026] Indicates that the question bank contains knowledge points The number of topics, Indicates that it also contains knowledge points And knowledge points The number of questions;

[0027] Indicates the total number of questions in the question bank;

[0028] S12: Based on knowledge points The relevance of knowledge points , for knowledge points Calculate the importance of association and use the result of the calculation of the importance of association as the knowledge point Importance of knowledge points :

[0029] ;

[0030] in:

[0031] represents an exponential function with a natural constant as base;

[0032] Indicates the preset knowledge relevance threshold;

[0033] represents the smoothing parameter, Set to 2;

[0034] represents a symbolic function;

[0035] Indicates selection The maximum value in ;

[0036] S13: Select questions from the question bank and construct a question containing knowledge points The question sequence is D, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained. The mean difficulty, discrimination and guessing parameters of all questions are taken as the knowledge points. Difficulty of knowledge points , Knowledge point differentiation And the knowledge point guessing factor ;

[0037] S14: Construct knowledge points Knowledge point index vector , .

[0038] Optionally, the step S13 constructs a The question sequence is obtained, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained, including:

[0039] S131: Constructing knowledge points The title sequence :

[0040] ;

[0041] in:

[0042] Represents a sequence of questions The Road topic, , Indicates the total number of questions in the constructed question sequence;

[0043] S132: Get D answerers in the question sequence The answer results of the dth answerer in the question The answer result is , Indicates that the dth respondent answered the question correctly , Indicates that the dth respondent answered the question incorrectly , ;

[0044] S133: Construct the objective function of question information estimation based on the answer results of D respondents :

[0045] ;

[0046] in:

[0047] represents the vector to be estimated, represents the ability parameter vector of the respondent, Indicates the ability of the dth respondent; Represents a sequence of questions The discrimination vector of the questions in , Indicate the topic The discrimination degree of b represents the question sequence The difficulty vector of the question in , Indicate the topic Difficulty; Represents a sequence of questions The guess parameter vector for the question in , Indicate the topic Guessed parameters of

[0048] S134: The objective function is solved by using a gradient descent method to obtain the difficulty, discrimination and guessing parameters of each question.

[0049] Optionally, the S2 step includes:

[0050] Get the learner's answer questions and answer results as the learner's answer process :

[0051] ;

[0052] in:

[0053] Indicates the information of the kth question answered by the learner, including the difficulty of the question , Discrimination , guess parameters And the knowledge points included, Indicates the total number of questions answered;

[0054] represents the answer result of the learner in the kth question. Indicates that the learner answered the kth question correctly, Indicates that the learner answered the kth question incorrectly;

[0055] According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector. The interactive fusion processing process is as follows:

[0056] S21: Calculate the learner's ability value based on the difficulty, discrimination and guessing parameters of the answer questions ;

[0057] S22: Based on ability value Rank the learners in reverse order to get the ranking results of the learners ; The higher the ability value, the lower the ranking;

[0058] S23: According to the ranking results Calculate the learner's effort characteristics :

[0059] ;

[0060] in:

[0061] represents the total number of learners;

[0062] S24: Combined with learners’ ability values and effort characteristics , decompose and fuse the knowledge point index vector of the knowledge point to obtain the learner's mastery weight coefficient at different knowledge points, where the learner's mastery weight coefficient at knowledge points The mastery weight coefficient is :

[0063] ;

[0064] in:

[0065] Indicates that learners are at the knowledge point The probability of mastering Indicates that you have fully mastered the knowledge points In the case of wrong answer The probability of the corresponding question;

[0066] Indicates ability value Answer the knowledge points correctly The probability of the corresponding question, Representing knowledge points The correct answer probability threshold;

[0067] Indicates that learners are at the knowledge point The correlation mastery coefficient, Indicates that K questions contain knowledge points The proportion of topics Indicates that K questions contain knowledge points The proportion of topics Representing knowledge points And knowledge points The relationship between Representing knowledge points And knowledge points The degree of correlation between

[0068] T stands for transpose;

[0069] S25: The learner's mastery weight coefficients of all knowledge points form the learner's knowledge fusion ability vector w: .

[0070] Optionally, the step of optimizing model parameters in combination with the knowledge point mastery evaluation model structure to obtain optimal model parameters includes:

[0071] S31: Obtain the learner knowledge fusion ability vectors and weak knowledge points of U learners to form the model parameter training data set data:

[0072] ;

[0073] in:

[0074] represents the learner knowledge fusion ability vector of the u-th learner, represents the weak knowledge point of the u-th learner;

[0075] S32: marking the model parameters to be optimized in the knowledge point mastery degree assessment model structure, wherein the model parameters to be optimized include the encoding matrix in the knowledge encoding layer and the training matrix in the knowledge point assessment layer;

[0076] S33: Initialize and generate Z groups of model parameters, wherein the zth group of model parameters generated by initialization is , , set the current iteration number of the model parameters to c, and the maximum iteration number to C, then the iteration result of the zth group of model parameters in the cth iteration is ,in Represents model parameters The encoding matrix in the corresponding knowledge encoding layer, Represents model parameters The two corresponding training matrices in ;

[0077] S34: Combine the model parameter training data set data and construct a fitness function with the model parameters as input, where the model parameters The fitness function of the input is :

[0078] ;

[0079] in:

[0080] represents the discriminant function, if If they are equal, the discriminant function is 1, otherwise the discriminant function is 0;

[0081] Indicates the use of model parameters Construct a knowledge point mastery evaluation model;

[0082] Represents the knowledge point mastery evaluation model The calculated weak points of the u-th learner;

[0083] S35: Substitute the Z groups of model parameters obtained in the cth iteration into the fitness function to obtain the fitness value of each group of model parameters, and use the model parameters with the highest fitness value as the optimal model parameters for the cth iteration ;

[0084] S36: Generate the inverse solution of the Z group of model parameters obtained in the cth iteration, where the zth group of model parameters The reverse solution is :

[0085] ;

[0086] in:

[0087] represents the lower bound of the model parameters, represents the upper bound of the model parameters;

[0088] like , then keep ,neglect , otherwise let ; Indicates the preset fitness threshold;

[0089] S37: Iterate the model parameters, where the iteration formula is:

[0090] ;

[0091] in:

[0092] represents the mean value of the Z group model parameters obtained in the cth iteration, Represents a random number between 0 and 1;

[0093] S38: Let c=c+1, return to step S35, until the maximum number of iterations is reached, and select the model parameter with the highest current fitness value as the optimal model parameter.

[0094] Optionally, in step S4, using the knowledge point mastery degree assessment model to receive the learner's knowledge fusion ability vector and perform knowledge point mastery degree prediction includes:

[0095] S41: The input layer receives the learner’s knowledge fusion capability vector w;

[0096] S42: The knowledge encoding layer encodes the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector w to obtain the encoding vector of the knowledge point, where the knowledge point The encoding vector of :

[0097] ;

[0098] in:

[0099] represents the encoding matrix of 1 row and 2 columns in the optimal model parameters, That is, a 1-row, N-column encoding vector; the encoding vector The Nth column in the table is the Nth knowledge point. The probability of mastering the knowledge points The degree of credibility support;

[0100] S43: The confidence feature extraction layer extracts the encoding vector Extract the credibility features of knowledge points from The reliability characteristics are :

[0101] ;

[0102] in:

[0103] Represents the encoding vector The nth column in ;

[0104] represents the activation function;

[0105] S44: Information fusion layer for credibility features and the encoding vector Fusion is performed to obtain the mastery information features of knowledge points, where knowledge points The information characteristics of :

[0106] ;

[0107] S45: The knowledge point evaluation layer maps the mastery information features of the knowledge points to obtain the mastery degree prediction results of the knowledge points, where the knowledge points Characteristics of mastering information The mapping result is :

[0108] ;

[0109] in:

[0110] are two sets of training matrices in the optimal model parameters;

[0111] express function;

[0112] represents the L2 norm.

[0113] Optionally, in step S21, the learner's ability value is calculated based on the difficulty, discrimination and guessing parameters of the answering questions. ,include:

[0114] S211: Initialize learner capabilities , set the current iteration number of the capability to t, the maximum iteration number to Max, and the initial value of t to 0, Express ability The t-th iteration result of ;

[0115] S212: Constructing the log-likelihood function for estimating learner ability :

[0116] ;

[0117] S213: Calculated The corresponding gradient :

[0118] ;

[0119] S214: Combined Gradients right Iteration is performed, where the iteration formula is:

[0120] ;

[0121] S215: Set t = t + 1, return to step S213, until the maximum number of iterations is reached, and use the ability value obtained by the iteration at this time as the learner's ability value .

[0122] In order to solve the above problem, the present invention provides an electronic device, the electronic device comprising:

[0123] A memory storing at least one instruction;

[0124] Communication interface, enabling electronic equipment to communicate; and

[0125] The processor executes the instructions stored in the memory to implement the above-mentioned big data driven knowledge point positioning method.

[0126] 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, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned big data-driven knowledge point positioning method.

[0127] Compared with the prior art, the present invention proposes a knowledge point location method driven by big data, which has the following advantages:

[0128] First, this scheme proposes a method for constructing a knowledge point index vector, constructing a question bank containing all knowledge points, combining the frequency of occurrence of knowledge points in the question bank and the joint frequency of occurrence, calculating the importance of knowledge points and their correlation with other knowledge points, and selecting a question sequence containing knowledge points from the question bank for respondents to answer, and combining the respondents' answers to estimate the difficulty, discrimination and guessing parameters of the questions, and then characterizing the difficulty and discrimination of the knowledge points, and constructing a knowledge point index vector that characterizes the multi-dimensional information of the knowledge points.

[0129] At the same time, this scheme proposes a knowledge point positioning method. According to the learner's answering process, the learner's ability value and the effort characteristics that characterize the learner's ranking are extracted. The learner's mastery weight coefficients at different knowledge points are obtained through fusion calculation, including the learner's mastery probability at the knowledge point and the associated mastery coefficient. The knowledge point mastery evaluation model is used to encode the learner's mastery weight coefficients at different knowledge points, extract reliability features, and perform fusion mapping calculations to obtain the knowledge point mastery prediction results. The knowledge points whose knowledge point mastery prediction results are lower than the specified threshold are regarded as weak knowledge points for knowledge point positioning. In the process of solving the model parameters, the local optimum is jumped out through the reverse solution, and random number iteration is performed in combination with the iterative optimal solution and the mean of the solution to improve the traversal ability of the model parameters, and the iteration direction of the model parameters is controlled to quickly solve the optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0130] Figure 1 A flowchart of a big data driven knowledge point location method provided in accordance with an embodiment of the present invention.

[0131] 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

[0132] 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.

[0133] The embodiment of the present application provides a big data driven knowledge point positioning method. The execution subject of the big data driven knowledge point positioning method includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the big data driven knowledge point positioning method 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.

[0134] A big data driven knowledge point location method comprises the following steps:

[0135] S1: Construct a multi-dimensional knowledge point evaluation index system, combine the multi-dimensional knowledge point evaluation index system to evaluate and quantify the knowledge points, and obtain the knowledge point index vector, including:

[0136] Construct a multi-dimensional knowledge point evaluation index system and a question bank containing all knowledge points, wherein the multi-dimensional knowledge point evaluation index system includes knowledge point relevance, knowledge point difficulty, knowledge point distinction, knowledge point importance and knowledge point guessing factor. There are several questions in the question bank, each question contains at least one knowledge point, and the set of the knowledge points is :

[0137] ;

[0138] in:

[0139] Indicates the first Knowledge points, Represents the total number of knowledge points in the set;

[0140] Combined with the multi-dimensional knowledge point evaluation index system, the knowledge points are evaluated and quantified to obtain the knowledge point index vector, where the knowledge point The corresponding knowledge point index vector is , knowledge points The evaluation and quantification process is as follows:

[0141] S11: Calculate the knowledge points The correlation between other knowledge points in the knowledge point set is constructed to obtain the knowledge point The relevance of knowledge points :

[0142] ;

[0143] in:

[0144] Representing knowledge points And knowledge points The degree of correlation between

[0145] Indicates that the question bank contains knowledge points The number of topics, Indicates that it also contains knowledge points And knowledge points The number of questions;

[0146] Indicates the total number of questions in the question bank;

[0147] S12: Based on knowledge points The relevance of knowledge points , for knowledge points Calculate the importance of association and use the result of the calculation of the importance of association as the knowledge point Importance of knowledge points :

[0148] ;

[0149] in:

[0150] represents an exponential function with a natural constant as base;

[0151] Indicates the preset knowledge relevance threshold;

[0152] represents the smoothing parameter, Set to 2;

[0153] represents a symbolic function;

[0154] Indicates selection The maximum value in ;

[0155] S13: Select questions from the question bank and construct a question containing knowledge points The question sequence is D, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained. The mean difficulty, discrimination and guessing parameters of all questions are taken as the knowledge points. Difficulty of knowledge points , Knowledge point differentiation And the knowledge point guessing factor ;

[0156] S14: Construct knowledge points Knowledge point index vector , .

[0157] The step S13 constructs a knowledge point The question sequence is obtained, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained, including:

[0158] S131: Constructing knowledge points The title sequence :

[0159] ;

[0160] in:

[0161] Represents a sequence of questions The Road topic, , Indicates the total number of questions in the constructed question sequence;

[0162] S132: Get D answerers in the question sequence The answer results of the dth answerer in the question The answer result is , Indicates that the dth respondent answered the question correctly , Indicates that the dth respondent answered the question incorrectly , ;

[0163] S133: Construct the objective function of question information estimation based on the answer results of D respondents :

[0164] ;

[0165] in:

[0166] represents the vector to be estimated, represents the ability parameter vector of the respondent, Indicates the ability of the dth respondent; Represents a sequence of questions The discrimination vector of the questions in , Indicate the topic The discrimination degree of b represents the question sequence The difficulty vector of the question in , Indicate the topic Difficulty; Represents a sequence of questions The guess parameter vector for the question in , Indicate the topic Guessed parameters of

[0167] S134: The objective function is solved by using a gradient descent method to obtain the difficulty, discrimination and guessing parameters of each question.

[0168] S2: According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector, including:

[0169] Get the learner's answer questions and answer results as the learner's answer process :

[0170] ;

[0171] in:

[0172] Indicates the information of the kth question answered by the learner, including the difficulty of the question , Discrimination , guess parameters And the knowledge points included, Indicates the total number of questions answered;

[0173] represents the answer result of the learner in the kth question. Indicates that the learner answered the kth question correctly, Indicates that the learner answered the kth question incorrectly;

[0174] According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector. The interactive fusion processing process is as follows:

[0175] S21: Calculate the learner's ability value based on the difficulty, discrimination and guessing parameters of the answer questions, including:

[0176] S211: Initialize learner capabilities , set the current iteration number of the capability to t, the maximum iteration number to Max, and the initial value of t to 0, Express ability The tth iteration result of ;

[0177] S212: Constructing the log-likelihood function for estimating learner ability :

[0178] ;

[0179] in:

[0180] represents an intermediate variable;

[0181] S213: Calculated The corresponding gradient :

[0182] ;

[0183] S214: Combined Gradients right Iteration is performed, where the iteration formula is:

[0184] ;

[0185] S215: Set t = t + 1, return to step S213, until the maximum number of iterations is reached, and use the ability value obtained by the iteration at this time as the learner's ability value ;

[0186] S22: Based on ability value Rank the learners in reverse order to get the ranking results of the learners ; The higher the ability value, the lower the ranking;

[0187] S23: According to the ranking results Calculate the learner's effort characteristics :

[0188] ;

[0189] in:

[0190] represents the total number of learners;

[0191] S24: Combined with learners’ ability values and effort characteristics , decompose and fuse the knowledge point index vector of the knowledge point to obtain the learner's mastery weight coefficient at different knowledge points, where the learner's mastery weight coefficient at knowledge points The mastery weight coefficient is :

[0192] ;

[0193] in:

[0194] Indicates that learners are at the knowledge point The probability of mastering Indicates that you have fully mastered the knowledge points In the case of wrong answer The probability of the corresponding question;

[0195] Indicates ability value Answer the knowledge points correctly The probability of the corresponding question, Representing knowledge points The correct answer probability threshold;

[0196] Indicates that learners are at the knowledge point The correlation mastery coefficient, Indicates that K questions contain knowledge points The proportion of topics Indicates that K questions contain knowledge points The proportion of topics Representing knowledge points And knowledge points The relationship between Representing knowledge points And knowledge points The correlation between ;

[0197] T stands for transpose;

[0198] S25: The learner's mastery weight coefficients of all knowledge points form the learner's knowledge fusion ability vector w: .

[0199] S3: Construct a knowledge point mastery degree assessment model structure, and optimize the model parameters to obtain the optimal model parameters. According to the knowledge point mastery degree assessment model structure and the optimal model parameters, a knowledge point mastery degree assessment model is constructed, including:

[0200] Constructing a knowledge point mastery degree assessment model structure, wherein the knowledge point mastery degree assessment model structure includes an input layer, a knowledge encoding layer, a reliability feature extraction layer, an information fusion layer, and a knowledge point assessment layer;

[0201] The input layer is used to receive the learner's knowledge fusion ability vector;

[0202] The knowledge encoding layer is used to encode the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector;

[0203] The reliability feature extraction layer is used to extract the reliability features of knowledge points from the encoded mastery weight coefficients;

[0204] The information fusion layer is used to fuse the credibility features and the encoded mastery weight coefficients to obtain the mastery information features of the knowledge points;

[0205] The knowledge point evaluation layer is used to map the mastery information features of the knowledge points to obtain the predicted results of the mastery degree of the knowledge points, and regard the knowledge points whose predicted results of the mastery degree of the knowledge points are lower than the specified threshold as weak knowledge points;

[0206] The model parameters are optimized and solved in combination with the knowledge point mastery degree assessment model structure to obtain the optimal model parameters. A knowledge point mastery degree assessment model is constructed according to the knowledge point mastery degree assessment model structure and the optimal model parameters. The knowledge point mastery degree assessment model takes the learner's knowledge fusion ability vector as input and takes the learner's mastery degree prediction results at different knowledge points as output.

[0207] The method of optimizing the model parameters by combining the knowledge point mastery evaluation model structure to obtain the optimal model parameters includes:

[0208] S31: Obtain the learner knowledge fusion ability vectors and weak knowledge points of U learners to form the model parameter training data set data:

[0209] ;

[0210] in:

[0211] represents the learner knowledge fusion ability vector of the u-th learner, represents the weak knowledge point of the u-th learner;

[0212] S32: marking the model parameters to be optimized in the knowledge point mastery degree assessment model structure, wherein the model parameters to be optimized include the encoding matrix in the knowledge encoding layer and the training matrix in the knowledge point assessment layer;

[0213] S33: Initialize and generate Z groups of model parameters, wherein the zth group of model parameters generated by initialization is , , set the current iteration number of the model parameters to c, and the maximum iteration number to C, then the iteration result of the zth group of model parameters in the cth iteration is ,in Represents model parameters The encoding matrix in the corresponding knowledge encoding layer, Represents model parameters The two corresponding training matrices in ;

[0214] S34: Combine the model parameter training data set data and construct a fitness function with the model parameters as input, where the model parameters The fitness function of the input is :

[0215] ;

[0216] in:

[0217] represents the discriminant function, if If they are equal, the discriminant function is 1, otherwise the discriminant function is 0;

[0218] Indicates the use of model parameters Construct a knowledge point mastery evaluation model;

[0219] Represents the knowledge point mastery evaluation model The calculated weak points of the u-th learner;

[0220] S35: Substitute the Z groups of model parameters obtained in the cth iteration into the fitness function to obtain the fitness value of each group of model parameters, and use the model parameters with the highest fitness value as the optimal model parameters for the cth iteration ;

[0221] S36: Generate the inverse solution of the Z group of model parameters obtained in the cth iteration, where the zth group of model parameters The reverse solution is :

[0222] ;

[0223] in:

[0224] represents the lower bound of the model parameters, represents the upper bound of the model parameters;

[0225] like , then keep ,neglect , otherwise let ; Indicates the preset fitness threshold;

[0226] S37: Iterate the model parameters, where the iteration formula is:

[0227] ;

[0228] in:

[0229] represents the mean value of the Z group model parameters obtained in the cth iteration, Represents a random number between 0 and 1;

[0230] S38: Let c=c+1, return to step S35, until the maximum number of iterations is reached, and select the model parameter with the highest current fitness value as the optimal model parameter.

[0231] S4: Use the knowledge point mastery evaluation model to receive the learner's knowledge fusion ability vector and predict the knowledge point mastery. The knowledge points whose knowledge point mastery prediction results are lower than the specified threshold are regarded as weak knowledge points, and the knowledge points are located.

[0232] In step S4, the knowledge point mastery degree evaluation model is used to receive the learner's knowledge fusion ability vector and predict the knowledge point mastery degree, including:

[0233] S41: The input layer receives the learner’s knowledge fusion capability vector w;

[0234] S42: The knowledge encoding layer encodes the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector w to obtain the encoding vector of the knowledge point, where the knowledge point The encoding vector of :

[0235] ;

[0236] in:

[0237] represents the encoding matrix of 1 row and 2 columns in the optimal model parameters, That is, a 1-row, N-column encoding vector; the encoding vector The Nth column in the table is the Nth knowledge point. The probability of mastering the knowledge points The degree of credibility support;

[0238] S43: The confidence feature extraction layer extracts the encoding vector Extract the credibility features of knowledge points from The reliability characteristics are :

[0239] ;

[0240] in:

[0241] Represents the encoding vector The nth column in ;

[0242] represents an activation function; in the embodiment of the present invention, the selected activation function is a ReLU function;

[0243] S44: Information fusion layer for credibility features and the encoding vector Fusion is performed to obtain the mastery information features of knowledge points, where knowledge points The information characteristics of :

[0244] ;

[0245] S45: The knowledge point evaluation layer maps the mastery information features of the knowledge points to obtain the mastery degree prediction results of the knowledge points, where the knowledge points Characteristics of mastering information The mapping result is :

[0246] ;

[0247] in:

[0248] are two sets of training matrices in the optimal model parameters;

[0249] express function;

[0250] represents the L2 norm.

[0251] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0252] It should be noted that the serial numbers of the above-mentioned 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.

[0253] 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.

[0254] 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 big data driven knowledge point location method, characterized in that: The method comprises: S1: Construct a multi-dimensional knowledge point evaluation index system, combine the multi-dimensional knowledge point evaluation index system to evaluate and quantify the knowledge points, and obtain the knowledge point index vector; S2: According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector; S3: construct the knowledge point mastery degree assessment model structure, and optimize the model parameters to obtain the optimal model parameters, and construct the knowledge point mastery degree assessment model based on the knowledge point mastery degree assessment model structure and the optimal model parameters; The knowledge point mastery degree assessment model structure includes an input layer, a knowledge encoding layer, a reliability feature extraction layer, an information fusion layer and a knowledge point assessment layer; The input layer is used to receive the learner's knowledge fusion ability vector; The knowledge encoding layer is used to encode the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector; The reliability feature extraction layer is used to extract the reliability features of knowledge points from the encoded mastery weight coefficients; The information fusion layer is used to fuse the credibility features and the encoded mastery weight coefficients to obtain the mastery information features of the knowledge points; The knowledge point evaluation layer is used to map the mastery information features of the knowledge points to obtain the predicted results of the mastery degree of the knowledge points, and regard the knowledge points whose predicted results of the mastery degree of the knowledge points are lower than the specified threshold as weak knowledge points; S4: Use the knowledge point mastery evaluation model to receive the learner's knowledge fusion ability vector and predict the knowledge point mastery. The knowledge points whose knowledge point mastery prediction results are lower than the specified threshold are regarded as weak knowledge points, and the knowledge points are located.

2. A big data driven knowledge point location method as claimed in claim 1, characterized in that: The S1 step includes: Construct a multi-dimensional knowledge point evaluation index system and a question bank containing all knowledge points, wherein the multi-dimensional knowledge point evaluation index system includes knowledge point relevance, knowledge point difficulty, knowledge point distinction, knowledge point importance and knowledge point guessing factor. There are several questions in the question bank, each question contains at least one knowledge point, and the set of the knowledge points is : ； in: Indicates the first Knowledge points, Represents the total number of knowledge points in the set; Combined with the multi-dimensional knowledge point evaluation index system, the knowledge points are evaluated and quantified to obtain the knowledge point index vector, where the knowledge point The corresponding knowledge point index vector is , knowledge points The evaluation and quantification process is as follows: S11: Calculate the knowledge points The correlation between other knowledge points in the knowledge point set is constructed to obtain the knowledge point The relevance of knowledge points : ； in: Representing knowledge points And knowledge points The degree of correlation between Indicates that the question bank contains knowledge points The number of topics, Indicates that it also contains knowledge points And knowledge points The number of questions; Indicates the total number of questions in the question bank; S12: Based on knowledge points The relevance of knowledge points , for knowledge points Calculate the importance of association and use the result of the calculation of the importance of association as the knowledge point Importance of knowledge points : ； in: represents an exponential function with a natural constant as base; Indicates the preset knowledge relevance threshold; represents the smoothing parameter, Set to 2; represents a symbolic function; Indicates selection The maximum value in ; S13: Select questions from the question bank and construct a question containing knowledge points The question sequence is D, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained. The mean difficulty, discrimination and guessing parameters of all questions are taken as the knowledge points. Difficulty of knowledge points , Knowledge point differentiation And the knowledge point guessing factor ; S14: Construct knowledge points Knowledge point index vector , .

3. A big data driven knowledge point location method as claimed in claim 2, characterized in that: The step S13 constructs a knowledge point The question sequence is obtained, and the answer results of D respondents are collected. The question information of each question is estimated according to the answer results of the respondents, and the difficulty, discrimination and guessing parameters of each question are obtained, including: S131: Constructing knowledge points The title sequence : ； in: Represents a sequence of questions The Road topic, , Indicates the total number of questions in the constructed question sequence; S132: Get D answerers in the question sequence The answer results of the dth answerer are The answer result is , Indicates that the dth respondent answered the question correctly , Indicates that the dth respondent answered the question incorrectly , ; S133: Construct the objective function of question information estimation based on the answer results of D respondents : ； in: represents the vector to be estimated, represents the ability parameter vector of the respondent, Indicates the ability of the dth respondent; Represents a sequence of questions The discrimination vector of the questions in , Indicate the topic The discrimination degree of b represents the question sequence The difficulty vector of the question in Indicate the topic Difficulty; Represents a sequence of questions The guess parameter vector for the question in , Indicate the topic Guessed parameters of S134: The objective function is solved by using a gradient descent method to obtain the difficulty, discrimination and guessing parameters of each question.

4. A big data driven knowledge point location method as claimed in claim 2, characterized in that: The S2 step includes: Get the learner's answer questions and answer results as the learner's answer process : ； in: Indicates the information of the kth question answered by the learner, including the difficulty of the question , Discrimination , guess parameters And the knowledge points included, Indicates the total number of questions answered; represents the answer result of the learner in the kth question. Indicates that the learner answered the kth question correctly, Indicates that the learner answered the kth question incorrectly; According to the learner's answering process, the knowledge point indicator vector is interactively fused to obtain the learner's knowledge fusion ability vector. The interactive fusion processing process is as follows: S21: Calculate the learner's ability value based on the difficulty, discrimination and guessing parameters of the answer questions ; S22: Based on ability value Rank the learners in reverse order to get the ranking results of the learners ; The higher the ability value, the lower the ranking; S23: According to the ranking results Calculate the learner's effort characteristics : ； in: represents the total number of learners; S24: Combined with learners’ ability values and effort characteristics , decompose and fuse the knowledge point index vector of the knowledge point to obtain the learner's mastery weight coefficient at different knowledge points, where the learner's mastery weight coefficient at knowledge points The mastery weight coefficient is : ； in: Indicates that learners are at the knowledge point The probability of mastering Indicates that you have fully mastered the knowledge points In the case of wrong answer The probability of the corresponding question; Indicates ability value Answer the knowledge points correctly The probability of the corresponding question, Representing knowledge points The correct answer probability threshold; Indicates that learners are at the knowledge point The correlation mastery coefficient, Indicates that K questions contain knowledge points The proportion of topics Indicates that K questions contain knowledge points The proportion of topics Representing knowledge points And knowledge points The relationship between Representing knowledge points And knowledge points The degree of correlation between T stands for transpose; S25: The learner's mastery weight coefficients of all knowledge points form the learner's knowledge fusion ability vector w: .

5. A big data driven knowledge point location method as claimed in claim 4, characterized in that: The step of optimizing the model parameters to obtain the optimal model parameters includes: S31: Obtain the learner knowledge fusion ability vectors and weak knowledge points of U learners to form the model parameter training data set data: ； in: represents the learner knowledge fusion ability vector of the u-th learner, represents the weak knowledge point of the u-th learner; S32: marking the model parameters to be optimized in the knowledge point mastery degree assessment model structure, wherein the model parameters to be optimized include the encoding matrix in the knowledge encoding layer and the training matrix in the knowledge point assessment layer; S33: Initialize and generate Z groups of model parameters, wherein the zth group of model parameters generated by initialization is , , set the current iteration number of the model parameters to c, and the maximum iteration number to C, then the iteration result of the zth group of model parameters in the cth iteration is ,in Represents model parameters The encoding matrix in the corresponding knowledge encoding layer, Represents model parameters The two corresponding training matrices in ; S34: Combine the model parameter training data set data and construct a fitness function with the model parameters as input, where the model parameters The fitness function of the input is : ； in: represents the discriminant function, if If they are equal, the discriminant function is 1, otherwise the discriminant function is 0; Indicates the use of model parameters Construct a knowledge point mastery evaluation model; Represents the knowledge point mastery evaluation model The calculated weak points of the u-th learner; S35: Substitute the Z groups of model parameters obtained in the cth iteration into the fitness function to obtain the fitness value of each group of model parameters, and use the model parameters with the highest fitness value as the optimal model parameters for the cth iteration ; S36: Generate the inverse solution of the Z group of model parameters obtained in the cth iteration, where the zth group of model parameters The reverse solution is : ； in: represents the lower bound of the model parameters, represents the upper bound of the model parameters; like , then keep ,neglect , otherwise let ; Indicates the preset fitness threshold; S37: Iterate the model parameters, where the iteration formula is: ； in: represents the mean value of the Z group model parameters obtained in the cth iteration, Represents a random number between 0 and 1; S38: Let c=c+1, return to step S35, until the maximum number of iterations is reached, and select the model parameter with the highest current fitness value as the optimal model parameter.

6. A big data driven knowledge point location method as claimed in claim 5, characterized in that: In step S4, the knowledge point mastery degree evaluation model is used to receive the learner's knowledge fusion ability vector and predict the knowledge point mastery degree, including: S41: The input layer receives the learner’s knowledge fusion capability vector w; S42: The knowledge encoding layer encodes the mastery weight coefficients of different knowledge points in the learner's knowledge fusion ability vector w to obtain the encoding vector of the knowledge point, where the knowledge point The encoding vector of : ； in: represents the encoding matrix of 1 row and 2 columns in the optimal model parameters, That is, a 1-row, N-column encoding vector; the encoding vector The Nth column in the table is the Nth knowledge point. The probability of mastering the knowledge points The degree of credibility support; S43: The confidence feature extraction layer extracts the encoding vector Extract the credibility features of knowledge points from The reliability characteristics are : ； in: Represents the encoding vector The nth column in ; represents the activation function; S44: Information fusion layer for credibility features and the encoding vector Fusion is performed to obtain the mastery information features of knowledge points, where knowledge points The information characteristics of : ； S45: The knowledge point evaluation layer maps the mastery information features of the knowledge points to obtain the mastery degree prediction results of the knowledge points, where the knowledge points Characteristics of mastering information The mapping result is : ； in: are two sets of training matrices in the optimal model parameters; express function; represents the L2 norm.

7. A big data driven knowledge point location method as claimed in claim 4, characterized in that: In step S21, the learner's ability value is calculated based on the difficulty, discrimination and guessing parameters of the answering questions. ,include: S211: Initialize learner capabilities , set the current iteration number of the capability to t, the maximum iteration number to Max, and the initial value of t to 0, Express ability The tth iteration result of ; S212: Constructing the log-likelihood function for estimating learner ability : ； in: represents an intermediate variable; S213: Calculated The corresponding gradient : ； S214: Combined Gradients right Iteration is performed, where the iteration formula is: ； S215: Set t = t + 1, return to step S213, until the maximum number of iterations is reached, and use the ability value obtained by the iteration at this time as the learner's ability value .

Images