Resource scheduling method based on knowledge point incidence matrix

Abstract

The invention discloses a resource scheduling method based on a knowledge point incidence matrix. The method comprises the following steps: S1, constructing a user demand matrix and a resource pool matrix; s2, constructing a knowledge point incidence matrix; s3, performing resource matching in a resource pool according to user requirements; s4, performing resource selection on the matched resource pool according to the knowledge point incidence matrix; s5, updating the user demand matrix and the knowledge point incidence matrix based on feedback of historical scheduling information, and finally taking the updated matrix as an initial condition of a next scheduling cycle; from the perspective of knowledge points, research on a resource scheduling technology based on a knowledge point incidence matrix is provided, relevance of the knowledge points is fused into a resource scheduling process, and an active discovery network and a feedback network of user requirements are established in combination with a related feedback mechanism to continuously finely adjust and optimize the scheduling process. The resource scheduling efficiency and accuracy of the cloud VR virtual simulation teaching platform are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud computing, and in particular to a resource scheduling method based on a knowledge point association matrix. Background Art

[0002] Cloud VR virtual simulation teaching combines cloud computing, virtual reality and simulation technology. It is an innovative teaching model that provides students with an immersive learning experience. Students learn through resources such as virtual devices, virtual scenes and virtual teaching materials. These resources include 3D models, experimental scenes, teaching videos, interactive teaching materials, etc.

[0003] However, due to the richness and complexity of cloud VR teaching resources, how to reasonably and efficiently schedule resources has become a key issue. Existing resource scheduling methods mainly rely on user behavior data and simple resource usage frequency. The growing types and quantities of resources in cloud computing and the ever-changing user needs have brought great challenges to resource scheduling. One of them is how to find effective resources associated with knowledge points in a complex cloud computing environment. The current resource scheduling algorithm not only fails to fully consider the correlation between knowledge points, resulting in insufficient efficiency and accuracy of resource scheduling, but also lacks an effective feedback mechanism to dynamically adjust the resource scheduling strategy, making it difficult to meet personalized teaching needs. Summary of the invention

[0004] The purpose of the present invention is to provide a resource scheduling method based on a knowledge point association matrix in view of the deficiencies of the prior art.

[0005] The technical solution of the present invention to solve the above problem is: a resource scheduling method based on a knowledge point association matrix, comprising the following steps:

[0006] Step S1, constructing a user demand matrix and a resource pool matrix;

[0007] Step S2, constructing a knowledge point association matrix;

[0008] Step S3: Match resources in the resource pool according to user needs;

[0009] According to the demand values ​​in the user demand matrix, the resources in the resource pool are preliminarily screened to select the resources that meet the basic needs. Then, based on the screening results, the resources are sorted according to the demand values ​​to form a preliminary matching resource list, and the resources that meet the user needs are selected;

[0010] Step S4: selecting resources from the matched resource pool according to the knowledge point association matrix;

[0011] Step S5: Based on the feedback of historical scheduling information, the user demand matrix and the knowledge point association matrix are updated, and finally the updated matrix is ​​used as the initial condition for the next scheduling cycle.

[0012] Furthermore, step S1 specifically includes:

[0013] a) Build a user needs matrix based on the user’s learning needs;

[0014] The constructed user needs matrix is ​​as follows:

[0015] ur i =[ur i1 ,ur i2 ,ur i3 ,ur i4 ]

[0016] ur i1 ,ur i2 ,ur i3 ,ur i4 They are time requirements, security requirements, reliability requirements and trustworthiness requirements;

[0017] b) Construct a resource pool matrix based on existing teaching resources;

[0018] Each system resource is described by multiple resource features. Assuming the number of resource features is f, the defined resource pool is as follows:

[0019] rp j =[rp j1 ,rp j2 ,…,rp jf ].

[0020] Furthermore, the specific process of constructing the knowledge point association matrix in step S2 is as follows:

[0021] Step S2.1, score rate: calculate the user's score rate for each knowledge point;

[0022] Step S2.2, user interest: evaluate the user's interest in each knowledge point;

[0023] Step S2.3, knowledge point relevance: construct a relevance matrix based on the relevance between knowledge points;

[0024] Step S2.4, knowledge point importance: Use the knowledge point centrality formula to determine the importance of the knowledge point;

[0025] Step S2.5: Based on the teaching syllabus and expert knowledge, and the data collected from the user during the learning process, including the knowledge point score rate, the interest of the knowledge point, the relevance of the knowledge point, and the importance of the knowledge point, the knowledge point relevance matrix is ​​finally obtained as shown below:

[0026] kc i =[G i ,I i ,G i ,K i ]

[0027] G i ,I i ,C i ,K i They are the knowledge point scoring rate, knowledge point interest, knowledge point relevance and knowledge point importance.

[0028] Furthermore, the specific method of step S2.4 is: after improving the knowledge point relevance matrix using the knowledge point centrality formula, the importance distribution matrix of the knowledge points in each experimental step is obtained;

[0029] The formula of knowledge point centrality is as follows:

[0030]

[0031] where relation ij is the number of associations between knowledge point i and knowledge point j, and sum is the total number of knowledge points involved in the experimental steps.

[0032] Furthermore, the specific steps of step S3 are as follows:

[0033] a) Demand extraction: Extract the specific demand value of each user for each knowledge point from the user demand matrix; suppose that the demand value of user i for the resource in the kth dimension of knowledge point j is ur ik ;

[0034] b) Resource pool search: Use the resource pool matrix rp to find the resources that contain the knowledge points required by the user; each element of the resource pool matrix represents the degree to which resource i contains knowledge point j;

[0035] c) Matching resources: For each user requirement, filter out all resources that meet the requirement and form a preliminary matching resource list;

[0036] The resource matching formula is as follows:

[0037]

[0038] Candidate Resource Set CR i It is expressed as follows:

[0039] CR i = {r j ∣compare(ur i ,rp j )≥θ,rj ∈R,j=1,2,…,n}.

[0040] Furthermore, the specific method of step S4 is: comprehensively score each resource based on the user score rate, user interest, knowledge point relevance and knowledge point importance, re-rank the preliminary matching resources according to the comprehensive score, and give priority to resources with high comprehensive scores; the resource selection formula is expressed as follows:

[0041]

[0042] where g k ij Represents user task u i Resource requirements and system resources j The “gap” in the kth feature dimension, kc ik It represents the degree of the i-th user on the k-th dimension of the knowledge point matrix;

[0043] g k ij The calculation is expressed as follows:

[0044]

[0045] Furthermore, the specific steps of step S5 are as follows:

[0046] a) Collect feedback;

[0047] Collecting scheduling information from users’ learning behaviors and feedback, the scheduling information includes learning results and user feedback;

[0048] b) Update the user demand matrix;

[0049] c) Update the knowledge point association matrix;

[0050] The update formula of the user demand and knowledge point association matrix is ​​as follows:

[0051]

[0052] Among them, p(r s ) is r s The probability of being used as positive feedback can be calculated from the user's historical scheduling information; rp sk It means r s The gap between (the resources selected by the user in the resource selection stage) in the kth dimension.

[0053] The present invention has the beneficial effects:

[0054] The present invention provides a resource scheduling method based on the knowledge point association matrix. Starting from the perspective of knowledge points, the present invention proposes a research on resource scheduling technology based on the knowledge point association matrix, integrates the relevance of knowledge points into the resource scheduling process, and combines relevant feedback mechanisms to establish an active discovery network and feedback network of user needs to continuously fine-tune and optimize the scheduling process, so as to improve the resource scheduling efficiency and accuracy of the cloud VR virtual simulation teaching platform. DETAILED DESCRIPTION

[0055] A resource scheduling method based on a knowledge point association matrix comprises the following steps:

[0056] Step S1: construct a user demand matrix and a resource pool matrix.

[0057] a) Build a user needs matrix based on the user’s learning needs;

[0058] The constructed user needs matrix is ​​as follows:

[0059] ur i =[ur i1 ,ur i2 ,ur i3 ,ur i4 ]

[0060] Each element of the matrix represents the degree of demand of a certain user for a certain knowledge point. i1 ,ur i2 ,ur i3 ,ur i4 They are time requirements, security requirements, reliability requirements and trustworthiness requirements.

[0061] b) Construct a resource pool matrix based on existing teaching resources;

[0062] Each element of the matrix represents the knowledge point information contained in a resource. Each system resource is described by multiple resource features. Assuming the number of resource features is f, the defined resource pool is as follows:

[0063] rp j =[rp j1 ,rp j2 ,…,rp jf ]

[0064] Step S2: construct a knowledge point association matrix.

[0065] The knowledge point association matrix is ​​constructed by listing the influencing factors that may be involved in the association of knowledge points. The specific process of construction is as follows:

[0066] Step S2.1, score rate: calculate the user's score rate for each knowledge point.

[0067] The score rate describes the user's score on the experimental step. If the experiment passes, the score rate of this step is 1, otherwise it is 0.

[0068] Step S2.2: User interest: Evaluate the user's interest in each knowledge point.

[0069] The interest level describes the user's interest in the experimental step. If the user is interested in the knowledge point of the step, the interest level of the step is 1, otherwise it is 0.

[0070] Step S2.3, knowledge point correlation: construct a correlation matrix based on the correlation between knowledge points.

[0071] The experimental steps are broken down into multiple knowledge points. If the knowledge point is examined in the step, it is recorded as 1, otherwise it is recorded as 0.

[0072] Step S2.4, knowledge point importance: Use the knowledge point centrality formula to determine the importance of the knowledge point.

[0073] The more connections between knowledge points, the more important the knowledge point is; conversely, the fewer connections, the less important it is. The knowledge point centrality formula describes the degree of connection and importance between each knowledge point. After improving the knowledge point association matrix using the knowledge point centrality formula, the knowledge point importance distribution matrix in each experimental step is obtained.

[0074] The formula of knowledge point centrality is as follows:

[0075]

[0076] where relation ij is the number of associations between knowledge point i and knowledge point j, and sum is the total number of knowledge points involved in the experimental steps.

[0077] Step S2.5: Based on the teaching syllabus and expert knowledge, and the data collected from the user during the learning process, including the knowledge point score rate, the interest of the knowledge point, the relevance of the knowledge point, and the importance of the knowledge point, the knowledge point relevance matrix is ​​finally obtained as shown below:

[0078] kc i =[G i ,I i ,G i ,K i ]

[0079] G i ,I i ,C i ,K i They are the knowledge point scoring rate, knowledge point interest, knowledge point relevance and knowledge point importance.

[0080] Step S3: Match resources in the resource pool according to user needs.

[0081] According to the demand values ​​in the user demand matrix, the resources in the resource pool are preliminarily screened to select the resources that meet the basic needs. Then, based on the screening results, the resources are sorted according to the demand values ​​to form a preliminary matching resource list, and the resources that meet the user needs are selected.

[0082] The specific steps are as follows:

[0083] c) Demand extraction: Extract the specific demand value of each user for each knowledge point from the user demand matrix. Assume that the demand value of user i for knowledge point j on the kth dimension resource is ur ik .

[0084] d) Resource pool search: Use the resource pool matrix rp to find the resources that contain the knowledge points required by the user. Each element of the resource pool matrix represents the degree to which resource i contains knowledge point j.

[0085] c) Matching resources: For each user requirement, filter out all resources that meet the requirement and form a preliminary matching resource list.

[0086] The resource matching formula is as follows:

[0087]

[0088] Use the cosine similarity formula to calculate the given user task u i (u i ∈T, 1≤i≤m), then the resource requirement of the user task is equal to any resource r in the resource set j (r j ∈R, 1≤j≤n) features, T is the task set, the resource set consists of m independent meta-tasks; R is the resource set, and the cloud system used can be abstracted into n heterogeneous resources. i , rp j ) is the user task u i Resource requirements and resource j Similarity in feature description; ik For user task u i The resource requirement of the kth dimension, ur ik With rp jk If the calculated similarity is greater than the given matching threshold θ, then resource r is considered j The characteristics and user tasks u i The resource requirements are similar, and all resources in the resource set R are repeatedly calculated, and finally returned to the user task u i, the matching threshold θ is the best matching value given manually to determine whether the matching resources are suitable. A candidate resource set CR i It is expressed as follows:

[0089] CR i = {r j ∣compare(ur i ,rp j )≥θ,r j ∈R,j=1,2,…,n}

[0090] Step S4: Select resources from the matched resource pool according to the knowledge point association matrix.

[0091] On the basis of preliminary matching resources, the knowledge point association matrix is ​​further used to optimize the selection of resources. Combined with the user score rate, user interest, knowledge point association and knowledge point importance, each resource is comprehensively scored, and the preliminary matching resources are re-ranked according to the comprehensive score, giving priority to resources with high comprehensive scores. The resource selection formula is expressed as follows:

[0092]

[0093] where g k ij Represents user task u i Resource requirements and system resources j The “gap” in the kth feature dimension, kc ik It represents the degree of the i-th user on the k-th dimension of the knowledge point matrix.

[0094] User Task i Resource requirements and system resources j The smaller the "gap", the closer the characteristics of the resource are to the requirements of the user's task, and the task should be assigned to such resources, and the resources should have a higher priority.

[0095] To this end, given g k ij The calculation is expressed as follows:

[0096]

[0097] In the calculation of this "gap", rp is directly used jk and ik Subtract. If rp jk -ur ik >0, it means that the resources can meet the resource requirements of the user task, and the "gap" is directly reflected as the difference between the two numbers, i.e., rp jk -ur ik ;

[0098] If rp jk -ur ik <0, it means that the resource cannot meet the resource requirements of the user task. At this time, if there are other resources that can meet the user's needs, those resources should be considered first. Therefore, the absolute value of these negative differences is taken and added to the maximum absolute value of the difference in this dimension, so that the processed value must be greater than the difference of the original positive difference resources, ensuring that the priority of the resources with negative differences is lower than the priority of the resources with positive differences. Moreover, in the case that there are no resources that meet the user's needs, the values ​​with negative differences are still processed in this way. The smaller the difference, the greater the priority, and the greater the possibility of being selected for operation;

[0099] In rp jk -ur ik =0, in order to avoid the denominator of the benefit function being 0, the minimum absolute value of all differences in this dimension is taken to ensure that this situation has the highest priority.

[0100] Step S5: Based on the feedback of historical scheduling information, update the user demand matrix and the knowledge point association matrix, and finally use the updated matrix as the initial condition for the next scheduling cycle. The specific steps are as follows:

[0101] a) Collect feedback

[0102] Gather scheduling information from user learning behaviors and feedback, including:

[0103] 1. Learning results: The learning results of users after using recommended resources, such as test scores, learning progress, etc.

[0104] 2. User feedback: users’ subjective evaluation and feedback on recommended resources, such as satisfaction ratings, usage experience, etc.

[0105] b) Update the user requirements matrix.

[0106] c) Update the knowledge point association matrix.

[0107] In step S5 of the above solution, the user demand matrix and the knowledge point association matrix are dynamically adjusted through the feedback of historical scheduling information to optimize the resource scheduling effect.

[0108] The update formula of the user demand and knowledge point association matrix is ​​as follows:

[0109]

[0110] Among them, p(r s ) is r s The probability of being used as positive feedback can be calculated from the user's historical scheduling information; rpsk It means r s The gap between (the resources selected by the user in the resource selection stage) in the kth dimension.

[0111] The updated task requirement matrix and task preference matrix will serve as the initial conditions for the next scheduling cycle to realize the active discovery and collection of user demand information. Moreover, this feedback process will be iteratively updated to make the subsequent scheduling process more in line with the true intention of the inquirer and realize the automatic adjustment of user needs.

[0112] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A resource scheduling method based on a knowledge point association matrix, characterized in that: The following steps are involved: Step S1, constructing a user demand matrix and a resource pool matrix; Step S2, constructing a knowledge point association matrix; Step S3: Match resources in the resource pool according to user needs; According to the demand values ​​in the user demand matrix, the resources in the resource pool are preliminarily screened to select the resources that meet the basic needs. Then, based on the screening results, the resources are sorted according to the demand values ​​to form a preliminary matching resource list, and the resources that meet the user needs are selected; Step S4: selecting resources from the matched resource pool according to the knowledge point association matrix; Step S5: Based on the feedback of historical scheduling information, the user demand matrix and the knowledge point association matrix are updated, and finally the updated matrix is ​​used as the initial condition for the next scheduling cycle.

2. A resource scheduling method based on a knowledge point association matrix as claimed in claim 1, characterized in that: Step S1 specifically includes: a) Build a user needs matrix based on the user’s learning needs; The constructed user needs matrix is ​​as follows: clock i =[clock i1 ,clock i2 ,clock i3 ,clock i4 ] ur i1 ,ur i2 ,ur i3 ,ur i4 They are time requirements, security requirements, reliability requirements and trustworthiness requirements; b) Construct a resource pool matrix based on existing teaching resources; Each system resource is described by multiple resource features. Assuming the number of resource features is f, the defined resource pool is as follows: rp j =[rp j1 ,rp j2 ,…,rp jf ]。 3. A resource scheduling method based on a knowledge point association matrix as claimed in claim 1, characterized in that: The specific process of constructing the knowledge point association matrix in step S2 is as follows: Step S2.1, score rate: calculate the user's score rate for each knowledge point; Step S2.2, user interest: evaluate the user's interest in each knowledge point; Step S2.3, knowledge point relevance: construct a relevance matrix based on the relevance between knowledge points; Step S2.4, knowledge point importance: Use the knowledge point centrality formula to determine the importance of the knowledge point; Step S2.5: Based on the teaching syllabus and expert knowledge, and the data collected from the user during the learning process, including the knowledge point score rate, the interest of the knowledge point, the relevance of the knowledge point, and the importance of the knowledge point, the knowledge point relevance matrix is ​​finally obtained as shown below: kc i =[G i ,I i ,G i ,K i ] G i ,I i ,C i ,K i They are the knowledge point scoring rate, knowledge point interest, knowledge point relevance and knowledge point importance.

4. A resource scheduling method based on a knowledge point association matrix as claimed in claim 3, characterized in that: The specific method of step S2.4 is: after improving the knowledge point relevance matrix using the knowledge point centrality formula, the importance distribution matrix of the knowledge points in each experimental step is obtained; The formula of knowledge point centrality is as follows: where relation ij is the number of associations between knowledge point i and knowledge point j, and sum is the total number of knowledge points involved in the experimental steps.

5. A resource scheduling method based on a knowledge point association matrix as claimed in claim 1, characterized in that: The specific steps of step S3 are as follows: a) Demand extraction: Extract the specific demand value of each user for each knowledge point from the user demand matrix; suppose that the demand value of user i for the resource in the kth dimension of knowledge point j is ur ik ; b) Resource pool search: Use the resource pool matrix rp to find the resources that contain the knowledge points required by the user; each element of the resource pool matrix represents the degree to which resource i contains knowledge point j; c) Matching resources: For each user requirement, filter out all resources that meet the requirement and form a preliminary matching resource list; The resource matching formula is as follows: Candidate Resource Set CR i It is expressed as follows: CR i ={r j ∣compare(ur i ,rp j )≥θ,r j ∈R,j=1,2,…,n}。 6. A resource scheduling method based on a knowledge point association matrix as claimed in claim 1, characterized in that: The specific method of step S4 is: comprehensively score each resource based on the user score rate, user interest, knowledge point relevance and knowledge point importance, re-rank the preliminary matching resources according to the comprehensive score, and give priority to resources with high comprehensive scores; The resource selection formula is expressed as follows: where g k ij Represents user task u i Resource requirements and system resources j The "gap" in the kth feature dimension, kc ik It represents the degree of the i-th user on the k-th dimension of the knowledge point matrix; g k ij The calculation is expressed as follows:

7. A resource scheduling method based on a knowledge point association matrix as claimed in claim 1, characterized in that: The specific steps of step S5 are as follows: a) Collect feedback; Collecting scheduling information from users’ learning behaviors and feedback, the scheduling information includes learning results and user feedback; b) Update the user demand matrix; c) Update the knowledge point association matrix; The update formula of the user demand and knowledge point association matrix is ​​as follows: Among them, p(r s ) is r s The probability of being used as positive feedback can be calculated from the user's historical scheduling information; rp sk It means r s The gap between (the resources selected by the user in the resource selection stage) in the kth dimension.

Images