A method for generating optimal learning path based on learning situation model
Through the optimal learning path generation method based on the academic situation model, the academic situation analysis model is constructed and a comprehensive academic situation distribution is generated, which solves the problem that the existing online learning platform cannot achieve personalized learning, realizes the generation of personalized learning paths, and improves improve learning efficiency.
Patent Information
- Application Number
- CN202411463659.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2025-01-24
- Estimated Expiration
- 2044-10-21
AI Technical Summary
The existing online learning platforms lack intelligent learning path planning capabilities, making them difficult to achieve truly personalized learning, and cannot effectively respond to the diverse learning needs of users of different backgrounds.
Adopt the best learning path generation method based on the learning model. By obtaining the user's academic information and the existing learning list, to build a learning analysis model, generate the initial academic distribution value, and use the comprehensive academic sentiment distribution and the existing learning list with the existing learning list. The matching of the best learning path is obtained.
Realize personalized learning, meet the differentiated needs of different users, improve learning efficiency, and reduce ineffective learning time.
Smart Images
Figure CN118967397B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of data analysis, and specifically relates to a method for generating an optimal learning path based on a learning situation model. Background Art
[0002] With the rapid development of Internet technology and the continuous improvement of the public's learning concept, online learning activities using the Internet as a medium have become a very common social activity. However, the accompanying massive knowledge and data have also brought serious information overload problems to the field of online education. How to provide personalized learning solutions for users from different backgrounds has become an important topic in online learning activities. The traditional one-size-fits-all teaching model can no longer meet the diverse learning needs of users. Although some existing learning platforms provide a large number of learning resources, they lack intelligent learning path planning capabilities and are difficult to achieve truly personalized learning. Summary of the invention
[0003] In order to solve the above problems, the present invention proposes an optimal learning path generation method based on a learning situation model.
[0004] The technical solution of the present invention is: a method for generating an optimal learning path based on a learning situation model comprises the following steps:
[0005] S1. Obtain the user's learning information and existing learning list;
[0006] S2. Construct a learning situation analysis model, input the user's learning situation information into the learning situation analysis model, generate an initial learning situation distribution value, and obtain a comprehensive learning situation distribution through the initial learning situation distribution value;
[0007] S3. Match the comprehensive learning situation distribution with the existing learning list to obtain the optimal learning path.
[0008] The user's learning information records the courses that the user has studied, and the current learning list records the names of existing courses.
[0009] Furthermore, S2 includes the following sub-steps:
[0010] S21, extracting a high-frequency field set from the user's learning information;
[0011] S22, based on the high-frequency field set, extracting the field with the largest academic search value from the user's academic information;
[0012] S23, constructing a field operator for each field in the learning information;
[0013] S24, constructing a learning situation distribution model, inputting the field operator corresponding to the field of the maximum learning situation search value into the learning situation distribution model, and obtaining an initial learning situation distribution value;
[0014] S25. Obtain the comprehensive learning situation distribution according to the initial learning situation distribution value and the field operators of the remaining fields.
[0015] The beneficial effect of the above further scheme is that in the present invention, in S21, the stop words of the learning information can be eliminated to reduce the influence of irrelevant words on the extraction of high-frequency field sets. When extracting high-frequency field sets, the words ranked first in the frequency order can be extracted. , N represents the number of fields in the learning information. Based on the high-frequency field set, the argmax function is used to filter the field with the largest learning search value from the learning information. Then, the field operator is constructed for it using the parameters such as the word frequency and inverse text frequency of the field itself, and the initial learning distribution value of the field with the largest learning search value is determined through the learning distribution model. Finally, the field operators of the remaining fields are combined with the field operators of the special field, and the spectral radius is used to filter the appropriate fields as the comprehensive learning distribution. After the above process, the field that best reflects the user's characteristics can be selected from the user's learning information to match the existing learning list.
[0016] Furthermore, in S22, the calculation formula of the field i with the maximum academic sentiment search value is: ; In the formula, I represents the set of all fields of learning information, represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of field m in the high-frequency field set, where M represents the high-frequency field set. represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of the field with the minimum word frequency in the learning information, β i represents the weight of field i, and c represents a constant.
[0017] If field i appears in the high-frequency field set, then β i =1, otherwise β i =0
[0018] Furthermore, in S23, the expression of the field operator σ of the field is: ; In the formula, y represents the word embedding vector of the field, N represents the number of fields in the learning information, T represents the inverse document frequency of the field, and F represents the word frequency of the field in the learning information.
[0019] Furthermore, in S24, the expression of the learning situation distribution model Q is: ; In the formula, σ i The field operator corresponding to field i representing the maximum search value, R i represents the spectral radius of the field operator corresponding to field i of the maximum knowledge search value, ln(·) represents the logarithmic function, Indicates the maximum field operator value of the remaining fields in the learning information. Indicates the minimum field operator value of the remaining fields in the learning information.
[0020] Each field operator is a matrix, and the maximum matrix value is selected as , select the minimum matrix value as .
[0021] Further, S25 includes the following sub-steps:
[0022] S251, combining the field operator of every three fields of the learning information with the field operator corresponding to the field of the maximum learning search value to obtain a plurality of local distribution operators;
[0023] S252, deleting the fields whose spectral radius of the local distribution operator is smaller than the initial learning situation distribution value from the learning situation information;
[0024] S253. Generate a comprehensive learning situation distribution in the remaining fields.
[0025] In S251, for example, the first field, the second field, and the third field in the learning information are connected to the field of the maximum learning search value to obtain a local distribution operator Z, which is expressed as: , where σ1 represents the field operator of the first field, σ2 represents the field operator of the second field, σ3 represents the field operator of the third field, and σ i The field operator of field i representing the maximum academic information search value. If the number of remaining fields in the academic information is not a multiple of 3, the missing fields are filled with the unit matrix. If the spectral radius of the local distribution operator Z is smaller than the initial academic information distribution value, the first field, the second field, and the third field are deleted from the academic information.
[0026] Furthermore, in S253, the method for generating the comprehensive learning situation distribution is: taking the intersection of the remaining fields and the high-frequency field set as the comprehensive learning situation distribution.
[0027] Furthermore, in S3, all learning names in the existing learning list whose similarity with the comprehensive learning situation distribution is greater than a threshold are taken as the optimal learning path.
[0028] The threshold can be determined artificially.
[0029] The beneficial effects of the present invention are: the optimal learning path generation method constructs a field operator for the field of the user's learning situation information, then uses the learning situation distribution model to generate an initial learning situation distribution value, and then screens the comprehensive learning situation distribution from the learning situation information based on characteristic parameters such as the spectral radius of the field operator; uses the matching results of the comprehensive learning situation distribution and the existing learning list to generate the optimal learning path, so that users can achieve truly personalized learning, meet the differentiated needs of different users, improve learning efficiency, and reduce ineffective learning time. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 Flowchart of the method for generating the optimal learning path based on the learning situation model. DETAILED DESCRIPTION
[0031] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings.
[0032] like Figure 1 As shown, the present invention provides a method for generating an optimal learning path based on a learning situation model, comprising the following steps:
[0033] S1. Obtain the user's learning information and existing learning list;
[0034] S2. Construct a learning situation analysis model, input the user's learning situation information into the learning situation analysis model, generate an initial learning situation distribution value, and obtain a comprehensive learning situation distribution through the initial learning situation distribution value;
[0035] S3. Match the comprehensive learning situation distribution with the existing learning list to obtain the optimal learning path.
[0036] The user's learning information records the courses that the user has studied, and the current learning list records the names of existing courses.
[0037] In this embodiment of the present invention, S2 includes the following sub-steps:
[0038] S21, extracting a high-frequency field set from the user's learning information;
[0039] S22, based on the high-frequency field set, extracting the field with the largest academic search value from the user's academic information;
[0040] S23, constructing a field operator for each field in the learning information;
[0041] S24, constructing a learning situation distribution model, inputting the field operator corresponding to the field of the maximum learning situation search value into the learning situation distribution model, and obtaining an initial learning situation distribution value;
[0042] S25. Obtain the comprehensive learning situation distribution according to the initial learning situation distribution value and the field operators of the remaining fields.
[0043] In the present invention, in S21, stop words in the learning information can be removed to reduce the influence of irrelevant words on the extraction of high-frequency field sets. When extracting high-frequency field sets, the words ranked first in frequency can be extracted. , N represents the number of fields in the learning information. Based on the high-frequency field set, the argmax function is used to filter the field with the largest learning search value from the learning information. Then, the field operator is constructed for it using the parameters such as the word frequency and inverse text frequency of the field itself, and the initial learning distribution value of the field with the largest learning search value is determined through the learning distribution model. Finally, the field operators of the remaining fields are combined with the field operators of the special field, and the spectral radius is used to filter the appropriate fields as the comprehensive learning distribution. After the above process, the field that best reflects the user's characteristics can be selected from the user's learning information to match the existing learning list.
[0044] In the embodiment of the present invention, in S22, the calculation formula of the field i with the maximum sentiment search value is: ; In the formula, I represents the set of all fields of learning information, represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of field m in the high-frequency field set, where M represents the high-frequency field set. represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of the field with the minimum word frequency in the learning information, β i represents the weight of field i, c represents a constant, and argmax(·) represents a function that finds the maximum value of the independent variable point set.
[0045] If field i appears in the high-frequency field set, then β i =1, otherwise β i =0
[0046] In the embodiment of the present invention, in S23, the expression of the field operator σ of the field is: ; In the formula, y represents the word embedding vector of the field, N represents the number of fields in the learning information, T represents the inverse document frequency of the field, and F represents the word frequency of the field in the learning information.
[0047] In the embodiment of the present invention, in S24, the expression of the learning situation distribution model Q is: ; In the formula, σ i The field operator corresponding to field i representing the maximum search value, R i represents the spectral radius of the field operator corresponding to field i of the maximum knowledge search value, ln(·) represents the logarithmic function, Indicates the maximum field operator value of the remaining fields in the learning information. Indicates the minimum field operator value of the remaining fields in the learning information.
[0048] Each field operator is a matrix, and the maximum matrix value is selected as , select the minimum matrix value as .
[0049] In this embodiment of the present invention, S25 includes the following sub-steps:
[0050] S251, combining the field operator of every three fields of the learning information with the field operator corresponding to the field of the maximum learning search value to obtain a plurality of local distribution operators;
[0051] S252, deleting the fields whose spectral radius of the local distribution operator is smaller than the initial learning situation distribution value from the learning situation information;
[0052] S253. Generate a comprehensive learning situation distribution in the remaining fields.
[0053] In S251, for example, the first field, the second field, and the third field in the learning information are connected to the field of the maximum learning search value to obtain a local distribution operator Z, which is expressed as: , where σ1 represents the field operator of the first field, σ2 represents the field operator of the second field, σ3 represents the field operator of the third field, and σ i The field operator of field i representing the maximum academic information search value. If the number of remaining fields in the academic information is not a multiple of 3, the missing fields are filled with the unit matrix. If the spectral radius of the local distribution operator Z is smaller than the initial academic information distribution value, the first field, the second field, and the third field are deleted from the academic information.
[0054] In an embodiment of the present invention, in S253, the method for generating a comprehensive learning situation distribution is: taking the intersection of the remaining fields and the high-frequency field set as the comprehensive learning situation distribution.
[0055] In the embodiment of the present invention, in S3, all learning names in the existing learning list whose similarities with the comprehensive learning situation distribution are greater than a threshold are taken as the optimal learning path.
[0056] The threshold can be determined artificially.
[0057] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.
Claims
1. A method for generating an optimal learning path based on a learning situation model, characterized in that: The following steps are involved: S1. Obtain the user's learning information and existing learning list; S2. Construct a learning situation analysis model, input the user's learning situation information into the learning situation analysis model, generate an initial learning situation distribution value, and obtain a comprehensive learning situation distribution through the initial learning situation distribution value; S3, matching the comprehensive learning situation distribution with the existing learning list to obtain the optimal learning path; The S2 comprises the following sub-steps: S21, extracting a high-frequency field set from the user's learning information; S22, based on the high-frequency field set, extracting the field with the largest academic search value from the user's academic information; S23, constructing a field operator for each field in the learning information; S24, constructing a learning situation distribution model, inputting the field operator corresponding to the field of the maximum learning situation search value into the learning situation distribution model, and obtaining an initial learning situation distribution value; S25. Obtain comprehensive learning situation distribution according to the initial learning situation distribution value and the field operators of the remaining fields; In S22, the calculation formula of the field i with the maximum academic search value is: In the formula, I represents the set of all fields of learning information, represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of field m in the high-frequency field set, M represents the high-frequency field set, and s′ i represents the Euclidean distance between the word embedding vector of field i and the word embedding vector of the field with the minimum word frequency in the learning information, β i represents the weight of field i, c represents a constant, and argmax(·) represents a function for finding the maximum value of the independent variable point set; In S23, the expression of the field operator σ of the field is: In the formula, y represents the word embedding vector of the field, N represents the number of fields in the learning information, T represents the inverse document frequency of the field, and F represents the word frequency of the field in the learning information; In S24, the expression of the learning situation distribution model Q is: In the formula, σ i The field operator corresponding to field i representing the maximum search value, R i represents the spectral radius of the field operator corresponding to field i with the maximum academic information search value, ln(·) represents the logarithmic function, σ″ represents the maximum field operator value of the remaining fields in the academic information, and σ′ represents the minimum field operator value of the remaining fields in the academic information; The S25 comprises the following sub-steps: S251, combining the field operator of every three fields of the learning information with the field operator corresponding to the field of the maximum learning search value to obtain a plurality of local distribution operators; S252, deleting the fields whose spectral radius of the local distribution operator is smaller than the initial learning situation distribution value from the learning situation information; S253, generating a comprehensive learning situation distribution in the remaining fields; In S253, the method for generating the comprehensive learning situation distribution is: taking the intersection of the remaining fields and the high-frequency field set as the comprehensive learning situation distribution.
2. The optimal learning path generation method based on the learning situation model according to claim 1 is characterized in that: In S3, all learning names in the existing learning list whose similarities with the comprehensive learning situation distribution are greater than a threshold are taken as the optimal learning path.
Citation Information
Patent Citations
System and method for a computerized learning system
CA2730456A1
Explanatable learning path recommendation method fusing learning scene
CN118297154A