Learning resource recommendation method and device, equipment and storage medium
By building a knowledge graph and evaluation index system, combining learning resources and the records of training objects, and recommending learning resources, the problem of insufficient recommendations in the existing technology is solved, and high accuracy and comprehensive learning resource recommendations are achieved.
Patent Information
- Application Number
- CN202310583644.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing learning resource recommendation method is based on the association between users in the online community, and cannot provide recommendations when users without similar knowledge levels, and rely on single user posts, resulting in insufficient accuracy and comprehensiveness of the recommended content.
Build a knowledge graph, describe the relationship between learning resources through courses, knowledge points, course labels and teacher users, and describe the relationship between the exam and learning resources through the knowledge points contained in the course, combining the courses of the training object Learning records and assessment records, identify knowledge weaknesses, and recommend relevant learning resources.
Recommending learning resources without historical users solves the problem of cold start, significantly improves the accuracy and comprehensiveness of recommended content, and ensures good recommendation results.
Smart Images

Figure CN119025738A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of Internet technology, and in particular to a learning resource recommendation method, device, equipment and storage medium. Background Art
[0002] In recent years, with the rapid development of the Internet and other new generation information technologies, emerging technologies represented by artificial intelligence have been widely used in many fields, triggering a profound change in learning concepts and methods. In this context, online learning has transcended the limitations of time and space, providing learners with more possibilities to learn anytime and anywhere, and has thus flourished.
[0003] Taking the application of learning resource recommendation in online communities as an example, the relevant technology calculates the semantic similarity between user posts in the online community, determines the unique user with a similar knowledge level to the learner, obtains the user's learning path and recommends posts; if the user's learning path is missing, the unique user with a higher level of knowledge points is determined through the learner's knowledge background, and the user's posts are recommended to the learner; if the learner's knowledge level is already at the highest level, no recommendation is made.
[0004] However, the above recommendation method is only based on the association relationship between users in the online community. If there are no users with a similar knowledge level or a higher level of users as the learner, no recommendation can be made. In addition, the above recommendation method can only be based on a unique user. This single recommendation source is too dependent on the high quality of user posts, and it is difficult to ensure the accuracy and comprehensiveness of the recommended content, and the recommendation effect cannot be guaranteed. Summary of the invention
[0005] The embodiment of the present application provides a learning resource recommendation method, device, equipment and storage medium, which solves the cold start problem and greatly improves the accuracy and comprehensiveness of the recommended content, and has a good recommendation effect. The technical solution is as follows:
[0006] In one aspect, a learning resource recommendation method is provided, the method comprising:
[0007] Constructing a knowledge graph; wherein the knowledge graph is used to describe the association between learning resources through courses, knowledge points, course tags, and teacher users, and to describe the association between exams and learning resources through the knowledge points contained in the exams and courses;
[0008] Constructing evaluation indicators, and determining the knowledge mastery of the training subjects through the evaluation indicators; wherein the evaluation indicators are used to measure the learning effect of the training subjects;
[0009] Determine the knowledge weaknesses of the trainees according to their knowledge mastery;
[0010] Obtaining the course study records and assessment records of the trainees;
[0011] In combination with the knowledge graph, learning resources are recommended to the training subject according to at least one of the weak knowledge items, the course learning records and the assessment records.
[0012] In a possible implementation, constructing a knowledge graph includes:
[0013] Collect data and process the collected data;
[0014] According to the processed data, a learning resource network and a knowledge network are constructed respectively to obtain the knowledge graph;
[0015] The learning resource network is used to describe the association between learning resources through courses, knowledge points, course labels and teacher users; the knowledge network is used to describe the association between exams and learning resources through knowledge points contained in exams and courses.
[0016] In a possible implementation, the entity types in the knowledge graph include: courses, exams, questions, knowledge points, course labels, positions, and teacher users;
[0017] The association relationships described in the knowledge graph include: the association relationship between the course and the knowledge point, the association relationship between the course and the course label, the association relationship between the course and the position, the association relationship between the course and the teacher user, the association relationship between the test and the question, and the association relationship between the question and the knowledge point;
[0018] The course label is a keyword used to describe the course; the course label includes: course category, business field to which the course belongs, training program to which the course belongs, course catalog, course name, course content and course requirements.
[0019] In a possible implementation, constructing the evaluation index includes:
[0020] According to the business field to which the group belongs and the knowledge graph, the knowledge system of the group is divided into multiple knowledge domains; wherein one knowledge domain corresponds to one evaluation indicator;
[0021] Under each evaluation indicator, multiple evaluation levels are divided according to the degree of knowledge mastery; wherein, for any evaluation indicator, the degree of knowledge mastery represented by the multiple evaluation levels under the evaluation indicator is progressive in ascending order;
[0022] Determine the learning evaluation criteria for each evaluation indicator for trainees in different positions in the group;
[0023] Among them, the learning evaluation standards include: excellence standards, competence standards and qualification standards.
[0024] In a possible implementation, determining the knowledge mastery of the trainee by using the evaluation index; and determining the knowledge weaknesses of the trainee according to the knowledge mastery of the trainee include:
[0025] Obtaining the assessment records of the training subjects;
[0026] Determine the knowledge points corresponding to each question in the history assessment of the training subject;
[0027] Determine the evaluation index to which the knowledge point corresponding to each question is classified;
[0028] For any knowledge point, weighted average the answer scores of the trainees on the knowledge point to obtain the score of the knowledge point;
[0029] For any evaluation indicator, the scores of each knowledge point under the evaluation indicator are weighted averaged to obtain the score of the training object under the evaluation indicator;
[0030] Converting the score of the training object under the evaluation indicator into an evaluation grade to obtain the current grade of the training object under the evaluation indicator;
[0031] In response to the current level of the training subject under the evaluation indicator not meeting the job level requirement, the knowledge domain corresponding to the evaluation indicator is determined to be a knowledge weakness of the training subject.
[0032] In a possible implementation, combining the knowledge graph and recommending learning resources to the training subject according to at least one of the weak knowledge items, the course learning record, and the assessment record includes:
[0033] In response to the trainee's on-the-job training duration being less than a preset duration and the trainee not completing an on-the-job training course and a related examination, obtaining a first course sequence; wherein the first course sequence includes an unfinished on-the-job training course;
[0034] In response to the trainee entering a new position and the trainee not completing the position training course and related examinations, obtaining a second course sequence; wherein the second course sequence includes the unfinished position training course;
[0035] Acquiring a third course sequence according to the course learning record of the training subject;
[0036] In combination with the knowledge graph, according to the knowledge point-course path, courses associated with the knowledge points included in the weak knowledge items are retrieved to obtain a fourth course sequence;
[0037] In response to the trainee not meeting the assessment indicator in the previous assessment, according to the wrong questions of the trainee, combined with the knowledge graph, and according to the question-knowledge point-course path, the courses associated with the knowledge point corresponding to the wrong questions are retrieved to obtain a fifth course sequence;
[0038] Merge the first course sequence, the second course sequence, the third course sequence, the fourth course sequence and the fifth course sequence, and remove duplicate courses to obtain a final course sequence;
[0039] Recommending learning resources to the training object according to the final course sequence.
[0040] In a possible implementation, the acquiring the third course sequence according to the course learning record of the training object includes:
[0041] Sort the courses that the training subject has studied in order of time from recent to oldest, to obtain a sequence of studied courses;
[0042] Starting from the first course in the learned course sequence, obtaining the similarity between the first course and other courses that do not belong to the learned course sequence;
[0043] In response to the similarity between a certain course and the first course being greater than a similarity threshold, taking the course as a candidate course, and storing the attribute information of the candidate course in a course dictionary;
[0044] The attribute information includes: the course code of the candidate course, the time when the trainee studied the first course, the similarity between the candidate course and the first course, and the popularity of the candidate course;
[0045] After traversing the learned course sequence, the attribute information currently stored in the course dictionary is sorted at least according to the learning time, the similarity between courses and the popularity of the courses to obtain the third course sequence.
[0046] In a possible implementation, the method further includes:
[0047] In response to the course code of the candidate course already existing in the course dictionary, and the newly acquired similarity is greater than the saved similarity of the candidate course, the saved similarity is replaced with the newly acquired similarity in the course dictionary.
[0048] In a possible implementation, the method further includes:
[0049] On the learning page of each course, in combination with the knowledge graph, other courses associated with the current learning course are recommended to the training object.
[0050] In a possible implementation, the similarity between the courses is obtained in the following manner:
[0051] For any course, mapping the association relationship between the entities and descriptions included in the knowledge graph to a vector space to obtain a first feature vector of the course;
[0052] Mapping the first type of text of the course to a vector space to obtain a second feature vector of the course; wherein the first type of text includes: the course name and course content of the course;
[0053] Mapping the second type of text of the course to the vector space to obtain a third feature vector of the course; wherein the second type of text includes: the course duration and course label of the course;
[0054] Performing feature fusion on the first feature vector, the second feature vector and the third feature vector of the course to obtain a final feature vector of the course;
[0055] The similarity between different courses is determined by obtaining the cosine similarity between the final feature vectors of different courses;
[0056] Among them, for any two courses, the higher the similarity between the final feature vectors of the two courses, the higher the similarity between the two courses.
[0057] In another aspect, a learning resource recommendation device is provided, the device comprising:
[0058] The first construction module is configured to construct a knowledge graph; wherein the knowledge graph is used to describe the association relationship between learning resources through courses, knowledge points, course tags and teacher users, and to describe the association relationship between exams and learning resources through knowledge points included in exams and courses;
[0059] The second building module is configured to build an evaluation index; wherein the evaluation index is used to measure the learning effect of the training object;
[0060] A determination module is configured to determine the knowledge mastery of the training object through the evaluation index; and determine the knowledge weaknesses of the training object according to the knowledge mastery of the training object;
[0061] An acquisition module is configured to acquire the course learning record and assessment record of the training object;
[0062] The recommendation module is configured to recommend learning resources to the training object in combination with the knowledge graph according to at least one of the weak knowledge items, the course learning records and the assessment records.
[0063] In a possible implementation, the first building module is configured as follows:
[0064] Collect data and process the collected data;
[0065] According to the processed data, a learning resource network and a knowledge network are constructed respectively to obtain the knowledge graph;
[0066] The learning resource network is used to describe the association between learning resources through courses, knowledge points, course labels and teacher users; the knowledge network is used to describe the association between exams and learning resources through knowledge points contained in exams and courses.
[0067] In a possible implementation, the entity types in the knowledge graph include: courses, exams, questions, knowledge points, course labels, positions, and teacher users;
[0068] The association relationships described in the knowledge graph include: the association relationship between the course and the knowledge point, the association relationship between the course and the course label, the association relationship between the course and the position, the association relationship between the course and the teacher user, the association relationship between the test and the question, and the association relationship between the question and the knowledge point;
[0069] The course label is a keyword used to describe the course; the course label includes: course category, business field to which the course belongs, training program to which the course belongs, course catalog, course name, course content and course requirements.
[0070] In a possible implementation, the second building module is configured as follows:
[0071] According to the business field to which the group belongs and the knowledge graph, the knowledge system of the group is divided into multiple knowledge domains; wherein one knowledge domain corresponds to one evaluation indicator;
[0072] Under each evaluation indicator, multiple evaluation levels are divided according to the degree of knowledge mastery; wherein, for any evaluation indicator, the degree of knowledge mastery represented by the multiple evaluation levels under the evaluation indicator is progressive in ascending order;
[0073] Determine the learning evaluation criteria for each evaluation indicator for trainees in different positions in the group;
[0074] Among them, the learning evaluation standards include: excellence standards, competence standards and qualification standards.
[0075] In a possible implementation, the determining module is configured to:
[0076] Obtaining the assessment records of the training subjects;
[0077] Determine the knowledge points corresponding to each question in the history assessment of the training subject;
[0078] Determine the evaluation index to which the knowledge point corresponding to each question is classified;
[0079] For any knowledge point, weighted average the answer scores of the trainees on the knowledge point to obtain the score of the knowledge point;
[0080] For any evaluation indicator, the scores of each knowledge point under the evaluation indicator are weighted averaged to obtain the score of the training object under the evaluation indicator;
[0081] Converting the score of the training object under the evaluation indicator into an evaluation grade to obtain the current grade of the training object under the evaluation indicator;
[0082] In response to the current level of the training subject under the evaluation indicator not meeting the job level requirement, the knowledge domain corresponding to the evaluation indicator is determined to be a knowledge weakness of the training subject.
[0083] In a possible implementation, the recommendation module is configured as follows:
[0084] In response to the trainee's on-the-job training duration being less than a preset duration and the trainee not completing an on-the-job training course and a related examination, obtaining a first course sequence; wherein the first course sequence includes an unfinished on-the-job training course;
[0085] In response to the trainee entering a new position and the trainee not completing the position training course and related examinations, obtaining a second course sequence; wherein the second course sequence includes the unfinished position training course;
[0086] Acquiring a third course sequence according to the course learning record of the training subject;
[0087] In combination with the knowledge graph, according to the knowledge point-course path, courses associated with the knowledge points included in the weak knowledge items are retrieved to obtain a fourth course sequence;
[0088] In response to the trainee not meeting the assessment indicator in the previous assessment, according to the wrong questions of the trainee, combined with the knowledge graph, and according to the question-knowledge point-course path, the courses associated with the knowledge point corresponding to the wrong questions are retrieved to obtain a fifth course sequence;
[0089] Merge the first course sequence, the second course sequence, the third course sequence, the fourth course sequence and the fifth course sequence, and remove duplicate courses to obtain a final course sequence;
[0090] Recommending learning resources to the training object according to the final course sequence.
[0091] In a possible implementation, the recommendation module is configured as follows:
[0092] Sort the courses that the training subject has studied in order of time from recent to oldest, to obtain a sequence of studied courses;
[0093] Starting from the first course in the learned course sequence, obtaining the similarity between the first course and other courses that do not belong to the learned course sequence;
[0094] In response to the similarity between a certain course and the first course being greater than a similarity threshold, taking the course as a candidate course, and storing the attribute information of the candidate course in a course dictionary;
[0095] The attribute information includes: the course code of the candidate course, the time when the trainee studied the first course, the similarity between the candidate course and the first course, and the popularity of the candidate course;
[0096] After traversing the learned course sequence, the attribute information currently stored in the course dictionary is sorted at least according to the learning time, the similarity between courses and the popularity of the courses to obtain the third course sequence.
[0097] In a possible implementation, the recommendation module is further configured to:
[0098] In response to the course code of the candidate course already existing in the course dictionary, and the newly acquired similarity is greater than the saved similarity of the candidate course, the saved similarity is replaced with the newly acquired similarity in the course dictionary.
[0099] In a possible implementation, the recommendation module is further configured to:
[0100] On the learning page of each course, in combination with the knowledge graph, other courses associated with the current learning course are recommended to the training object.
[0101] In a possible implementation, the similarity between the courses is obtained in the following manner:
[0102] For any course, mapping the association relationship between the entities and descriptions included in the knowledge graph to a vector space to obtain a first feature vector of the course;
[0103] Mapping the first type of text of the course to a vector space to obtain a second feature vector of the course; wherein the first type of text includes: the course name and course content of the course;
[0104] Mapping the second type of text of the course to the vector space to obtain a third feature vector of the course; wherein the second type of text includes: the course duration and course label of the course;
[0105] Performing feature fusion on the first feature vector, the second feature vector and the third feature vector of the course to obtain a final feature vector of the course;
[0106] The similarity between different courses is determined by obtaining the cosine similarity between the final feature vectors of different courses;
[0107] Among them, for any two courses, the higher the similarity between the final feature vectors of the two courses, the higher the similarity between the two courses.
[0108] On the other hand, a computer device is provided, the device comprising a processor and a memory, the memory storing at least one program code, the at least one program code being loaded and executed by the processor to implement the above-mentioned learning resource recommendation method.
[0109] On the other hand, a computer-readable storage medium is provided, wherein at least one program code is stored in the storage medium, and the at least one program code is loaded and executed by a processor to implement the above-mentioned learning resource recommendation method.
[0110] On the other hand, a computer program product or a computer program is provided, which includes a computer program code, which is stored in a computer-readable storage medium, and a processor of a computer device reads the computer program code from the computer-readable storage medium, and the processor executes the computer program code, so that the computer device executes the above-mentioned learning resource recommendation method.
[0111] The embodiment of the present application provides a learning resource recommendation scheme, which can recommend learning resources without historical users, that is, the scheme solves the cold start problem. In detail, the scheme will construct a knowledge graph (learning resource knowledge graph) and an evaluation index (evaluation index system) respectively; wherein, the learning resource knowledge graph associates learning resources through courses, course labels, knowledge points and teacher users on the one hand, and associates exams and learning resources through knowledge points contained in exams and courses on the other hand, thereby forming a knowledge map with rich content; and by constructing an evaluation index system, the knowledge mastery of the training object can be evaluated, and the knowledge weaknesses of the training object can be analyzed accordingly. Next, combined with the learning resource knowledge graph, based on at least one of the training object's course learning records, assessment records and knowledge weaknesses, it is possible to recommend personalized learning resources to the training object, such as recommending learning resources associated with knowledge weaknesses. Since the scheme not only combines the learning resource knowledge graph when recommending resources, but also comprehensively considers the training object's course learning records, assessment records and knowledge weaknesses, it can effectively understand the knowledge situation of the training object, so the scheme can greatly improve the accuracy and comprehensiveness of the recommended content, and the recommendation effect is good. BRIEF DESCRIPTION OF THE DRAWINGS
[0112] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0113] Figure 1 It is a schematic diagram of an implementation environment of a learning resource recommendation method provided in an embodiment of the present application;
[0114] Figure 2 is a flow chart of a learning resource recommendation method provided in an embodiment of the present application;
[0115] Figure 3 is a schematic diagram of a learning resource network provided in an embodiment of the present application;
[0116] Figure 4 This is an overall execution flow chart of a learning resource recommendation provided by an embodiment of the present application;
[0117] Figure 5 is a flow chart of another learning resource recommendation method provided in an embodiment of the present application;
[0118] Figure 6 This is a schematic diagram of the correspondence between an exam, knowledge points and evaluation indicators provided in an embodiment of the present application;
[0119] Figure 7 is a structural diagram of a learning resource recommendation device provided in an embodiment of the present application;
[0120] Figure 8 It is a structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0121] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.
[0122] In this application, the terms "first", "second", etc. are used to distinguish the same or similar items with substantially the same role and function. It should be understood that there is no logical or temporal dependency between "first", "second", and "nth", nor is there a limitation on the quantity and execution order. It should also be understood that although the following description uses the terms first, second, etc. to describe various elements, these elements should not be limited by the terms.
[0123] These terms are only used to distinguish one element from another element. For example, without departing from the scope of various examples, a first element can be referred to as a second element, and similarly, a second element can also be referred to as a first element. Both the first element and the second element can be elements, and in some cases, can be separate and different elements.
[0124] Here, at least one means one or more than one, for example, at least one element can be one element, two elements, three elements, or any other integer greater than or equal to one. And multiple means two or more than two, for example, multiple elements can be two elements, three elements, or any other integer greater than or equal to two.
[0125] The "and / or" mentioned in this article indicates that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship.
[0126] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0127] Figure 1 It is a schematic diagram of an implementation environment of a learning resource recommendation method provided in an embodiment of the present application.
[0128] See also Figure 1 , the implementation environment includes: a terminal 101 and a learning resource recommendation device 102 .
[0129] In a possible implementation, the terminal 101 is installed with a target application. The target application is an independent application that supports users to conduct online learning. Exemplarily, the target application provides a function of displaying or playing learning resources. For example, the learning resource recommendation device 102 pushes learning resources to the terminal 101 through the target application, and then the user can obtain the learning resources recommended by the learning resource recommendation device 102. That is, the learning resource recommendation device 102 is responsible for executing the learning resource recommendation method provided in the embodiment of the present application.
[0130] The learning resource recommendation device 102 is a computer device, for example, the computer device is a server that provides services for the target application.
[0131] In another possible implementation, the target application may also be a sub-application of an independent application, such as a mini-program of a social application, which is not limited in this application.
[0132] It should be noted that the learning resources mentioned in the embodiments of the present application all refer to online learning resources, such as courses. In addition, the users mentioned in the embodiments of the present application are also referred to as training objects.
[0133] In another possible implementation, the learning resource recommendation method provided in the embodiment of the present application is applied to enterprise employee training. As an assistant for employee learning, it can help employees improve their job capabilities, cultivate employees' interest in continuous learning, analyze and evaluate employees' knowledge mastery, help enterprise managers understand employees' situations, and make improvements to employee training.
[0134] Figure 2 is a flow chart of a learning resource recommendation method provided by an embodiment of the present application. The execution subject of the method is a computer device, such as a learning resource recommendation device. Figure 2 , the method comprising:
[0135] 201. A computer device constructs a knowledge graph; wherein the knowledge graph is used to describe the association between learning resources through courses, knowledge points, course tags, and teacher users, and to describe the association between exams and learning resources through the knowledge points contained in the exams and courses.
[0136] In the embodiment of the present application, the above-mentioned knowledge graph is also referred to as a learning resource knowledge graph, wherein the learning resource knowledge graph is used to describe the relationship between learning resources and between learning resources and examinations.
[0137] Specifically, on the one hand, the above-mentioned knowledge graph associates learning resources through courses, knowledge points, course tags, and teacher users (also known as lecturers), forming a Figure 3 The learning resource network shown. Exemplarily, the functions of this part of the knowledge graph include but are not limited to: first, supporting quick retrieval of courses; that is, supporting fuzzy query of learning resources by keywords; second, supporting recommendation of similar courses; for the courses browsed by users, this part of the knowledge graph can quickly recommend related courses (courses that users may be interested in) and guide users to click to watch.
[0138] On the other hand, the above-mentioned knowledge graph associates exams and learning resources through the knowledge points contained in exams and courses to form a knowledge network. Among them, each question in the test paper corresponds to one or more knowledge points, and the course also corresponds to multiple knowledge points. Therefore, the questions can be associated with the course through the knowledge points, that is, the exam and the course can be associated. Exemplarily, the functions of this part of the knowledge graph include but are not limited to: supporting the recommendation of learning resources based on the user's test scores, such as recommending related courses corresponding to the knowledge points based on the questions that the user answered incorrectly in the exam.
[0139] 202. The computer equipment constructs evaluation indicators, and the knowledge mastery of the training subjects is determined through the evaluation indicators; wherein the evaluation indicators are used to measure the learning effect of the training subjects.
[0140] This step is used to construct an evaluation index system.
[0141] Taking corporate employee training as an example, see Figure 4 ,The learning evaluation model is used to quantitatively evaluate the learning effect of the ,employees participating in training based on the constructed evaluation index system, ,such as measuring the knowledge literacy of the ,employees based on their training test scores and other assessment ,results to obtain the scores of the employees under each evaluation ,index.
[0142] In an embodiment of the present application, the learning evaluation model evaluates the learning effects of employees from multiple dimensions, and based on the employees' knowledge weaknesses, retrieves and recommends personalized learning courses to employees through the above-mentioned knowledge graph.
[0143] 203. The computer equipment determines the trainees’ weak points in knowledge based on their mastery of knowledge.
[0144] In the embodiment of the present application, the knowledge weaknesses of the training subjects are determined based on the scores of the employees under various evaluation indicators. Please refer to the following description for the detailed process.
[0145] 204. Computer equipment is used to obtain the training subjects’ course learning records and assessment records.
[0146] See also Figure 4,When recommending learning resources, we will refer to employees’ course learning records and assessment records.
[0147] 205. The computer device combines the knowledge graph to recommend learning resources to the trainee based on at least one of the trainee's knowledge weaknesses, course learning records, and assessment records.
[0148] This step is done by Figure 4 The learning resource recommendation module shown is executed. Among them, learning resource recommendation is to recommend personalized learning courses to employees.
[0149] For example, the embodiments of the present application make recommendations from three aspects: required learning for the position, ability improvement, and interest cultivation:
[0150] 1. Required courses for positions. For new employees and employees who have changed positions, we recommend required courses for their positions. For example, by searching the training project labels corresponding to the positions in the above knowledge graph, we can recall the required courses for the related positions and output the courses that the employees have not yet learned.
[0151] 2. Capacity improvement. Use the learning evaluation model to analyze employees’ knowledge weaknesses and recommend courses that they need to improve based on their knowledge weaknesses. For example, through the learning evaluation model and the above knowledge graph, recall the courses related to employees’ knowledge weaknesses, sort them, and output them.
[0152] 3. Interest cultivation. Recommend courses that employees may be interested in based on their browsing history on the corporate training website to improve their learning enthusiasm. For example, based on the courses that employees have browsed, recall the related courses starting from the first course in order of browsing time or upload time from recent to ancient, and output them after sorting them according to multiple conditions such as course popularity, course similarity, browsing time or upload time. For example, the popularity of a course can be the number of views of the course, which is not limited in this application.
[0153] Combine the following Figure 4 The overall execution process of the learning resource recommendation method provided in the embodiment of the present application is introduced.
[0154] Take corporate employee training as an example. Figure 4 As shown, the overall execution process of this method is divided into two parts, namely employee learning evaluation and learning resource recommendation.
[0155] Regarding employee learning evaluation, this method constructs an evaluation index system for employees. The learning evaluation model is based on employees' daily assessment records. Through the constructed evaluation index system, the employee's score under each evaluation index can be determined, and based on the employee's score under each evaluation index, the employee's knowledge weaknesses can be determined.
[0156] Regarding learning resource recommendations, for new employees, induction training courses can be recommended to them; and after new employees have clarified their job positions or old employees have adjusted their positions, job courses for the corresponding positions can be recommended to them. Among them, induction training courses and job training courses correspond to the aforementioned required learning for the position, self-study courses correspond to the aforementioned ability improvement, and recommended improvement courses correspond to the aforementioned interest improvement. In addition, this method can combine the above-mentioned knowledge graph to recommend personalized courses to employees based on at least one of the employees' course learning records, assessment records, and knowledge weaknesses. Among them, the learning resource recommendation process involves mapping learning resources (courses) to vector space, and then completing the recommendation based on the encoded representation of the learning resources.
[0157] The embodiment of the present application provides a learning resource recommendation scheme, which can recommend learning resources without historical users, that is, the scheme solves the cold start problem. In detail, the scheme will construct a learning resource knowledge graph and an evaluation index system respectively; wherein, the learning resource knowledge graph associates learning resources through courses, course labels, knowledge points and teacher users on the one hand, and associates exams and learning resources through knowledge points contained in exams and courses on the other hand, thereby forming a knowledge map with rich content; and by constructing an evaluation index system, the knowledge mastery of employees can be evaluated, and the knowledge weaknesses of the training objects can be analyzed accordingly. Next, combined with the learning resource knowledge graph, based on at least one of the training object's course learning records, assessment records and knowledge weaknesses, personalized learning resources can be recommended to the training object. Since the scheme not only combines the learning resource knowledge graph when recommending resources, but also comprehensively considers the training object's course learning records, assessment records and knowledge weaknesses, it can effectively understand the knowledge situation of the training object, so the scheme can greatly improve the accuracy and comprehensiveness of the recommended content, and the recommendation effect is good.
[0158] The above is only a brief introduction to some technical details of the recommended solution provided in the embodiment of this application. Figure 5 The illustrated embodiment provides a detailed description of the recommended solution.
[0159] Figure 5 is a flow chart of another learning resource recommendation method provided by an embodiment of the present application. The execution subject of the method is a computer device, such as a learning resource recommendation device. Figure 5 , the method comprising:
[0160] 501. The computer device constructs a knowledge graph; wherein the knowledge graph is used to describe the association between learning resources through courses, knowledge points, course labels, and teacher users, and to describe the association between exams and learning resources through the knowledge points contained in the exams and courses.
[0161] This step first collects data and processes the collected data; then, constructs a learning resource knowledge graph.
[0162] Exemplarily, the collected data includes video courses, test data, and instructor information, etc., which are not limited in this application. The above data processing includes but is not limited to: segmenting the course name and course content of the video course and extracting keywords, supplementing or improving the course label of the course, the knowledge points corresponding to the course, etc., which are also not limited in this application.
[0163] In an embodiment of the present application, a learning resource network and a knowledge network are constructed respectively according to the data after data processing to obtain a learning resource knowledge graph; wherein, the learning resource network is used to describe the association relationship between learning resources through courses, knowledge points, course labels and teacher users; the knowledge network is used to describe the association relationship between exams and learning resources through the knowledge points contained in the exams and courses.
[0164] In a possible implementation, entity types in the learning resource knowledge graph include, but are not limited to: courses, exams, questions, knowledge points, course labels, positions, and teacher users. The association relationships described in the learning resource knowledge graph include, but are not limited to: the association relationship between courses and knowledge points, the association relationship between courses and course labels, the association relationship between courses and positions, the association relationship between courses and teacher users, the association relationship between exams and questions, and the association relationship between questions and knowledge points.
[0165] Course tags are keywords used to describe courses; they include, but are not limited to, keywords for course category, business field, training program, course catalog, course name, course content, and course requirements. For example, the course tags for the "Gas Station Daily Inspection Operation Procedures" include: equipment management, safety management, gas station safety officer training topics, gas station inspection, inspection operation, and equipment maintenance.
[0166] 502. Computer equipment constructs evaluation indicators, wherein the evaluation indicators are used to measure the learning effect of the training subjects.
[0167] This step is used to build an evaluation index system. Taking corporate employee training as an example, the group mentioned in this step refers to the enterprise.
[0168] In a possible implementation, the evaluation index is constructed in the following manner: according to the business field to which the group belongs and the knowledge graph constructed, the knowledge system of the group is divided into multiple knowledge domains; wherein one knowledge domain corresponds to one evaluation index; under each evaluation index, multiple evaluation levels are divided according to the degree of knowledge mastery; wherein, for any evaluation index, the degree of knowledge mastery represented by the multiple evaluation levels under the evaluation index is progressive in ascending order; and the learning evaluation standards for the training objects in different positions in the group under each evaluation index are determined. Exemplarily, the above-mentioned learning evaluation standards include: excellent standards, competent standards and qualified standards.
[0169] The following example illustrates the process of constructing the evaluation index system.
[0170] Taking oil and gas sales companies as an example, the company's knowledge system can be divided into five knowledge domains: management knowledge, marketing knowledge, safety knowledge, financial knowledge, and information knowledge, which are used as five evaluation indicators for evaluating employee learning. Among them, each evaluation indicator is divided into 6 levels according to the employee's knowledge mastery; among them, level 1 means no mastery, levels 2-5 are progressive, and level 6 means full mastery. Due to the different work content and responsibility requirements of different positions, employees in different positions need different knowledge systems. For example, the learning evaluation standard for gas station salespersons is (competence standard): management knowledge reaches level 3, marketing knowledge reaches level 4, safety knowledge reaches level 4, financial knowledge reaches level 3, and information knowledge reaches level 3.
[0171] 503. Computer equipment determines the knowledge mastery of trainees through evaluation indicators; based on the knowledge mastery of trainees, determine the trainees' knowledge weaknesses.
[0172] This step is used to evaluate the knowledge mastery of the trainees through the constructed evaluation index system. In a possible implementation, the knowledge mastery of the trainees is determined through the evaluation index; based on the knowledge mastery of the trainees, the knowledge weaknesses of the trainees are determined, including but not limited to the following methods:
[0173] Obtain the assessment records of the trainees; determine the knowledge points corresponding to each question in the trainees' historical assessments; determine the evaluation indicators to which the knowledge points corresponding to each question are classified;
[0174] For any knowledge point, the weighted average of the training subjects’ answers on the knowledge point is taken to obtain the score of the knowledge point;
[0175] For any evaluation indicator, the scores of each knowledge point under the evaluation indicator are weighted averaged to obtain the score of the training object under the evaluation indicator; the score of the training object under the evaluation indicator is converted into an evaluation grade to obtain the current grade of the training object under the evaluation indicator; in response to the current grade of the training object under the evaluation indicator not meeting the job grade requirement, the knowledge domain corresponding to the evaluation indicator is determined to be the knowledge weakness of the training object.
[0176] Combine the following Figure 6 The employee learning evaluation process is illustrated by an example.
[0177] The knowledge points determined in step 501 are classified into five knowledge domains (evaluation indicators) respectively, and the correlation between the test questions and knowledge points is obtained through the learning resource knowledge graph, so as to correspond the employee's test results to the five evaluation indicators, such as Figure 6 As shown. To analyze the employee's historical test scores, firstly, for each knowledge point, the employee's answer scores are weighted averaged; then for each evaluation indicator, the corresponding knowledge point scores are weighted averaged to obtain the employee's score under each evaluation indicator; finally, the scores are converted into grades. Compare the employee's current grade with the corresponding job grade requirements. If the requirements are not met, it means that the knowledge domain is the employee's weak knowledge area.
[0178] The following steps 504-509 are a personalized course recommendation process. For example, the recommended courses are displayed on the user homepage of the learning website, and this application does not limit this.
[0179] 504. In response to the trainee's on-the-job training duration being less than a preset duration and the trainee not completing an on-the-job training course and a related examination, the computer device acquires a first course sequence; wherein the first course sequence includes an unfinished on-the-job training course.
[0180] Exemplarily, the above preset time period may be 1 year, which is not limited in this application.
[0181] This step is used to determine whether the training object is a new employee; if it is a new employee, check whether the employee has completed the induction training course and related examinations; if not, output the unfinished induction training course sequence a1, that is, the first course sequence; otherwise, a1 is empty.
[0182] 505. In response to the trainee entering a new position and the trainee not completing the position training course and related examinations, the computer device obtains a second course sequence; wherein the second course sequence includes the unfinished position training course.
[0183] This step is used to determine whether the trainee has entered a new job position (for example, whether the employee has worked in the position for less than 1 year; if so, check whether the employee has completed the job training course and related examinations; if not, output the unfinished job training course sequence a2, that is, the second course sequence; otherwise, a2 is empty.
[0184] 506. The computer device obtains a third course sequence according to the course learning record of the trainee.
[0185] In a possible implementation, the third course sequence is obtained according to the course learning record of the training object, including but not limited to the following methods:
[0186] 5061. Sort the courses that the trainees have studied in order of time from recent to oldest, and obtain the sequence of studied courses.
[0187] This step is to sort the courses that the employee has studied in chronological order from recent to oldest, and obtain a course list lessons = [s1, s2, s3, ...].
[0188] Among them, the above time can be learning time or uploading time, and this application does not limit this.
[0189] 5062. Starting from the first course in the learned course sequence, obtain the similarity between the first course and other courses that do not belong to the learned course sequence; in response to the similarity between a certain course and the first course being greater than a similarity threshold, take the course as a candidate course, and store the attribute information of the course in the course dictionary.
[0190] Among them, starting from the first course s1, the similarity between s1 and other courses that do not belong to lessons is obtained. The embodiment of the present application sets the similarity threshold to k. If the similarity between any course in lessons and course r is greater than k, course r is recalled as a candidate course, and the relevant information of course r is saved in the course dictionary D.
[0191] For example, the attribute information includes but is not limited to: the course code id of the candidate course, the date when the trainee studied the first course, the similarity between the candidate course and the first course, and the heat of the candidate course. That is, the attribute information of course r can be represented as {id:[date,similar,heat]} in the course dictionary D.
[0192] In addition, if the recalled course already exists, that is, the id of the course already exists in D, then it is determined whether the latest recalled similarity value is greater than the saved similarity value; if so, the similarity value of the course in D is updated. That is, the method provided by the embodiment of the present application also includes: in response to the course code of the candidate course already existing in the course dictionary, and the newly obtained similarity is greater than the saved similarity of the candidate course, the saved similarity is replaced in the course dictionary with the newly obtained similarity.
[0193] 5063. After traversing the learned course sequence, the attribute information currently stored in the course dictionary is sorted at least according to the learning time, the similarity between courses and the course popularity to obtain a third course sequence.
[0194] For this step, after traversing lessons, the recalled course dictionary D is obtained, and the courses stored in D are sorted according to the attribute information date, similar, and heat, so that the third course sequence a3 can be obtained; wherein a3 is the key sequence of D.
[0195] The similarity between courses is obtained as follows:
[0196] 1. Extract course features and embed the courses into vector space; that is, the embodiment of the present application also provides a course feature extraction method.
[0197] 1-1. For any course, map the association relationship between entities and descriptions included in the learning resource knowledge graph to the vector space to obtain the first eigenvector of the course.
[0198] This step is to map the entities and associations in the learning resource knowledge graph into the vector space to obtain the vector representation v1 of the course, which retains the structural features between courses. For example, the Node2Vec algorithm can be used to map entities and associations into the vector space, which is not limited in this application.
[0199] 1-2. Map the first type of text of the course to the vector space to obtain the second eigenvector of the course; wherein the first type of text includes: the course name and course content of the course.
[0200] This step is to map the long text content such as the course name and course content into the vector space to obtain the vector representation v2 of the course. Exemplarily, the long text content can be mapped into the vector space by a text embedding representation method, which is not limited in this application.
[0201] 1-3. Map the second type of text of the course to the vector space to obtain the third eigenvector of the course; wherein the second type of text includes: the course duration and course label of the course.
[0202] This step is to map short text contents such as course duration, course requirements, and course labels into the vector space to obtain the vector representation v3 of the course. For example, the words can be encoded using one-hot encoding and the numbers can be normalized, which is not limited in this application.
[0203] The first point that needs to be explained is that for the first type of text, the above steps 1-2 can be performed even if it only includes the course name or course content; for the second type of text, the above steps 1-3 can be performed even if it only includes the course duration or course label; in addition, this application does not limit the content of the course label.
[0204] The second point that needs to be explained is that the difference between the first type of text and the second type of text is not the absolute length. The first type of text is usually a longer text containing at least two words, such as a sentence or paragraph; the second type of text refers to phrases, words, proper nouns, etc. The mapping methods of the two types of text are different. In the embodiment of the present application, "course name" and "course content" are classified as the first type of text, and "course duration" and "course label" are classified as the second type of text, and the corresponding mapping methods are used. In addition, even in some special cases, such as "course name" is a phrase (such as the "Fire Protection Law"), it is mapped according to the mapping method of the first type of text. Similarly, even if the "course label" is very long (such as "Work Management System Comprehensive Quality Evaluation Form"), it is mapped according to the mapping method of the second type of text.
[0205] The third point that needs to be explained is that the above steps 1-2 and the above steps 1-3 only illustrate the different processing methods for long texts and short texts. This application does not limit which ones belong to long texts and which ones belong to short texts. Theoretically, any information that can directly describe the course can be used as the source data for calculating the feature vector. Different mapping methods correspond to different data types of this information. For example, long texts are usually represented by text embedding, while short texts are usually represented by one-hot encoding, and numbers are usually represented by normalization (or regularization).
[0206] 1-4. Perform feature fusion on the first eigenvector, the second eigenvector and the third eigenvector of the course to obtain the final eigenvector of the course.
[0207] This step is to fuse the features of the vector representations of the three parts, that is, to obtain the final feature vector representation v of the course. In addition, for the vector representations with relatively large dimensions, they can be first reduced in dimension and then feature fused.
[0208] 1-5. Determine the similarity between different courses by obtaining the cosine similarity between the final feature vectors of different courses; wherein, for any two courses, the higher the similarity between the final feature vectors of the two courses, the higher the similarity between the two courses.
[0209] 507. The computer device combines the knowledge graph and searches for courses associated with the knowledge points included in the knowledge weaknesses of the training object according to the knowledge point-course path to obtain a fourth course sequence.
[0210] In an embodiment of the present application, for employees who fail the learning evaluation, that is, employees who have knowledge weaknesses, the learning resource knowledge graph is used to retrieve related courses according to the knowledge point--course path for the knowledge points corresponding to their failed evaluation indicators to form a fourth course sequence a4.
[0211] 508. In response to the trainee not reaching the assessment indicator in the historical assessment, the computer device retrieves the courses associated with the knowledge points corresponding to the wrong questions according to the trainee's wrong questions in combination with the knowledge graph, and follows the question-knowledge point-course path to obtain a fifth course sequence.
[0212] In the embodiment of the present application, for the exams that the employee has participated in, it is determined whether the score requirements have been met. If not, the learning resource knowledge graph is used to start from the wrong questions and retrieve related courses according to the question-knowledge point-course path to form a fifth course sequence a5.
[0213] 509. The computer device merges the first course sequence, the second course sequence, the third course sequence, the fourth course sequence and the fifth course sequence, and removes duplicate courses to obtain a final course sequence; and recommends learning resources to the trainees based on the final course sequence.
[0214] This step is to merge the above course sequences to obtain the course sequence A = [a1, a2, a4, a3, a5], and remove duplicate elements in A to obtain recommended learning resources.
[0215] 510. On the learning page of each course, the computer device combines the knowledge graph to recommend other courses related to the current learning course to the trainees.
[0216] This step is a parallel step of the above steps 504 to 509. This step is used to make quick course recommendations, that is, to recommend courses related to the courses that the employees have already learned.
[0217] Exemplarily, this type of course recommendation is applied to the learning page of each course and displayed in the sidebar as courses that may be of interest. In other words, the embodiment of the present application utilizes the fast retrieval feature of the learning resource knowledge graph to retrieve and output other courses associated with the current learning course as fast recommended courses.
[0218] Among them, personalized course recommendations are mainly used to guide employees' independent learning and ability improvement. Quick course recommendations are a supplement to personalized recommendations, which are used to discover employees' interests and guide them to broaden their knowledge.
[0219] The recommendation scheme provided by the embodiment of the present application can be applied to the recommendation of online learning resources with a large number of similar courses. In detail, the scheme will construct a learning resource knowledge graph; wherein, the learning resource knowledge graph associates learning resources through courses, course labels, knowledge points and teacher users on the one hand, and associates exams and learning resources through the knowledge points contained in exams and courses on the other hand, thereby forming a knowledge map with rich content. Combined with the knowledge graph, the scheme can retrieve and recommend learning resources through the knowledge graph based on the user's own learning needs, course learning records and assessment records. In other words, since the scheme not only combines the learning resource knowledge graph when recommending resources, but also comprehensively considers the course learning records, assessment records and knowledge weaknesses of the training subjects, the scheme can greatly improve the accuracy and comprehensiveness of the recommended content, and the recommendation effect is good.
[0220] In addition, this solution only needs to maintain learning resource data and does not need to rely on other user behaviors. That is, this solution does not rely on other user behaviors when recommending learning resources, so it can complete recommendations without historical users, providing an effective cold start method.
[0221] In addition, the solution solves the problems of incomplete knowledge system and insufficient data accumulation by adding labels to learning resources, expanding course requirements, course descriptions, lecturers and other information and supplementing them into the knowledge graph, and enhancing the correlation between courses.
[0222] In addition, the program also conducts a comprehensive evaluation of employees’ course learning and examination results during the analysis of employees, which can more effectively understand the employees’ situations and recommend learning resources for employees’ weak points in knowledge.
[0223] In addition, the program recommends learning resources based on employees' job requirements and knowledge mastery. For example, among a large number of similar and continuously updated courses, it prioritizes courses that are popular and uploaded more recently, and takes the completion of knowledge points as the basis for consideration, that is, the learning path is dynamically changing. For example, after an employee has completed the induction training course and has identified his or her position, that is, after completing the required learning for the position, the program can continue to recommend personalized training courses to the employee from the two perspectives of ability improvement and interest cultivation.
[0224] In summary, the solution first forms a knowledge map by constructing a knowledge graph of learning resources, and provides keyword-based course search and association functions; the search function can be a course entity automatically matched based on keywords, and the association function can be a course entity obtained by associating other entities. Secondly, an evaluation index system is constructed to comprehensively evaluate the knowledge literacy of employees, and quantitative analysis is conducted on the employees' continuous learning ability and knowledge mastery level, to analyze the strengths and weaknesses of each employee's knowledge system. Then, combined with the employees' course learning records and assessment records, personalized training courses are recommended to each employee from three perspectives: job requirements, ability improvement, and interest cultivation.
[0225] Based on the above description, it can be seen that this solution improves the effectiveness of corporate training and the enthusiasm of employees to learn by scientifically evaluating employees and recommending learning resources that are combined with business practices. In detail, this solution can be applied to the company's new employee induction training and business capability training for employees in different positions. It can plan personalized learning plans based on the employee's knowledge level and assessment records, and carry out differentiated training. It can effectively solve the problems of large differences in personal experience backgrounds of corporate employees, high mobility, high costs of unified and centralized training, and difficulty in evaluating and tracking training results. It not only improves the efficiency and effectiveness of employee training, but also improves the professional skills, cognitive ability and comprehensive quality of employees.
[0226] Figure 7 is a structural diagram of a learning resource recommendation device provided in an embodiment of the present application. Figure 7 , the device comprises:
[0227] The first construction module 701 is configured to construct a knowledge graph; wherein the knowledge graph is used to describe the association relationship between learning resources through courses, knowledge points, course tags and teacher users, and to describe the association relationship between exams and learning resources through knowledge points included in exams and courses;
[0228] The second construction module 702 is configured to construct an evaluation index; wherein the evaluation index is used to measure the learning effect of the training object;
[0229] The determination module 703 is configured to determine the knowledge mastery of the training object through the evaluation index; and determine the knowledge weaknesses of the training object according to the knowledge mastery of the training object;
[0230] An acquisition module 704 is configured to acquire the course learning record and assessment record of the training object;
[0231] The recommendation module 705 is configured to recommend learning resources to the training object in combination with the knowledge graph according to at least one of the weak knowledge items, the course learning records and the assessment records.
[0232] The embodiment of the present application provides a learning resource recommendation scheme, which can recommend learning resources without historical users, that is, the scheme solves the cold start problem. In detail, the scheme will construct a knowledge graph (learning resource knowledge graph) and an evaluation index (evaluation index system) respectively; wherein, the learning resource knowledge graph associates learning resources through courses, course labels, knowledge points and teacher users on the one hand, and associates exams and learning resources through knowledge points contained in exams and courses on the other hand, thereby forming a knowledge map with rich content; and by constructing an evaluation index system, the knowledge mastery of the training object can be evaluated, and the knowledge weaknesses of the training object can be analyzed accordingly. Next, combined with the learning resource knowledge graph, based on at least one of the training object's course learning records, assessment records and knowledge weaknesses, it is possible to recommend personalized learning resources to the training object, such as recommending learning resources associated with knowledge weaknesses. Since the scheme not only combines the learning resource knowledge graph when recommending resources, but also comprehensively considers the training object's course learning records, assessment records and knowledge weaknesses, it can effectively understand the knowledge situation of the training object, so the scheme can greatly improve the accuracy and comprehensiveness of the recommended content, and the recommendation effect is good.
[0233] In a possible implementation, the first building module 701 is configured to:
[0234] Collect data and process the collected data;
[0235] According to the processed data, a learning resource network and a knowledge network are constructed respectively to obtain the knowledge graph;
[0236] The learning resource network is used to describe the association between learning resources through courses, knowledge points, course labels and teacher users; the knowledge network is used to describe the association between exams and learning resources through knowledge points contained in exams and courses.
[0237] In a possible implementation, the entity types in the knowledge graph include: courses, exams, questions, knowledge points, course labels, positions, and teacher users;
[0238] The association relationships described in the knowledge graph include: the association relationship between the course and the knowledge point, the association relationship between the course and the course label, the association relationship between the course and the position, the association relationship between the course and the teacher user, the association relationship between the test and the question, and the association relationship between the question and the knowledge point;
[0239] The course label is a keyword used to describe the course; the course label includes: course category, business field to which the course belongs, training program to which the course belongs, course catalog, course name, course content and course requirements.
[0240] In a possible implementation, the second building module 702 is configured to:
[0241] According to the business field to which the group belongs and the knowledge graph, the knowledge system of the group is divided into multiple knowledge domains; wherein one knowledge domain corresponds to one evaluation indicator;
[0242] Under each evaluation indicator, multiple evaluation levels are divided according to the degree of knowledge mastery; wherein, for any evaluation indicator, the degree of knowledge mastery represented by the multiple evaluation levels under the evaluation indicator is progressive in ascending order;
[0243] Determine the learning evaluation criteria for each evaluation indicator for trainees in different positions in the group;
[0244] Among them, the learning evaluation standards include: excellence standards, competence standards and qualification standards.
[0245] In a possible implementation, the determination module 703 is configured to:
[0246] Obtaining the assessment records of the training subjects;
[0247] Determine the knowledge points corresponding to each question in the history assessment of the training subject;
[0248] Determine the evaluation index to which the knowledge point corresponding to each question is classified;
[0249] For any knowledge point, weighted average the answer scores of the trainees on the knowledge point to obtain the score of the knowledge point;
[0250] For any evaluation indicator, the scores of each knowledge point under the evaluation indicator are weighted averaged to obtain the score of the training object under the evaluation indicator;
[0251] Converting the score of the training object under the evaluation indicator into an evaluation grade to obtain the current grade of the training object under the evaluation indicator;
[0252] In response to the current level of the training subject under the evaluation indicator not meeting the job level requirement, the knowledge domain corresponding to the evaluation indicator is determined to be a knowledge weakness of the training subject.
[0253] In a possible implementation, the recommendation module 705 is configured to:
[0254] In response to the trainee's on-the-job training duration being less than a preset duration and the trainee not completing an on-the-job training course and a related examination, obtaining a first course sequence; wherein the first course sequence includes an unfinished on-the-job training course;
[0255] In response to the trainee entering a new position and the trainee not completing the position training course and related examinations, obtaining a second course sequence; wherein the second course sequence includes the unfinished position training course;
[0256] Acquiring a third course sequence according to the course learning record of the training subject;
[0257] In combination with the knowledge graph, according to the knowledge point-course path, courses associated with the knowledge points included in the weak knowledge items are retrieved to obtain a fourth course sequence;
[0258] In response to the trainee not meeting the assessment indicator in the previous assessment, according to the wrong questions of the trainee, combined with the knowledge graph, and according to the question-knowledge point-course path, the courses associated with the knowledge point corresponding to the wrong questions are retrieved to obtain a fifth course sequence;
[0259] Merge the first course sequence, the second course sequence, the third course sequence, the fourth course sequence and the fifth course sequence, and remove duplicate courses to obtain a final course sequence;
[0260] Recommending learning resources to the training object according to the final course sequence.
[0261] In a possible implementation, the recommendation module 705 is configured to:
[0262] Sort the courses that the training subject has studied in order of time from recent to oldest, to obtain a sequence of studied courses;
[0263] Starting from the first course in the learned course sequence, obtaining the similarity between the first course and other courses that do not belong to the learned course sequence;
[0264] In response to the similarity between a certain course and the first course being greater than a similarity threshold, taking the course as a candidate course, and storing the attribute information of the candidate course in a course dictionary;
[0265] The attribute information includes: the course code of the candidate course, the time when the trainee studied the first course, the similarity between the candidate course and the first course, and the popularity of the candidate course;
[0266] After traversing the learned course sequence, the attribute information currently stored in the course dictionary is sorted at least according to the learning time, the similarity between courses and the popularity of the courses to obtain the third course sequence.
[0267] In a possible implementation, the recommendation module 705 is further configured to:
[0268] In response to the course code of the candidate course already existing in the course dictionary, and the newly acquired similarity is greater than the saved similarity of the candidate course, the saved similarity is replaced with the newly acquired similarity in the course dictionary.
[0269] In a possible implementation, the recommendation module is further configured to:
[0270] On the learning page of each course, in combination with the knowledge graph, other courses associated with the current learning course are recommended to the training object.
[0271] In a possible implementation, the similarity between the courses is obtained in the following manner:
[0272] For any course, mapping the association relationship between the entities and descriptions included in the knowledge graph to the vector space to obtain the first feature vector of the course;
[0273] Mapping the first type of text of the course to a vector space to obtain a second feature vector of the course; wherein the first type of text includes: the course name and course content of the course;
[0274] Mapping the second type of text of the course to the vector space to obtain a third feature vector of the course; wherein the second type of text includes: the course duration and course label of the course;
[0275] Performing feature fusion on the first feature vector, the second feature vector and the third feature vector of the course to obtain a final feature vector of the course;
[0276] The similarity between different courses is determined by obtaining the cosine similarity between the final feature vectors of different courses;
[0277] Among them, for any two courses, the higher the similarity between the final feature vectors of the two courses, the higher the similarity between the two courses.
[0278] All the above optional technical solutions can be arbitrarily combined to form optional embodiments of the present application, which will not be described one by one here.
[0279] It should be noted that: the learning resource recommendation device provided in the above embodiment only uses the division of the above functional modules as an example when recommending learning resources. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the learning resource recommendation device provided in the above embodiment and the learning resource recommendation method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.
[0280] Figure 8 It is a structural diagram of a computer device 800 provided in an embodiment of the present application. The computer 800 may be a server. The computer device 800 may have relatively large differences due to different configurations or performances, and may include one or more processors (Central Processing Units, CPU) 801 and one or more memories 802, wherein at least one program code is stored in the memory 802, and the at least one program code is loaded and executed by the processor 801 to implement the learning resource recommendation method provided by the above-mentioned various method embodiments. Of course, the computer device 800 may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output, and the computer device 800 may also include other components for implementing device functions, which will not be repeated here.
[0281] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including a program code, and the program code can be executed by a processor in a computer device to complete the learning resource recommendation method in the above embodiment. For example, the computer-readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device.
[0282] In an exemplary embodiment, a computer program product or a computer program is also provided, which includes a computer program code, which is stored in a computer-readable storage medium. A processor of a computer device reads the computer program code from the computer-readable storage medium, and the processor executes the computer program code, so that the computer device executes the above-mentioned learning resource recommendation method.
[0283] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.
[0284] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. A learning resource recommendation method, characterized in that: The method comprises: Constructing a knowledge graph; wherein the knowledge graph is used to describe the association between learning resources through courses, knowledge points, course tags, and teacher users, and to describe the association between exams and learning resources through the knowledge points contained in the exams and courses; Constructing evaluation indicators, and determining the knowledge mastery of the training subjects through the evaluation indicators; wherein the evaluation indicators are used to measure the learning effect of the training subjects; Determine the knowledge weaknesses of the trainees according to their knowledge mastery; Obtaining the course study records and assessment records of the trainees; In combination with the knowledge graph, learning resources are recommended to the training subject according to at least one of the weak knowledge items, the course learning records and the assessment records.
2. The method according to claim 1, characterized in that The constructing of the knowledge graph includes: Collect data and process the collected data; According to the processed data, a learning resource network and a knowledge network are constructed respectively to obtain the knowledge graph; The learning resource network is used to describe the association between learning resources through courses, knowledge points, course labels and teacher users; the knowledge network is used to describe the association between exams and learning resources through knowledge points contained in exams and courses.
3. The method according to claim 1 or 2, characterized in that: The entity types in the knowledge graph include: courses, exams, questions, knowledge points, course labels, positions, and teacher users; The association relationships described in the knowledge graph include: the association relationship between the course and the knowledge point, the association relationship between the course and the course label, the association relationship between the course and the position, the association relationship between the course and the teacher user, the association relationship between the test and the question, and the association relationship between the question and the knowledge point; The course label is a keyword used to describe the course; the course label includes: course category, business field to which the course belongs, training program to which the course belongs, course catalog, course name, course content and course requirements.
4. The method according to claim 1, characterized in that The construction evaluation index includes: According to the business field to which the group belongs and the knowledge graph, the knowledge system of the group is divided into multiple knowledge domains; wherein one knowledge domain corresponds to one evaluation indicator; Under each evaluation indicator, multiple evaluation levels are divided according to the degree of knowledge mastery; wherein, for any evaluation indicator, the degree of knowledge mastery represented by the multiple evaluation levels under the evaluation indicator is progressive in ascending order; Determine the learning evaluation criteria for each evaluation indicator for trainees in different positions in the group; Among them, the learning evaluation standards include: excellence standards, competence standards and qualification standards.
5. The method according to claim 4, characterized in that The step of determining the knowledge mastery of the training object through the evaluation index; and determining the knowledge weaknesses of the training object according to the knowledge mastery of the training object, includes: Obtaining the assessment records of the training subjects; Determine the knowledge points corresponding to each question in the history assessment of the training subject; Determine the evaluation index to which the knowledge point corresponding to each question is classified; For any knowledge point, weighted average the answer scores of the trainees on the knowledge point to obtain the score of the knowledge point; For any evaluation indicator, the scores of each knowledge point under the evaluation indicator are weighted averaged to obtain the score of the training object under the evaluation indicator; Converting the score of the training object under the evaluation indicator into an evaluation grade to obtain the current grade of the training object under the evaluation indicator; In response to the current level of the training subject under the evaluation indicator not meeting the job level requirement, the knowledge domain corresponding to the evaluation indicator is determined to be a knowledge weakness of the training subject.
6. The method according to claim 1, characterized in that The combining the knowledge graph and recommending learning resources to the training object according to at least one of the weak knowledge items, the course learning records and the assessment records includes: In response to the trainee's on-the-job training duration being less than a preset duration and the trainee not completing an on-the-job training course and a related examination, obtaining a first course sequence; wherein the first course sequence includes an unfinished on-the-job training course; In response to the trainee entering a new position and the trainee not completing the position training course and related examinations, obtaining a second course sequence; wherein the second course sequence includes the unfinished position training course; Acquiring a third course sequence according to the course learning record of the training subject; In combination with the knowledge graph, according to the knowledge point-course path, courses associated with the knowledge points included in the weak knowledge items are retrieved to obtain a fourth course sequence; In response to the trainee not meeting the assessment indicator in the previous assessment, according to the wrong questions of the trainee, combined with the knowledge graph, and according to the question-knowledge point-course path, the courses associated with the knowledge point corresponding to the wrong questions are retrieved to obtain a fifth course sequence; Merge the first course sequence, the second course sequence, the third course sequence, the fourth course sequence and the fifth course sequence, and remove duplicate courses to obtain a final course sequence; Recommending learning resources to the training object according to the final course sequence.
7. The method according to claim 6, characterized in that The step of obtaining a third course sequence according to the course learning record of the training object includes: Sort the courses that the training subject has studied in order of time from recent to oldest, to obtain a sequence of studied courses; Starting from the first course in the learned course sequence, obtaining the similarity between the first course and other courses that do not belong to the learned course sequence; In response to the similarity between a certain course and the first course being greater than a similarity threshold, taking the course as a candidate course, and storing the attribute information of the candidate course in a course dictionary; The attribute information includes: the course code of the candidate course, the time when the trainee studied the first course, the similarity between the candidate course and the first course, and the popularity of the candidate course; After traversing the learned course sequence, the attribute information currently stored in the course dictionary is sorted at least according to the learning time, the similarity between courses and the popularity of the courses to obtain the third course sequence.
8. The method according to claim 7, characterized in that The method further comprises: In response to the course code of the candidate course already existing in the course dictionary, and the newly acquired similarity is greater than the saved similarity of the candidate course, the saved similarity is replaced with the newly acquired similarity in the course dictionary.
9. The method according to claim 1, characterized in that: The method further comprises: On the learning page of each course, in combination with the knowledge graph, other courses associated with the current learning course are recommended to the training object.
10. The method according to any one of claims 1 to 9, characterized in that The similarity between the courses is obtained as follows: For any course, mapping the association relationship between the entities and descriptions included in the knowledge graph to a vector space to obtain a first feature vector of the course; Mapping the first type of text of the course to a vector space to obtain a second feature vector of the course; wherein the first type of text includes: the course name and course content of the course; Mapping the second type of text of the course to the vector space to obtain a third feature vector of the course; wherein the second type of text includes: the course duration and course label of the course; Performing feature fusion on the first feature vector, the second feature vector and the third feature vector of the course to obtain a final feature vector of the course; The similarity between different courses is determined by obtaining the cosine similarity between the final feature vectors of different courses; Among them, for any two courses, the higher the similarity between the final feature vectors of the two courses, the higher the similarity between the two courses.
11. A learning resource recommendation device, characterized in that: The device comprises: The first construction module is configured to construct a knowledge graph; wherein the knowledge graph is used to describe the association relationship between learning resources through courses, knowledge points, course tags and teacher users, and to describe the association relationship between exams and learning resources through knowledge points included in exams and courses; The second building module is configured to build an evaluation index; wherein the evaluation index is used to measure the learning effect of the training object; A determination module is configured to determine the knowledge mastery of the training object through the evaluation index; and determine the knowledge weaknesses of the training object according to the knowledge mastery of the training object; An acquisition module is configured to acquire the course learning record and assessment record of the training object; The recommendation module is configured to recommend learning resources to the training object in combination with the knowledge graph according to at least one of the weak knowledge items, the course learning records and the assessment records.
12. A computer device, characterized in that: The device includes a processor and a memory, wherein at least one program code is stored in the memory, and the at least one program code is loaded and executed by the processor to implement the learning resource recommendation method according to any one of claims 1 to 10.
13. A computer-readable storage medium, characterized in that: The storage medium stores at least one program code, and the at least one program code is loaded and executed by the processor to implement the learning resource recommendation method according to any one of claims 1 to 10.
Citation Information
Patent Citations
Personalized teaching resource recommendation system based on knowledge map and ability evaluation
CN107085803A
Course recommendation method, electronic equipment and storage medium
CN111008340A
Personalized learning platform based on knowledge graph and construction method thereof
CN111831831A
Extracurricular learning tutoring system based on deep learning and knowledge graph and implementation method
CN112184500A
Personalized intelligent learning recommendation method, device and equipment and storage medium
CN112507140A
Cited By
Online course learning recommendation method based on knowledge graph
CN120256724A
High vocational course content recommendation method and system based on big data analysis
CN120336640A
Personalized course recommendation method and system based on knowledge graph
CN120407951A
Medical training content generating and pushing method based on knowledge graph
CN120873178A
Safety training model, method and system based on capability evaluation dynamic weight optimization model, and computer medium
CN120894205A
