Learning Path Recommendation Method and Device Based on Multi-Level Knowledge Graph

Abstract

The present invention discloses a learning path recommendation method and device based on a multi-level knowledge graph, relating to the technical field of learning path recommendation, including: constructing a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and there are predecessor relationships, successor relationships, and ancestor relationships between at least some of the knowledge points; constructing the shortest learning path or the complete learning path based on the multi-level knowledge graph. This application can separate different knowledge points, aggregate similar knowledge points, and has a more reasonable organizational structure.

Description

Technical Field

[0001] The present invention belongs to the field of learning path recommendation, and particularly relates to a learning path recommendation method and device based on a multi-level knowledge graph. Background Art

[0002] With the continuous development of science and technology, many online education platforms have emerged, and learning using online education platforms has become a new choice for many people.

[0003] With the continuous popularization of online education platforms, the learning path recommendation algorithms therein have also attracted extensive attention from researchers and become a new application direction. However, the recommendation of learning paths is different from the recommendations in other fields. The recommendation of learning paths needs to satisfy the successive relationship between knowledge points and recommend a reasonable learning order of knowledge points. In addition, for different users, everyone's available learning time is different, resulting in poor applicability of the existing learning path recommendations.

[0004] Therefore, there is an urgent need for a learning path recommendation method that can recommend a suitable learning path for users according to their available learning time. Summary of the Invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a learning path recommendation method and device based on a multi-level knowledge graph. The technical problems to be solved by the present invention are realized through the following technical solutions:

[0006] In a first aspect, the present application provides a learning path recommendation method based on a multi-level knowledge graph, including:

[0007] Construct a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and at least some of the knowledge points include a predecessor relationship, a successor relationship, and an ancestor relationship;

[0008] Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point and does not designate the starting knowledge point, construct the shortest learning path or the complete learning path;

[0009] Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and the starting knowledge point is path-connected to the target knowledge point, construct the shortest learning path or the complete learning path;

[0010] Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point and does not designate the starting knowledge point, construct the shortest learning path or the complete learning path;

[0011] Based on a multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and are connected by a path, construct the shortest learning path or the complete learning path;

[0012] Based on a multi-level knowledge graph, when the user designates a certain secondary knowledge point as the starting knowledge point and does not designate a target knowledge point, construct the shortest learning path or the complete learning path;

[0013] Based on a multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, construct the shortest learning path or the complete learning path;

[0014] Among them, the primary knowledge points include multiple secondary knowledge points.

[0015] In a second aspect, the present application also provides a learning path recommendation device based on a multi-level knowledge graph, including:

[0016] A construction module for constructing a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and at least some of the knowledge points include a predecessor relationship, a successor relationship, and an ancestor relationship;

[0017] A generation module for, based on the multi-level knowledge graph, when the user designates a certain primary knowledge point as the target knowledge point and does not designate a starting knowledge point, construct the shortest learning path or the complete learning path; based on the multi-level knowledge graph, when the user designates a certain primary knowledge point as the target knowledge point, designates another primary knowledge point as the starting knowledge point, and the starting knowledge point and the target knowledge point are connected by a path, construct the shortest learning path or the complete learning path; based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point and does not designate a starting knowledge point, construct the shortest learning path or the complete learning path; based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and are connected by a path, construct the shortest learning path or the complete learning path; based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the starting knowledge point and does not designate a target knowledge point, construct the shortest learning path or the complete learning path; based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, construct the shortest learning path or the complete learning path; among them, the primary knowledge points include multiple secondary knowledge points.

[0018] In a third aspect, the present application further provides an electronic device, including: a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the method described in any one of claims 1 to 8 is implemented.

[0019] Advantages of the present invention:

[0020] A learning path recommendation method and device based on a multi-level knowledge graph provided by the present invention address the problems of cold start and inconsistent recommended content with the sequence law of knowledge points in the learning path recommendation algorithm. The learning path recommendation method based on the multi-level knowledge graph of the present application can make recommendations based on the user's available time without relying on the user's learning history. The multi-level knowledge graph can meet the coarse-grained and fine-grained display of knowledge points. Compared with a single-level knowledge graph, the multi-level knowledge graph can separate different knowledge points, aggregate similar knowledge points, and has a more reasonable organizational structure.

[0021] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Description of the Drawings

[0022] Figure 1 is a flowchart of a learning path recommendation method based on a multi-level knowledge graph provided by an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of the relationship between knowledge points provided by an embodiment of the present invention;

[0024] Figure 3 is another schematic diagram of the relationship between knowledge points provided by an embodiment of the present invention;

[0025] Figure 4 is a schematic diagram of a shortest learning path generation algorithm provided by an embodiment of the present invention;

[0026] Figure 5 is a schematic diagram of a complete learning path generation algorithm provided by an embodiment of the present invention;

[0027] Figure 6 is a schematic diagram of constructing a shortest learning path provided by an embodiment of the present invention;

[0028] Figure 7 is a schematic diagram of constructing a complete learning path provided by an embodiment of the present invention;

[0029] Figure 8 is a flowchart of a shortest learning path provided by an embodiment of the present invention;

[0030] Figure 9 is a flowchart of a complete learning path provided by an embodiment of the present invention. Detailed implementation manners

[0031] The present invention will be further described in detail below in conjunction with specific embodiments, but the implementation manners of the present invention are not limited thereto.

[0032] In the prior art, learning path recommendation includes methods of collaborative filtering and methods based on heuristic algorithms. Collaborative filtering, as a commonly used method in recommendation algorithms, has been widely applied in learning path recommendation. Among them, Vesin (Vesin B, A M et al. Applied recommender systems and adaptive hypermedia for e-learning personalization[J]. Computing and Informatics, 2013, 32(3): 629-659.) et al. adopted clustering and collaborative filtering methods to recommend learning content of various media types; Segal (A Segal, K Gal, G Shani, et al. A difficulty ranking approach to personalization in e-learning[J]. International Journal of Human-Computer Studies, 2019, 130: 261-272.) et al. proposed the EduRank algorithm, which clusters users according to attributes such as the answering time, number of attempts of the same question by users, and the grade of users, so as to rank the acceptance degree of users for the difficulty of knowledge and combine collaborative filtering to achieve personalized learning path recommendations for users; Dihua Xu (D Xu, Z Wang, K Chen, et al. Personalized learning path recommender based on user profile using social tags[J]. in 2012 Fifth International Symposium on Computational Intelligence and Design, 2012, 1: 511-514.) et al. used KNN and Bayesian methods to recommend learning content; Junfu Xi (Xi J, Chen Y, Wang G. Design of a Personalized Massive Open Online Course Platform[J]. International Journal of Emerging Technologies in Learning, 2018, 13(04).) et al. used linear regression and matrix factorization methods to recommend learning paths according to users' historical behaviors and different needs.

[0033] Many researchers have also made many attempts in learning path recommendation using heuristic algorithms. Among them, Dwivedi (Dwivedi P, Kant V, Bharadwaj KK. Learning path recommendation based on modified variable length genetic algorithm[J]. Education&Information Technologies, 2018, 23(2):1-18.) et al. used a variable length genetic algorithm to recommend learning sequences for users. The algorithm would find the historical learning sequence that was most similar to the current user based on characteristics such as learning habits, knowledge level, and goals from those who had completed the course, and then generate a learning path suitable for the current user through the Variable Length Genetic Algorithm (VLGA); Junmin Ye (Jun-Min YE, XU Song, XU Chen, et al. Research on Learning Path Recommendation Algorithms in Online Learning Community[J]. DEStech Transactions on Engineering and Technology Research, 2018(ecar).) et al. clustered users based on characteristics such as cognitive level and learning style among users, and then used the Ant colony algorithm for personalized learning path recommendation for different groups of users after clustering; Ahmad Kardan (Ahmad, A, Kardan, et al. A new personalized learning path generation method: ACO-MAP[J]. Indian Journal of Scientific Research, 2014.) et al. used the ant colony algorithm to give a learning path that satisfied Ausubel's learning theory; Sivakumar N (N. Sivakumar, R Praveena. Determining optimized learning path for an e-learning system using ant colony optimization algorithm[J]. International Journal of Computer Science&Engineering Technology, 2015, 6(2):61-66.) etc. use the ant colony algorithm to optimize the learning path according to the characteristics of the user's ability, goals, learning behaviors, etc.; the literature (Lei W U, Fang Q, Learning path optimization based on improved particle swarm optimization method[J]. Journal of Systems Science and Mathematical Sciences, 2016, 36(12):2272.) uses the particle swarm algorithm to generate the optimal learning path; Ibrahim et al. (Ibrahim M E, Yang Y, Ndzi D, et al. Ontology-based Personalized Course Recommendation Framework[J]. IEEE Access, 2018:1-1.) proposed a course recommendation method that integrates multiple recommendation strategies using the genetic algorithm. This method applies a customized genetic algorithm to the pre-stage of recommendation, uses training data to optimize the parameter configuration of the recommendation system, and then constructs a recommendation system model with this configuration.

[0034] Existing collaborative filtering-based methods and heuristic algorithm-based methods have many limitations when making learning path recommendations. For example, when the collaborative filtering algorithm makes recommendations, it calculates the similarity between users, and thus recommends the knowledge points learned by another user similar to the current user to the current user, ignoring the successor relationship between knowledge points, which greatly reduces the accuracy of the recommendation results. In addition, both collaborative filtering-based methods and heuristic algorithm-based methods require a large amount of user interaction data and rely on the user's learning history to make learning recommendations for users, and cannot solve the cold start problem in recommendation problems. Especially when a new user participates in a newly established new course, it is often difficult for this user to obtain effective learning path recommendation results.

[0035] In view of this, the present application provides a learning path recommendation method based on a multi-level knowledge graph. In the multi-level knowledge graph, with the user's available learning time as a limit, the graph search algorithm is used to solve the learning path recommendation problem; the proposed multi-level knowledge graph of this method is novel and has a low algorithm complexity, which can greatly alleviate the cold start problem in learning path recommendations and meet the learning needs of users.

[0036] Please refer to Figure 1 , Figure 1 is a flowchart of a learning path recommendation method based on a multi-level knowledge graph provided by an embodiment of the present invention. A learning path recommendation method based on a multi-level knowledge graph provided by the present application includes:

[0037] Construct a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and there are predecessor relationships, successor relationships, and ancestor relationships among at least some of the knowledge points;

[0038] Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point and does not designate the starting knowledge point, construct the shortest learning path or the complete learning path;

[0039] Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and there is a path connection between the starting knowledge point and the target knowledge point, construct the shortest learning path or the complete learning path;

[0040] Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point and does not designate the starting knowledge point, construct the shortest learning path or the complete learning path;

[0041] Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point, designates another second-level knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and there is a path connection, construct the shortest learning path or the complete learning path;

[0042] Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the starting knowledge point and does not designate the target knowledge point, construct the shortest learning path or the complete learning path;

[0043] Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point, designates another second-level knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, construct the shortest learning path or the complete learning path;

[0044] Among them, the first-level knowledge points include multiple second-level knowledge points.

[0045] Specifically, please continue to refer to Figure 1 As shown, a learning path recommendation method based on a multi-level knowledge graph provided in this embodiment can solve the problems of cold start and inconsistent recommended content with the successor rules of knowledge points in the learning path recommendation algorithm. The learning path recommendation method based on the multi-level knowledge graph in this application can recommend according to the user's available time without relying on the user's learning history. The multi-level knowledge graph can meet the coarse-grained and fine-grained display of knowledge points. Compared with the single-level knowledge graph, the multi-level knowledge graph can separate different knowledge points, aggregate similar knowledge points, and has a more reasonable organizational structure.

[0046] It should be noted that please refer to Figure 2 andFigure 3 As shown Figure 2 is a schematic diagram of the knowledge point relationship provided by an embodiment of the present invention; Figure 3 is another schematic diagram of the knowledge point relationship provided by an embodiment of the present invention. Among them, a knowledge point (Knowledge point, KP) is a generalization of knowledge content. In this article, there are knowledge points that can be further subdivided, such as Figure 2 and Figure 3 As shown, the sequence includes multiple knowledge points and indivisible knowledge points, such as Figure 3 the array in is an indivisible knowledge point; among them, each knowledge point can be divided into a first-level knowledge point and a second-level knowledge point according to its level; it can be understood that Figure 2 the sequence in is a first-level knowledge point, Figure 3 the sub-knowledge point array of the sequence in is a second-level knowledge point.

[0047] The target knowledge point (Target knowledge point, TKP) is the knowledge point that a student wants to learn during the learning process.

[0048] The start knowledge point (Start knowledge point, SKP) is the knowledge point that a student is learning during the learning process or the knowledge point specified by the student to start learning from.

[0049] The learning dependency (Learning dependency, LD) is a kind of necessary connection between knowledge points during the learning process, Figure 3 indicating the successor relationship between the knowledge points included in the sequence;

[0050] Among them, the knowledge point kp1 is the predecessor knowledge point of the knowledge point kp2, and its expression form is:

[0051]

[0052] Among them, the knowledge point kp1 is not the predecessor knowledge point of the knowledge point kp2, and its expression form is:

[0053]

[0054] The ancestor relationship (Ancestor relation, AR) is a kind of inclusion relationship between knowledge points. For example, Figure 3 in, the sequence is the ancestor of the array, the sequence contains the array, and is also the father of the array;

[0055] Among them, the knowledge point kp1 is the ancestor of the knowledge point kp2, and its expression form is:

[0056] kp1→kp2;

[0057] Among them, knowledge point kp1 is not an ancestor of knowledge point kp2, and its expression form is:

[0058] kp1!→kp2.

[0059] A knowledge graph is a directed graph composed of knowledge points. The nodes are composed of knowledge points, and the relationships between the graphs represent the successor relationship and the subordination relationship between knowledge points, such as Figure 3 , the array is a successor knowledge point of binary search and a sub-knowledge point of the sequence.

[0060] KG = (KP, KE);

[0061] Among them, KP is the set of knowledge points included in the knowledge graph and is also the node that composes the knowledge graph. KE is the relationship between knowledge points, including learning dependence relationships and ancestor relationships.

[0062] The adjacency matrix KG of the knowledge graph is defined as:

[0063] C = (c ij ) n*n , 0 < i ≤ n, 0 < j ≤ n, i ≠ j;

[0064] Among them, c ij satisfies the following relationship:

[0065]

[0066] Among them, in order to represent the connectivity relationship between kp i and kp j , the symbols and are defined. When there is a path from kp i to kp j in the knowledge graph KG, its expression is:

[0067]

[0068] When there is no path from kp i to kp j in the knowledge graph KG, its expression is:

[0069]

[0070] A learning path (LP) is a sequence composed of one or more knowledge points. The expression of the learning path is as follows:

[0071] lp = {kp i , kp j , …, kpm}kp i ,kp j ,kp m ∈KG.

[0072] The shortest learning path (SLP) has the following expression:

[0073] SLP = lp i ,lp i ∈ {lp1, lp2, …, lp n}

[0074] and lp i has the fewest kp in {lp1, lp2, …, lp n};

[0075] Wherein, lp i refers to the learning path LP = {lp1, lp2, …, lp n} from the starting knowledge point skp to the target knowledge point tkp, wherein the learning path lp i includes the fewest knowledge points kp; please refer to Figure 4 shown in Figure 4 , which is a schematic diagram of the shortest learning path generation algorithm provided by an embodiment of the present invention. The input is the secondary starting knowledge point skp - (optional) and the secondary target knowledge point tkp - (optional), as well as the primary target knowledge point tkp + , and the secondary starting knowledge point and the secondary target knowledge point must be sub-knowledge points of the primary starting knowledge point. According to the input knowledge points, there are the following processing methods:

[0076] When the secondary starting knowledge point is empty, all knowledge points without predecessor knowledge points are put into the set startKps. When the secondary target knowledge point tkp - is not empty, find the path connected to the target knowledge point in the set, and select the path with the fewest knowledge points as the result P shortest for output; when the secondary target knowledge point is empty, record the paths starting from each knowledge point in startKps and ending with the knowledge points without successor knowledge points and put them into the set result, and find the path with the fewest knowledge points in result as the result P shortest for output.

[0077] When the secondary knowledge point is not empty, find the path with the fewest knowledge points among all the paths connected by the starting knowledge point skp - and the target knowledge point tkp + as the shortest path Pshortest Perform the output.

[0078] A complete learning path (CLP), whose expression is:

[0079]

[0080] Among them, the learning path CLP starts with the starting knowledge point skp, ends with the target knowledge point tkp, and includes two knowledge points with the largest sum of in-degree and out-degree that have paths between the starting knowledge point and the target knowledge point, as well as all their successor knowledge points that have paths to the target knowledge point; please refer to Figure 5 as shown. Figure 5 is a schematic diagram of the complete learning path generation algorithm provided by an embodiment of the present invention. The input is the secondary starting knowledge point skp - (optional) and the secondary target knowledge point tkp - (optional), as well as the primary target knowledge point tkp + , and the secondary starting knowledge point and the secondary target knowledge point must be sub-knowledge points of the primary starting knowledge point. According to the input knowledge points, there are the following processing methods:

[0081] If both the secondary starting knowledge point and the secondary target knowledge point are not empty, find two knowledge points kp that are connected to both the secondary starting knowledge point and the secondary target knowledge point and have the largest sum of in-degree and out-degree in , kp out , and add all the paths connected to kp in and kp out to LP complete and output it as the result.

[0082] If either the secondary starting knowledge point or the secondary target knowledge point is empty, find two connected knowledge points kp with the largest sum of in-degree and out-degree from the sub-knowledge points of all target knowledge points tkp + , kp in , kp out , and find all the paths that start with no predecessor knowledge point, end with no successor knowledge point, and are connected to kp in , kp out , add them to LP complete and output it as the result.

[0083] If the secondary starting knowledge point is empty or the secondary target knowledge point is empty, assume the non-empty knowledge point is kp i , and find two knowledge points kp that are connected to kp and have the largest sum of in-degree and out-degree from the sub-knowledge points of all tkp + , kp i , and find two knowledge points kp with the largest sum of in-degree and out-degree that are connected to kpin , kp out , if the second-level starting knowledge point is empty, then use the knowledge point without a predecessor knowledge point as the starting knowledge point, and add the path composed of kp in , kp out to LP complete and output; if the second-level ending knowledge point is empty, then use the knowledge point without a successor knowledge point as the ending knowledge point, and add the path composed of kp in , kp out to LP complete and output.

[0084] In an alternative embodiment of the present application, please continue to refer to Figure 2 and Figure 3 as shown, the process of constructing a multi-level knowledge graph is as follows.

[0085] In order to be able to express the relationships between knowledge points and meet the requirements of viewing knowledge points at both coarse-grained and fine-grained levels, this embodiment proposes a multi-level knowledge graph. This knowledge graph further abstracts different indivisible knowledge points into higher-level knowledge points; in the multi-level knowledge graph, each high-level knowledge point can completely reflect the overall context of the entire course, while the low-level knowledge points can reflect the specific content contained in the high-level knowledge points at a finer granularity.

[0086] Figure 2 and Figure 3 respectively show a two-layer knowledge graph composed of multiple knowledge points in the data structure course. Figure 2 shows the first-level knowledge points. Each first-level knowledge point can be further divided into some second-level knowledge points, such as Figure 3 the sequence shown contains multiple second-level knowledge points.

[0087] Among them, the second-level knowledge points come from MOOCCube. MOOCCube is a data warehouse open to researchers related to natural language processing, knowledge graphs, data mining, etc. in large-scale online education, including 706 real online courses, 38,181 teaching videos, 114,563 knowledge points, hundreds of thousands of course selections and video viewing records of 199,199 MOOC users; a concept graph with relationships such as precedence and hyponymy between concepts and a supplementary resource library containing hundreds of thousands of academic papers related to in-class concepts. This embodiment uses the course information in MOOCCube and the knowledge point information contained in the courses.

[0088] The first-level knowledge points are manually abstracted and extracted from similar knowledge points included in the course.

[0089] In an alternative embodiment of the present application, according to the multi-level knowledge graph, a shortest learning path or a complete learning path is constructed with the user's learning time as a limit.

[0090] For users with less learning time, they can learn according to the results generated by the shortest learning path algorithm. The shortest learning path contains fewer knowledge points, enabling users to quickly understand the overall content of the target knowledge point.

[0091] For users with sufficient learning time, they can learn according to the results generated by the complete learning path algorithm. The complete learning path contains more knowledge points, covering most of the knowledge points related to the target knowledge point, enabling users to understand the target knowledge point more completely.

[0092] In an alternative embodiment of the present application, please refer to Figure 6 and Figure 7 , Figure 6 is a schematic diagram for constructing the shortest learning path provided by an embodiment of the present invention, Figure 7 is a schematic diagram for constructing the complete learning path provided by an embodiment of the present invention. Based on the multi-level knowledge graph, when the user designates a knowledge point at a certain level as the target knowledge point and does not specify the starting knowledge point, the process of constructing the shortest learning path or the complete learning path includes:

[0093] Obtain the secondary knowledge points included in the target knowledge point, and obtain the secondary knowledge points without predecessor knowledge points as the starting knowledge points from them, and obtain the secondary knowledge points without successor knowledge points as the ending knowledge points from them;

[0094] Obtain the learning path between the starting knowledge point and the ending knowledge point, and obtain the path with the fewest knowledge points as the shortest learning path; or,

[0095] Obtain the secondary knowledge points included in the target knowledge point, and obtain the secondary knowledge points kp1 and secondary knowledge points kp2 with a path connection between them, and the sum of the in-degree of the secondary knowledge points kp1 and the out-degree of the secondary knowledge points kp2 is the largest;

[0096] Obtain the secondary knowledge points that are connected by a path to the secondary knowledge points kp1 and have no predecessor knowledge points as the starting knowledge points, and record the path lp1 between the starting knowledge point and the secondary knowledge points kp1;

[0097] Obtain the secondary knowledge points that are connected by a path to the secondary knowledge points kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge point and the secondary knowledge points kp2;

[0098] Integrate the path lp1 and the path lp2 as the complete learning path.

[0099] In an alternative embodiment of the present application, please continue to refer to Figure 6 and Figure 7 As shown, based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and there is a path connection between the starting knowledge point and the target knowledge point, the process of constructing the shortest learning path or the complete learning path includes:

[0100] Obtain the path between the starting knowledge point and the target knowledge point, and obtain the path including the fewest knowledge points from it as the learning path LP1 composed of first-level knowledge points, and its expression is:

[0101]

[0102] Wherein, is any first-level knowledge point in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1;

[0103] Obtain any first-level knowledge point in the learning path LP1, including the second-level knowledge points, obtain the second-level knowledge points without predecessor knowledge points from them, and obtain the second-level knowledge points without successor knowledge points from them; obtain the learning path between the two second-level knowledge points; obtain the learning path LP2 corresponding to the second-level knowledge points included in all first-level knowledge points in the learning path LP1, and its expression is:

[0104]

[0105] Wherein, is any second-level knowledge point in the learning path LP2;

[0106] Obtain the learning path corresponding to the second-level knowledge points included in the i-th first-level knowledge point in the learning path LP1 as the shortest learning path, and this learning path includes the fewest knowledge points, and j is the number of the fewest knowledge points; or,

[0107] Obtain the path between the starting point and the target knowledge point, and obtain the path including the knowledge point with the largest in-degree and the knowledge point with the largest out-degree as the learning path LP1 composed of first-level knowledge points, and its expression is:

[0108]

[0109] Wherein, is any first-level knowledge point in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1;

[0110] Obtain any first-level knowledge point Included secondary knowledge points, from which secondary knowledge points kp1 and kp2 connected by a path are obtained, and the sum of the in-degree of secondary knowledge point kp1 and the out-degree of secondary knowledge point kp2 is the largest;

[0111] Obtain the secondary knowledge points that are connected by a path to secondary knowledge point kp1 and have no predecessor knowledge points, and record the path lp1 between this secondary knowledge point and secondary knowledge point kp1;

[0112] Obtain the secondary knowledge points that are connected by a path to secondary knowledge point kp2 and have no successor knowledge points, and record the path lp2 between this secondary knowledge point and secondary knowledge point kp2;

[0113] Integrate path lp1 and path lp2, that is, obtain the learning path corresponding to any secondary knowledge point included in learning path LP1 Obtain the learning path corresponding to the secondary knowledge points included in all the first-level knowledge points included in learning path LP1, and the learning path LP2 corresponding to the secondary knowledge points included in all the first-level knowledge points included in learning path LP1, and its expression is:

[0114]

[0115] Among them, is any secondary knowledge point in learning path LP2;

[0116] Obtain the learning path corresponding to the secondary knowledge points included in the i-th first-level knowledge point in learning path LP1 from it As the complete learning path, j is the number of knowledge points.

[0117] In an optional embodiment of the present application, please continue to refer to Figure 6 and Figure 7 As shown, based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point and does not designate the starting knowledge point, the process of constructing the shortest learning path or the complete learning path includes:

[0118] Obtain the secondary knowledge points that have the same ancestor as the target knowledge point, and obtain the secondary knowledge points that have no predecessor knowledge points and are connected by a path to the target knowledge point as the starting knowledge point;

[0119] Taking the target knowledge point as the end knowledge point, obtain the learning path between the starting knowledge point and the end knowledge point, and obtain the path including the fewest knowledge points as the shortest learning path; or,

[0120] Obtain the secondary knowledge points that have the same ancestor as the target knowledge point, and obtain the secondary knowledge points kp1 and kp2 that are connected by a path to the target knowledge point, and the sum of the in-degree of secondary knowledge point kp1 and the out-degree of secondary knowledge point kp2 is the largest;

[0121] Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp1 and have no predecessor knowledge points as the starting knowledge points, and record the path lp1 between the starting knowledge points and the secondary knowledge point kp1;

[0122] Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge points and the target knowledge point;

[0123] Integrate the path lp1 and the path lp2 as the complete learning path.

[0124] In an optional embodiment of the present application, please continue to refer to Figure 6 and Figure 7 As shown in, based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point and designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point have the same ancestor and are path-connected, the process of constructing the shortest learning path or the complete learning path includes:

[0125] Obtain the path between the starting knowledge point and the target knowledge point, and obtain the path including the fewest knowledge points therefrom as the shortest learning path; or,

[0126] Obtain the path between the starting knowledge point and the target knowledge point, and obtain the secondary knowledge point kp1 and the secondary knowledge point kp2 that are path-connected therefrom, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest;

[0127] Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp1, and obtain the path lp1 that the secondary knowledge point can reach the starting knowledge point;

[0128] Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge points and the secondary knowledge point kp2;

[0129] Integrate the path lp1 and the path lp2 as the complete learning path.

[0130] In an optional embodiment of the present application, please continue to refer to Figure 6 and Figure 7 As shown in, based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the starting knowledge point and does not designate the target knowledge point, the process of constructing the shortest learning path or the complete learning path includes:

[0131] Obtain the secondary knowledge points that have the same ancestor as the starting knowledge point, and obtain the secondary knowledge points without successor knowledge points therefrom as the ending knowledge points;

[0132] Obtain the path between the starting knowledge point and the ending knowledge point, and obtain the path including the fewest knowledge points from it as the shortest learning path; or,

[0133] Obtain the secondary knowledge points connected by a path to the starting knowledge point, and obtain the secondary knowledge point kp1 and the secondary knowledge point kp2 connected by a path from them, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest;

[0134] Obtain the secondary knowledge points connected by a path to the secondary knowledge point kp1, and obtain the path lp1 that this secondary knowledge point can reach the starting knowledge point;

[0135] Obtain the secondary knowledge point that is connected by a path to the secondary knowledge point kp2 and has no successor knowledge point as the ending knowledge point, and record the path lp2 between the ending knowledge point and the secondary knowledge point kp2;

[0136] Integrate the path lp1 and the path lp2 as the complete learning path.

[0137] In an alternative embodiment of the present application, please continue to refer to Figure 6 and Figure 7 As shown, based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, the process of constructing the shortest learning path or the complete learning path includes:

[0138] Obtain the first ancestor knowledge point of the starting knowledge point, obtain the second ancestor knowledge point of the target knowledge point, obtain the path between the first ancestor knowledge point and the second ancestor knowledge point, and obtain the path including the fewest knowledge points from it as the learning path LP1 composed of first-level knowledge points, and its expression is:

[0139]

[0140] Among them, is any first-level knowledge point in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1;

[0141] Obtain any first-level knowledge point included in the learning path LP1 and except, obtain the secondary knowledge points included in the learning path LP1 of all first-level knowledge points except

[0142]

[0143] Among them, is any secondary knowledge point in the learning path LP2;

[0144] Obtain the secondary knowledge points that have the same ancestors as the target knowledge point, and from them, obtain the secondary knowledge points that have no predecessor knowledge points and are path-connected to the target knowledge point as the starting knowledge points;

[0145] Taking the target knowledge point as the ending knowledge point, obtain the learning path between the starting knowledge point and the ending knowledge point, and obtain the path and the path Improve the learning path LP2, and its expression is:

[0146]

[0147] Obtain the learning path corresponding to the secondary knowledge points included in the i-th primary knowledge point in the learning path LP1 as the shortest learning path, which includes the fewest knowledge points, and j is the number of the fewest knowledge points; or,

[0148] Obtain the first ancestor knowledge point of the starting knowledge point, obtain the second ancestor knowledge point of the target knowledge point, obtain the path between the first ancestor knowledge point and the second ancestor knowledge point, and from it, obtain the path-connected secondary knowledge points kp1 and secondary knowledge point kp2, and the path with the largest sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 as the learning path LP1 composed of primary knowledge points, and its expression is:

[0149]

[0150] Among them, is any primary knowledge point in the learning path LP1, and n is the number of primary knowledge points in the learning path LP1;

[0151] Obtain any primary knowledge point in the learning path LP1, including the secondary knowledge points, and from them, obtain the path-connected secondary knowledge points kp1 and secondary knowledge point kp2, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest;

[0152] Obtain the secondary knowledge point that is path-connected to the secondary knowledge point kp1 and has no predecessor knowledge point, and record the path lp1 between this secondary knowledge point and the secondary knowledge point kp1;

[0153] Obtain the secondary knowledge point that is path-connected to the secondary knowledge point kp2 and has no successor knowledge point, and record the path lp2 between this secondary knowledge point and the secondary knowledge point kp2;

[0154] Integrate path lp1 and path lp2, that is, obtain any level of knowledge points in learning path LP1 including the learning paths corresponding to the secondary knowledge points, and obtain the learning paths corresponding to the secondary knowledge points included in the primary knowledge points in learning path LP1 except and except, and the learning path LP2 corresponding to the secondary knowledge points included in the primary knowledge points, and its expression is:

[0155]

[0156] Among them, is any secondary knowledge point in learning path LP2;

[0157] Obtain the secondary knowledge points that have the same ancestors as the target knowledge points, and obtain the secondary knowledge points kp1 and secondary knowledge points kp2 that are path-connected to the target knowledge points from them, and the sum of the in-degree of the secondary knowledge points kp1 and the out-degree of the secondary knowledge points kp2 is the largest;

[0158] Obtain the secondary knowledge points that are path-connected to the secondary knowledge points kp1 and have no predecessor knowledge points as the starting knowledge points, and record the path lp1 between the starting knowledge points and the secondary knowledge points kp1;

[0159] Obtain the secondary knowledge points that are path-connected to the secondary knowledge points kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge points and the target knowledge points;

[0160] Integrate path lp1 and path lp2, and obtain path and path Improve learning path LP2, and its expression is:

[0161]

[0162] Obtain the learning path corresponding to the secondary knowledge points included in the i-th primary knowledge point in learning path LP1 as the complete learning path, and j is the number of the least knowledge points.

[0163] Based on the same inventive concept, the present application also provides a learning path recommendation device based on a multi-level knowledge graph, which is applied to a learning path recommendation method based on a multi-level knowledge graph provided in the above embodiments of the present application, and will not be elaborated here; the device includes:

[0164] A construction module for constructing a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and at least some of the knowledge points include predecessor relationships, successor relationships, and ancestor relationships;

[0165] A generation module is configured to construct a shortest learning path or a complete learning path based on a multi-level knowledge graph when the user designates a knowledge point at a certain level as the target knowledge point and does not designate a starting knowledge point; construct a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a knowledge point at a certain level as the target knowledge point, designates another knowledge point at a certain level as the starting knowledge point, and there is a path connection between the starting knowledge point and the target knowledge point; construct a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a secondary knowledge point as the target knowledge point and does not designate a starting knowledge point; construct a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and there is a path connection; construct a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a secondary knowledge point as the starting knowledge point and does not designate a target knowledge point; construct a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor; wherein, the knowledge points at the first level include multiple secondary knowledge points.

[0166] Based on the same inventive concept, the present application further provides an electronic device, including: a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method provided in the above embodiments of the present application is implemented, which will not be elaborated here; in this embodiment, the types of the processor and the memory are not specifically limited. For example, the processor may be a microprocessor, a digital information processor, a programmable logic system on a chip, etc.; the memory may be a volatile memory, a non-volatile memory, or a combination thereof, etc. Its implementation principle and technical effect are the same as those of the above method, which will not be elaborated here.

[0167] In an optional embodiment of the present application, the following simulation conditions are used for verification.

[0168] In this embodiment, a simulation experiment is carried out using the python language on a central processing unit of AMD Ryzen 9 5900HX, a GPU of NVIDIA GeForce RTX3070 Laptop GPU, and a windows11 operating system.

[0169] Please refer to Figure 8 , Figure 8 is a flowchart of a shortest learning path provided by an embodiment of the present invention, taking a data object as a secondary starting knowledge point and not designating a secondary target knowledge point to obtain the shortest path.

[0170] Please refer toFigure 9 , Figure 9 is a flowchart of a complete learning path provided by an embodiment of the present invention. Using data as the secondary starting knowledge point, without specifying the secondary target knowledge point, and using the sequence as the target knowledge point.

[0171] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A learning path recommendation method based on a multi-level knowledge graph, characterized in that, Including: Constructing a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and there are predecessor relationships, successor relationships, and ancestor relationships among at least some of the knowledge points; Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point and does not designate a starting knowledge point, constructing the shortest learning path or the complete learning path; Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and there is a path connection between the starting knowledge point and the target knowledge point, constructing the shortest learning path or the complete learning path; Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point and does not designate a starting knowledge point, constructing the shortest learning path or the complete learning path; Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point, designates another second-level knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and there is a path connection, constructing the shortest learning path or the complete learning path; Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the starting knowledge point and does not designate a target knowledge point, constructing the shortest learning path or the complete learning path; Based on the multi-level knowledge graph, when the user designates a certain second-level knowledge point as the target knowledge point, designates another second-level knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, constructing the shortest learning path or the complete learning path; Wherein, the first-level knowledge points include multiple second-level knowledge points.

2. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, wherein The process of constructing the shortest learning path or the complete learning path based on the multi-level knowledge graph when the user designates a certain first-level knowledge point as the target knowledge point and does not designate a starting knowledge point includes: Obtaining the second-level knowledge points included in the target knowledge point, obtaining the second-level knowledge points without predecessor knowledge points therefrom as the starting knowledge points, and obtaining the second-level knowledge points without successor knowledge points therefrom as the ending knowledge points; Obtaining the learning path between the starting knowledge point and the ending knowledge point, and obtaining the path including the fewest knowledge points therefrom as the shortest learning path; or, Obtaining the second-level knowledge points included in the target knowledge point, obtaining the second-level knowledge points kp1 and the second-level knowledge point kp2 with a path connection therefrom, and the sum of the in-degree of the second-level knowledge point kp1 and the out-degree of the second-level knowledge point kp2 is the largest; Obtaining the second-level knowledge points that are path-connected to the second-level knowledge point kp1 and have no predecessor knowledge points as the starting knowledge points, and recording the path lp1 between the starting knowledge point and the second-level knowledge point kp1; Obtaining the second-level knowledge points that are path-connected to the second-level knowledge point kp2 and have no successor knowledge points as the ending knowledge points, and recording the path lp2 between the ending knowledge point and the second-level knowledge point kp2; Integrating the path lp1 and the path lp2 as the complete learning path.

3. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, characterized in that Based on the multi-level knowledge graph, when a user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and there is a path connection between the starting knowledge point and the target knowledge point, the process of constructing the shortest learning path or the complete learning path includes: Obtain the path between the starting knowledge point and the target knowledge point, and obtain the path including the fewest knowledge points from it as the learning path LP1 composed of the first-level knowledge points, and its expression is: Among them, is any level of knowledge points in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1; Obtain any level of knowledge points in the learning path LP1 including the secondary knowledge points, obtain the secondary knowledge points without predecessor knowledge points from them, and obtain the secondary knowledge points without successor knowledge points from them; obtain the learning path between these two secondary knowledge points; obtain the learning path LP2 corresponding to the secondary knowledge points included in all the primary knowledge points in the learning path LP1, and its expression is: Among them, is any secondary knowledge point in the learning path LP2; Obtain the learning path corresponding to the secondary knowledge points included in the i-th primary knowledge point in the learning path LP1 As the shortest learning path, this learning path includes the fewest knowledge points, where j is the number of the fewest knowledge points; or Obtain the path between the starting point and the target knowledge point, and obtain the path including the knowledge points with the largest in-degree and the largest out-degree from it as the learning path LP1 composed of the first-level knowledge points, and its expression is: Among them, is any level of knowledge points in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1; Obtain any level of knowledge points in the learning path LP1 including the secondary knowledge points, and obtain the secondary knowledge points kp1 and kp2 connected by paths from them, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest; Obtain the second-level knowledge points that are path-connected to the second-level knowledge point kp1 and have no predecessor knowledge points, and record the path lp1 between this second-level knowledge point and the second-level knowledge point kp1; Obtain the second-level knowledge points that are path-connected to the second-level knowledge point kp2 and have no successor knowledge points, and record the path lp2 between this second-level knowledge point and the second-level knowledge point kp2; Integrate path lp1 and path lp2, that is, obtain any level of knowledge points in learning path LP1 including the learning paths corresponding to the secondary knowledge points, and obtain the learning paths LP2 corresponding to the secondary knowledge points included in all the primary knowledge points in learning path LP1. Its expression is: Among them, is any secondary knowledge point in the learning path LP2; Obtain the learning path corresponding to the secondary knowledge points included in the $i$-th primary knowledge point in the learning path LP1 As a complete learning path, $j$ is the number of knowledge points.

4. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, wherein Based on the multi-level knowledge graph, when a user designates a certain second-level knowledge point as the target knowledge point and does not designate the starting knowledge point, the process of constructing the shortest learning path or the complete learning path includes: Obtain the second-level knowledge points that have the same ancestor as the target knowledge point, and obtain the second-level knowledge points that have no predecessor knowledge points and are path-connected to the target knowledge point as the starting knowledge point; Taking the target knowledge point as the end knowledge point, obtain the learning path between the starting knowledge point and the end knowledge point, and obtain the path including the fewest knowledge points from it as the shortest learning path; or, Obtain the second-level knowledge points that have the same ancestor as the target knowledge point, and obtain the second-level knowledge points kp1 and the second-level knowledge point kp2 that are path-connected to the target knowledge point, and the sum of the in-degree of the second-level knowledge point kp1 and the out-degree of the second-level knowledge point kp2 is the largest; Obtain the second-level knowledge points that are path-connected to the second-level knowledge point kp1 and have no predecessor knowledge points as the starting knowledge point, and record the path lp1 between the starting knowledge point and the second-level knowledge point kp1; Obtain the second-level knowledge points that are path-connected to the second-level knowledge point kp2 and have no successor knowledge points as the end knowledge point, and record the path lp2 between the end knowledge point and the target knowledge point; Integrate the path lp1 and the path lp2 as the complete learning path.

5. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, wherein Based on the multi-level knowledge graph, when a user designates a certain second-level knowledge point as the target knowledge point, designates another second-level knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor and have a path connection, the process of constructing the shortest learning path or the complete learning path includes: Obtain the path between the starting knowledge point and the target knowledge point, and obtain the path including the fewest knowledge points from it as the shortest learning path; or, Obtain the path between the starting knowledge point and the target knowledge point, and obtain the second-level knowledge points kp1 and the second-level knowledge point kp2 that are path-connected from it, and the sum of the in-degree of the second-level knowledge point kp1 and the out-degree of the second-level knowledge point kp2 is the largest; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp1, and obtain the path lp1 from this secondary knowledge point to the starting knowledge point; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge points and the secondary knowledge point kp2; Integrate the path lp1 and the path lp2 as the complete learning path.

6. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, characterized in that The process of constructing the shortest learning path or the complete learning path based on the multi-level knowledge graph when the user designates a certain secondary knowledge point as the starting knowledge point and does not designate the target knowledge point includes: Obtain the secondary knowledge points that have the same ancestor as the starting knowledge point, and obtain the secondary knowledge points that have no successor knowledge points among them as the ending knowledge points; Obtain the path between the starting knowledge point and the ending knowledge point, and obtain the path with the fewest knowledge points among them as the shortest learning path; or, Obtain the secondary knowledge points that are path-connected to the starting knowledge point, and obtain the secondary knowledge point kp1 and the secondary knowledge point kp2 that are path-connected among them, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp1, and obtain the path lp1 from this secondary knowledge point to the starting knowledge point; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp2 and have no successor knowledge points as the ending knowledge points, and record the path lp2 between the ending knowledge points and the secondary knowledge point kp2; Integrate the path lp1 and the path lp2 as the complete learning path.

7. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, wherein The process of constructing the shortest learning path or the complete learning path based on the multi-level knowledge graph when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor includes: Obtain the first ancestor knowledge point of the starting knowledge point, obtain the second ancestor knowledge point of the target knowledge point, obtain the path between the first ancestor knowledge point and the second ancestor knowledge point, and obtain the path with the fewest knowledge points among them as the learning path LP1 composed of primary knowledge points, and its expression is: wherein, is any level of knowledge points in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1; Obtain any level of knowledge points in the learning path LP1 The included secondary knowledge points, obtain the secondary knowledge points without predecessor knowledge points from them, and obtain the secondary knowledge points without successor knowledge points from them; obtain the learning path between these two secondary knowledge points; obtain the learning path LP2 corresponding to the secondary knowledge points included in all primary knowledge points except and in the learning path LP1, and its expression is: Among them, is any secondary knowledge point in the learning path LP2; Obtain the secondary knowledge points that have the same ancestor as the target knowledge point, and obtain the secondary knowledge points that have no predecessor knowledge points and are path-connected to the target knowledge point as the starting knowledge points; Taking the target knowledge point as the end knowledge point, obtain the learning path between the start knowledge point and the end knowledge point, and obtain the path from it and the path Improve the learning path LP2, and its expression is: Obtain the learning path corresponding to the secondary knowledge points included in the $i$-th primary knowledge point in the learning path LP1 As the shortest learning path, this learning path includes the fewest knowledge points, where $j$ is the number of the fewest knowledge points; or Obtain the first ancestor knowledge point of the starting knowledge point, obtain the second ancestor knowledge point of the target knowledge point, obtain the path between the first ancestor knowledge point and the second ancestor knowledge point, and obtain the path with the largest sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 among the path-connected secondary knowledge points as the learning path LP1 composed of primary knowledge points, and its expression is: Among them, is any first-level knowledge point in the learning path LP1, and n is the number of first-level knowledge points in the learning path LP1; Obtain any level of knowledge points in the learning path LP1 including the secondary knowledge points, and obtain the secondary knowledge points kp1 and kp2 connected by paths therefrom, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp1 and have no predecessor knowledge points, and record the path lp1 between this secondary knowledge point and the secondary knowledge point kp1; Obtain the secondary knowledge points that are path-connected to the secondary knowledge point kp2 and have no successor knowledge points, and record the path lp2 between this secondary knowledge point and the secondary knowledge point kp2; Integrate path lp1 and path lp2, that is, obtain any level of knowledge points in learning path LP1 The learning path corresponding to the secondary knowledge points included, obtain the learning path LP2 corresponding to the secondary knowledge points included in the primary knowledge points other than and in learning path LP1. Its expression is: Among them, is any second-level knowledge point in the learning path LP2; Obtain the secondary knowledge points that have the same ancestor as the target knowledge point, and from them, obtain the secondary knowledge points kp1 and kp2 that are path-connected to the target knowledge point, and the sum of the in-degree of the secondary knowledge point kp1 and the out-degree of the secondary knowledge point kp2 is the largest; Obtain the secondary knowledge point that is path-connected to the secondary knowledge point kp1 and has no predecessor knowledge point as the starting knowledge point, and record the path lp1 between the starting knowledge point and the secondary knowledge point kp1; Obtain the secondary knowledge point that is path-connected to the secondary knowledge point kp2 and has no successor knowledge point as the ending knowledge point, and record the path lp2 between the ending knowledge point and the target knowledge point; Integrate path lp1 and path lp2, and obtain a path therefrom and the path Improve learning path LP2, whose expression is: Obtain the learning path corresponding to the secondary knowledge points included in the $i$-th primary knowledge point in the learning path LP1 As a complete learning path, $j$ is the number of the least knowledge points.

8. The learning path recommendation method based on a multi-level knowledge graph according to claim 1, characterized in that The secondary knowledge points are obtained through the MOOCCube database.

9. A learning path recommendation device based on a multi-level knowledge graph, characterized in that, Include: A construction module for constructing a multi-level knowledge graph, where the knowledge graph includes multiple knowledge points, and at least some of the knowledge points include predecessor relationships, successor relationships, and ancestor relationships; A generation module for constructing a shortest learning path or a complete learning path based on the multi-level knowledge graph when the user designates a certain level of knowledge point as the target knowledge point and does not designate a starting knowledge point; Based on the multi-level knowledge graph, when the user designates a certain level of knowledge point as the target knowledge point, designates another level of knowledge point as the starting knowledge point, and the starting knowledge point and the target knowledge point are path-connected, construct a shortest learning path or a complete learning path; Based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point and does not designate a starting knowledge point, construct a shortest learning path or a complete learning path; Based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, the target knowledge point and the starting knowledge point have the same ancestor, and are path-connected, construct a shortest learning path or a complete learning path; Based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the starting knowledge point and does not designate a target knowledge point, construct a shortest learning path or a complete learning path; Based on the multi-level knowledge graph, when the user designates a certain secondary knowledge point as the target knowledge point, designates another secondary knowledge point as the starting knowledge point, and the target knowledge point and the starting knowledge point do not have the same ancestor, construct a shortest learning path or a complete learning path; where the primary knowledge points include multiple of the secondary knowledge points.

10. An electronic device, characterized in that, Include: A memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements the method described in any one of claims 1 to 8.

Images