Learning recommendation method and device for occupational development and storage medium

By building a job skills knowledge graph and a course evaluation hierarchical model, combining user learning history, identifying skills to be supplemented, and generating high-priority course resource recommendations, the problem of insufficiently refined course resource recommendations in the existing system is solved, and more effective Career planning and skill acquisition.

CN118966858APending Publication Date: 2024-11-15GUANGZHOU UNIVERSITY
View PDF 0 Cites 9 Cited by

Patent Information

Application Number
CN202410922469.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-10
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing learning recommendation system lacks consideration of the differences in user personalized needs and cannot make full use of information from multiple parties, resulting in the lack of refined curriculum resource recommendations and cannot meet the changes in skills and ability needs in the workplace environment.

Method used

By building a job skills knowledge graph, identifying the skills to be supplemented by users, matching course resources, combining the course evaluation hierarchy model and user learning history, high-priority and high-importance course resource recommendations are generated.

Benefits of technology

It improves the accuracy and relevance of course resource recommendations, promotes more effective career planning and skills acquisition, and adapts to changes in workplace needs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118966858A_ABST
    Figure CN118966858A_ABST
Patent Text Reader

Abstract

The invention discloses a learning recommendation method and device for occupational development and a storage medium. The method comprises the following steps: constructing a first recommended course resource list; performing position and skill association analysis according to the position information corpus, identifying position skills with a strong association relationship, and constructing a position skill knowledge graph; identifying and constructing a to-be-supplemented skill description set of the user; identifying course resources with high course cultivation skill matching degree, and generating a second recommended course resource data set; constructing a course resource evaluation hierarchy model; performing evaluation index weighting through theme modeling analysis or an expert scoring method, and performing comprehensive evaluation scoring on course resources in the second recommended course resource data set to generate a third recommended course resource data set; and recognizing and cultivating course resources of high-priority and high-importance knowledge points, and generating a fourth recommended course resource data set. The method can improve the accuracy and correlation of course resource recommendation, and can be widely applied to the technical field of computers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a learning recommendation method, device and storage medium for career development. Background Art

[0002] Personalized learning resource recommendation systems analyze user behaviors, preferences, and needs, search for the most relevant content from large amounts of information, and provide users with personalized course resource learning recommendations. Existing learning recommendation systems usually rely on static, one-size-fits-all model algorithms, lack consideration of the differences in user personalized needs, and often fail to adapt to the specific learning needs of individual users. At the same time, existing systems usually lack in-depth and comprehensive analysis of users' prior knowledge and skills, and therefore rarely consider the differences in user skills and knowledge and the gap between them and the knowledge and skills of the target position, making it difficult to provide the most important and highest priority learning recommendations for users. In addition, existing methods usually do not integrate real-time labor market data, and cannot fully utilize multi-party information consultations. Therefore, recommendations often fail to meet the ever-changing requirements for skills and abilities in the workplace, and fail to provide users with comprehensive and detailed course resource recommendations. Summary of the invention

[0003] The main purpose of the embodiments of the present invention is to propose a learning recommendation method, device and storage medium for career development, improve the accuracy and relevance of course resource recommendations, and promote more effective career planning and skill acquisition.

[0004] To achieve the above object, an embodiment of the present invention provides a learning recommendation method for career development, comprising the following steps:

[0005] Obtaining a first recommended course resource list according to user preset indicators, and constructing a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set;

[0006] Analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and build a position skill knowledge graph;

[0007] According to the job skill knowledge graph and the user's existing skill description corpus, identify and construct a set of user skill descriptions to be supplemented;

[0008] According to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, identifying course resources with a high degree of course training skill matching, and generating a second recommended course resource data set;

[0009] Constructing a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or collecting expert opinions through the Delphi method;

[0010] According to the model structure of the course resource evaluation hierarchical model, the evaluation indicators are weighted by subject modeling analysis or expert scoring method, and the course resources in the second recommended course resource data set are comprehensively evaluated and scored to generate a third recommended course resource data set;

[0011] According to the user's learning history information set and the third recommended course resource data set, course resources that cultivate high-priority and high-importance knowledge points are identified to generate a fourth recommended course resource data set.

[0012] In some embodiments, obtaining a first recommended course resource list according to user preset indicators and constructing a first recommended course resource data set includes the following steps:

[0013] Acquire the first recommended course resource list from the course resource database according to a first preset rule; wherein the first preset rule includes a matching condition for screening different attributes in the course resource corpus metadata of the course resource database;

[0014] Constructing the first recommended course resource data set according to the course resource meta information, course training objectives, course resource description and catalog content, and course evaluation information in the first recommended course resource list;

[0015] Among them, the course resource meta-information is used to describe the basic information of the course resources; the course training objectives are used to describe the training skill information of the course resources; the course resource description and directory content are used to describe the specific information contained in the course resource material content document; the course resource material content document is used for the keyword information of the course resource description and directory content; the course evaluation information is used to describe the user's evaluation level of the course, the course evaluation user and the course evaluation text information.

[0016] In some embodiments, performing position and skill association analysis based on a position information corpus, identifying position skills with strong associations, and constructing a position skill knowledge graph includes the following steps:

[0017] Obtain a job information corpus, and perform text preprocessing on the job information in the job information corpus to obtain preprocessed job information; the text preprocessing includes: using the jieba word segmentation library to perform word segmentation processing on the text data of occupation-related information, dividing the continuous text sequence into words or phrases, and obtaining a word sequence; cleaning the word sequence to obtain a cleaned word sequence; extracting text feature data based on the job information according to the word sequence, and obtaining a text encoding feature vector of the job information through a pre-trained BERT model; based on semantic analysis means, find out job skills in combination with contextual relationships, form a job skill set, and convert each job in the job skill set into a job skill vector based on the BERT model;

[0018] According to the position vector and the position skill vector, an association matrix between data items is established, the association between data items is calculated according to co-occurrence or correlation, frequent item sets are discovered using an association rule mining algorithm, association rules are generated based on the frequent item sets, and the association relationship between each position and position skill is generated;

[0019] According to the relationship between each position and position skills, a graph structure is constructed, and then a position skill knowledge graph of the relationship between positions and position skills is constructed.

[0020] In some embodiments, the method of establishing an association matrix between data items based on the position vector and the position skill vector, calculating the association between data items based on co-occurrence or correlation, discovering frequent item sets using an association rule mining algorithm, generating association rules based on the frequent item sets, and generating an association relationship between each position and position skill includes the following steps:

[0021] The BERT model obtained from the job description and skill requirement text embedding representation data is used to build a correlation matrix between data items;

[0022] Construct a correlation matrix by calculating the co-occurrence or correlation between the data represented by each embedding;

[0023] Use the Apriori algorithm to mine frequent itemsets on job descriptions and skill requirements, discover frequent itemsets, generate association rules based on frequent itemsets, and generate the association relationship between each job and job skills;

[0024] The process of mining frequent itemsets includes the following steps:

[0025] Extract features related to association rule mining from the collected job descriptions and skill requirements data, and set the minimum support threshold and minimum confidence threshold;

[0026] Convert the processed job description and skill requirement data into the input format of the Apriori algorithm, and determine the minimum support threshold and the minimum confidence threshold;

[0027] Treat a single feature as a candidate frequent item set, calculate its support, and filter out the frequent item sets that meet the conditions according to the minimum support threshold;

[0028] Combine the filtered frequent item sets in pairs and calculate the support of the newly generated candidate frequent item sets;

[0029] According to the minimum support threshold, filter out the frequent item sets that meet the conditions, and iterate until no new frequent item sets can be generated, and obtain all frequent item sets that meet the minimum support threshold conditions;

[0030] Generate association rules based on frequent item sets, calculate the confidence of each association rule, and filter out strong association rules that meet the conditions according to the minimum confidence threshold;

[0031] The method of generating association rules based on frequent item sets, calculating the confidence of each association rule, and selecting strong association rules that meet the conditions according to the minimum confidence threshold comprises the following steps:

[0032] Traverse all frequent item sets and split each frequent item set into antecedents and consequents;

[0033] For each combination of antecedent and consequent, calculate the confidence of the association rule;

[0034] According to the minimum confidence threshold, strong association rules that meet the conditions are screened out, and association rules with confidence less than the minimum confidence threshold are deleted;

[0035] After generating association rules based on frequent item sets, strong association rules are screened out by calculating confidence and using the minimum confidence threshold;

[0036] For the screened position-skill strong association rules, the evaluation indicators of lift, total confidence and Kulc index are calculated, and they are sorted according to the evaluation indicators to identify meaningful positions and position-skill associations.

[0037] In some embodiments, the step of identifying and constructing a set of user skill descriptions to be supplemented based on the job skill knowledge graph and the user's existing skill description corpus includes the following steps:

[0038] Extracting nodes and edges related to skills from the job skill knowledge graph, using natural language processing technology to perform text parsing on the description of the target job skill node, automatically extracting keywords related to the target job skill, and generating a target job skill keyword set; wherein the target job skill keyword set is composed of a user-preset keyword set and an automatically extracted keyword set; the automatically extracted keyword set is generated by extracting keywords from the description content of the target job skill node;

[0039] According to the description content of the user's existing skills in the user's existing skill description corpus, natural language processing technology is used to perform text analysis on the description content of the user's existing skills, automatically extract keywords related to the user's existing skills, and generate a keyword set of the user's existing skills;

[0040] According to the user's existing skill keyword set, a pre-trained word vector model is used to perform vector representation and addition processing on the user's existing skill keywords to obtain the semantic feature vector of the user's existing skill keywords;

[0041] Calculating the similarity between the semantic feature vector of the user's existing skill keywords and the semantic feature vector of the target position's skill keywords;

[0042] Generate a user target position skill matching report based on the similarity, and display the existing skills and similarities of users with higher matching degrees;

[0043] Construct a set of skill descriptions to be supplemented for the target position; wherein the set of skill descriptions to be supplemented for the target position is a set consisting of other target position skills and their skill description texts after removing the skills and their skill description texts whose similarity with the user's existing skills is greater than a similarity threshold from the target position skill description set.

[0044] In some embodiments, the step of identifying course resources with high course training skill matching according to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, and generating a second recommended course resource data set, comprises the following steps:

[0045] Use natural language processing technology to perform text analysis on the target position's skill description content in the user's skill description set to be supplemented, automatically extract keywords related to the target position's skill to be supplemented, and generate a target position's skill keyword set to be supplemented;

[0046] According to the target position skill keyword set to be supplemented, a pre-trained word vector model is used to perform vector representation and addition processing on the target position skill keywords to be supplemented therein, so as to obtain a semantic feature vector of the target position skill keywords to be supplemented;

[0047] For the course training goal data set, after removing the course training goal information of the course resources not included in the first recommended course resource list set, natural language processing technology is used to perform text parsing on the description of the course training skills, and course training skill related keywords are automatically extracted to generate a course training skill keyword set;

[0048] According to the course training skill keyword set, the pre-trained word vector model is used to perform vector representation and addition processing on the course training skill keywords therein to obtain the semantic feature vector of the course training skill keywords;

[0049] Calculate the similarity between the semantic feature vectors of the course training skill keywords in the first recommended course resource dataset and the semantic feature vectors of the target position's skill keywords to be supplemented, based on the semantic feature vectors of the course training skill keywords in the first recommended course resource dataset and the semantic feature vectors of the target position's skill keywords to be supplemented;

[0050] Display course resources and similarities with high course training skill matching, and build a second recommended course resource dataset.

[0051] In some embodiments, the step of weighting evaluation indicators by subject modeling analysis or expert scoring method according to the model structure of the course resource evaluation hierarchical model, and comprehensively evaluating and scoring the course resources in the second recommended course resource dataset to generate a third recommended course resource dataset includes the following steps:

[0052] Based on the constructed course resource evaluation hierarchical model and the determined relative weights of each evaluation main index and evaluation sub-index, the hierarchical analysis method is used to rank each course resource in the second recommended course resource data set to determine the overall relative advantage ranking of each course resource and the relative advantage ranking of each evaluation main index and evaluation sub-index of each course resource;

[0053] Based on the determined overall relative advantage ranking of each course resource and the relative advantage ranking of each course resource in terms of each main evaluation indicator and evaluation sub-indicator, according to the custom threshold, several course resources with high relative advantage ranking in the overall relative advantage ranking or in terms of each main evaluation indicator or evaluation sub-indicator are selected as the third recommended course resource data set.

[0054] In some embodiments, the step of identifying course resources that cultivate high-priority and high-importance knowledge points based on the user's learning history information set and the third recommended course resource data set to generate a fourth recommended course resource data set includes the following steps:

[0055] Determine the priority of the knowledge point for the user and determine the priority vector;

[0056] Determine the importance of the knowledge point to the course and determine the importance vector;

[0057] Determine the comprehensive priority vector of each knowledge point in the course resource according to the priority vector and the importance vector;

[0058] The comprehensive priority vectors of learning each knowledge point in the course resources are summed up to obtain the comprehensive priority of learning all knowledge points in the course resources;

[0059] The course resources in the third recommended course resource data set are arranged in descending order according to the comprehensive priority of learning all knowledge points of the course resources to obtain a fourth recommended course resource data set.

[0060] Another aspect of the embodiment of the present invention further provides a learning recommendation device for career development, comprising:

[0061] The first module is used to obtain a first recommended course resource list according to user preset indicators and construct a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set;

[0062] The second module is used to analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and construct a position skill knowledge graph;

[0063] The third module is used to identify and construct a set of user skill descriptions to be supplemented based on the job skill knowledge graph and the user's existing skill description corpus;

[0064] The fourth module is used to identify course resources with high course training skill matching according to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, and generate a second recommended course resource data set;

[0065] A fifth module is used to construct a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or expert opinions collected through the Delphi method;

[0066] The sixth module is used to weight the evaluation indicators through topic modeling analysis or expert scoring method according to the model structure of the course resource evaluation hierarchical model, and to comprehensively evaluate and score the course resources in the second recommended course resource data set to generate a third recommended course resource data set;

[0067] The seventh module is used to identify course resources that cultivate high-priority and high-importance knowledge points based on the user's learning history information set and the third recommended course resource data set, and generate a fourth recommended course resource data set.

[0068] Another aspect of an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the method described above to achieve the above-mentioned purpose.

[0069] Another aspect of an embodiment of the present invention provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor implements the above-mentioned method when executing the computer program.

[0070] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0071] The embodiments of the present invention include at least the following beneficial effects: The present invention provides a learning recommendation method, device and storage medium for career development, the method comprising: obtaining a first recommended course resource list according to user preset indicators, and constructing a first recommended course resource data set, the first recommended course resource data set including a course resource meta-information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set; performing position and skill association analysis according to a position information corpus, identifying position skills with strong associations, and constructing a position skill knowledge graph; identifying and constructing a user's to-be-supplemented skill description set according to the position skill knowledge graph and the user's existing skill description corpus; and A recommended course resource data set is concentrated in the course training target data set, and course resources with high course training skill matching are identified to generate a second recommended course resource data set; a course resource evaluation hierarchical model is constructed based on the course evaluation information data set concentrated in the second recommended course resource data set or expert opinions are collected through the Delphi method; based on the model structure of the course resource evaluation hierarchical model, evaluation indicators are weighted through topic modeling analysis or expert scoring method, and the course resources in the second recommended course resource data set are comprehensively evaluated and scored to generate a third recommended course resource data set; based on the user's learning history information set and the third recommended course resource data set, course resources that cultivate high priority and high importance knowledge points are identified to generate a fourth recommended course resource data set. The present invention can improve the accuracy and relevance of course resource recommendations and promote more effective career planning and skill acquisition. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a schematic diagram of an implementation environment provided by an embodiment of the present invention;

[0073] Figure 2 is a flow chart of the overall steps provided by an embodiment of the present invention;

[0074] Figure 3 This is a correlation diagram between positions and position skills formed by taking the position of data analyst as an example provided by an embodiment of the present invention;

[0075] Figure 4 This is an example diagram of selecting a teaching evaluation topic model based on semantic consistency and exclusivity indicators provided by an embodiment of the present invention;

[0076] Figure 5 This is a schematic diagram of the hierarchical model for the evaluation of online course resources in computer science, engineering and programming provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present invention. They are only examples of devices and methods consistent with some aspects of the embodiments of the present invention as detailed in the attached claims.

[0078] It is understood that the terms "first", "second", etc. used in the present invention may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0079] The terms "at least one", "multiple", "each", "any", etc. used in the present invention, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0080] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0081] The learning recommendation method, device and storage medium for career development provided by the embodiments of the present invention relate to the field of computer technology. The learning recommendation method for career development provided by the embodiments of the present invention can be applied to a terminal, can be applied to a server, or can be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements a learning recommendation method for career development, etc., but is not limited to the above forms.

[0082] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0083] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present invention. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to a network wirelessly or wired to complete data transmission and exchange.

[0084] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0085] In addition, the server 101 can also be a node server in the blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0086] The terminal 102 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc. The terminal 102 may also be a vehicle-mounted terminal of various device types described above, but is not limited thereto. The terminal 102 and the server 101 may be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment of the present invention.

[0087] Based on the example Figure 1 In the implementation environment shown, an embodiment of the present invention provides a learning recommendation method for career development. The following is an example of applying the learning recommendation method for career development to the server 101. It can be understood that the method can also be applied to the terminal 102.

[0088] Reference Figure 2 , Figure 2 The flowchart of the learning recommendation method for career development provided in the embodiment of the present invention is applied to a server. The execution subject of the method can be any of the aforementioned computer devices (including a server or a terminal).

[0089] Reference Figure 2 , the method may include the following steps:

[0090] Obtaining a first recommended course resource list according to user preset indicators, and constructing a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set;

[0091] Analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and build a position skill knowledge graph;

[0092] According to the job skill knowledge graph and the user's existing skill description corpus, identify and construct a set of user skill descriptions to be supplemented;

[0093] According to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, identifying course resources with a high degree of course training skill matching, and generating a second recommended course resource data set;

[0094] Constructing a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or collecting expert opinions through the Delphi method;

[0095] According to the model structure of the course resource evaluation hierarchical model, the evaluation indicators are weighted by subject modeling analysis or expert scoring method, and the course resources in the second recommended course resource data set are comprehensively evaluated and scored to generate a third recommended course resource data set;

[0096] According to the user's learning history information set and the third recommended course resource data set, course resources that cultivate high-priority and high-importance knowledge points are identified to generate a fourth recommended course resource data set.

[0097] Specifically, the implementation process of each step is described in detail below:

[0098] S101. Construction of course resource dataset.

[0099] According to the user preset indicators, a first recommended course resource list is obtained, and a first recommended course resource data set is constructed, wherein the first recommended course resource data set includes a course resource meta information data set, a course training target data set, a course resource description and catalog content data set, and a course evaluation information data set. The above step S101 includes S1011 and S1012:

[0100] Step S1011, course resource initial screening submodule: obtaining a first recommended course resource list from a course resource database according to a first preset rule;

[0101] The first recommendation list is a recommendation list of candidate courses.

[0102] The first preset rule includes fuzzy matching conditions and precise matching rules for screening different attributes in the course resource corpus metadata of the course resource database.

[0103] Specifically, in one embodiment of the present invention, input items such as the course name, the instructor and the school where he / she is enrolled are obtained from the course recommendation data set, and fuzzy matching is performed to obtain a preliminary recommended course resource list; based on the options selected by the user such as the subject, teaching stage, teaching grade, and evaluation level, an exact match is performed from the preliminary course recommendation list to obtain a first recommended course resource list.

[0104] Alternatively, in another embodiment of the present invention, fuzzy matching is performed from the course recommendation data set based on the options selected by the user, such as the subject, teaching stage, teaching grade, evaluation level, etc., to obtain a preliminary course recommendation list; and precise matching is performed from the preliminary course recommendation list based on input items input by the user, such as the course name, teaching teacher and the school where he / she is located, to obtain a first recommended course resource list.

[0105] A course resource corpus is constructed based on the first recommended course resource list, including course resource meta-information data, course training goal data, course resource description and catalog content data, and course evaluation information data.

[0106] Step S1012: Construct a first recommended course resource dataset based on the course resource corpus

[0107] The first recommended course resource dataset includes a course resource meta-information dataset, a course training goal dataset, a course resource description and catalog content dataset, and a course evaluation information dataset.

[0108] Specifically, in an embodiment of the present invention, a first recommended course resource data set is constructed according to the course resource meta information, course training objectives, course resource description and catalog content document information in the first recommended course resource list.

[0109] Among them, the course resource metadata dataset includes basic information used to describe course resources; the course training goal dataset includes training skill information used to describe course resources; the course resource description and catalog content dataset includes specific information used to describe course resource material content documents; the course resource material content document information includes keyword information used for course resource description and catalog content; the course evaluation information dataset includes information used to describe the user's evaluation level of the course (1, 2, ..., 5; 1 is the lowest score and 5 is the highest score), course evaluation users, and course evaluation text information.

[0110] A course evaluation text dataset is constructed based on the course evaluation text information in the course evaluation information dataset.

[0111] Optionally, the course resource meta-information subset includes course name, subject category, applicable grade, lecturer, and course offering institution.

[0112] S102. Construct a job skill and job data set based on the job information corpus.

[0113] The job skills and job data sets are constructed based on the job information corpus, and the job skills knowledge graph is constructed by identifying job skills with strong correlations. This includes but is not limited to the following steps: step S1021, step S1022, and step S1023.

[0114] Step S1021, Job Information Extraction Sub-module:

[0115] Specifically, the job information corpus can be job recruitment information and recruitment requirement data obtained from recruitment websites.

[0116] In an embodiment of the present invention, a crawler tool (such as BeautifulSoup) is used to obtain jobs and job recruitment information, such as job descriptions, skill requirements, work tasks, etc., from various sources such as the Ministry of Education website and related websites, major recruitment websites, etc., to construct a job dataset and a job skill dataset. Among them, the job dataset includes basic information for describing job names; the job skill dataset is used to describe basic information such as job descriptions, skill requirements, work tasks, etc.

[0117] Perform text preprocessing on the basic information of job names and the basic information such as job descriptions, skill requirements, work tasks, etc. in the job dataset and the job skill dataset.

[0118] Specifically, in an embodiment of the present invention, the jieba word segmentation library is used to perform word segmentation processing on the text data of occupation-related information to divide a continuous text sequence into words or phrases, which helps to eliminate some noise and redundant information in the text. The segmented word sequence can better represent the semantics and structure of the text, which helps the subsequent feature extraction work.

[0119] Clean the word sequence to obtain a cleaned word sequence. Among them, cleaning the word sequence includes at least one of the following operations: deleting duplicate words, removing punctuation marks, deleting special symbols, and stop word processing (pause words such as "of", "we", etc. that do not have much meaning), to obtain job information text feature data.

[0120] Based on the text feature data of job information, through a pre-trained BERT model, obtain job information text encoding feature vectors. Based on the BERT model, convert each job in the job set into a job vector;

[0121] Specifically, the job information text feature data is represented as T = [t1, t2,..., t n ; t1 represents the first job information text feature data, t2 represents the second job information text feature data, t n represents the nth job information text feature data; the text encoding feature vector is represented as represents the first job information text encoding feature vector, represents the second job information text encoding feature vector, represents the nth job information text encoding feature vector.

[0122] Based on semantic analysis, job skills are found in combination with contextual relationships to form a job skill set, and each job in the job skill set is converted into a job skill vector based on the BERT model;

[0123] Specifically, the text feature data of job skill information is expressed as S = [s1, s1, ..., s n ]; s1 represents the text feature data of the first position skill information, s2 represents the text feature data of the second position skill information, s n Represents the text feature data of the skill information of the nth position;

[0124] Text encoding feature vector representation Represents the first position skill information text encoding feature vector, Represents the second position skill information text encoding feature vector, Represents the text encoding feature vector of the nth job skill information.

[0125] Step S1022, association analysis submodule:

[0126] Based on the job skills and the job vectors, an association matrix between data items is established, the association between data items is calculated based on co-occurrence or correlation, frequent item sets are discovered using an association rule mining algorithm (such as Apriori), association rules are generated based on frequent item sets, and the association relationship between each of the jobs and job skills is generated.

[0127] Specifically, we first build a correlation matrix between data items from the BERT model embedding representation data obtained from the job description and skill requirement text conversion. We build a correlation matrix by calculating the co-occurrence or correlation between the text embedding representation data items to analyze the relationship between the data items. Then, we use the Apriori algorithm to mine frequent item sets for the job description and skill requirement data items, including the following steps:

[0128] 1. Extract features related to association rule mining from the collected job descriptions and skill requirements data, and set the minimum support threshold and minimum confidence threshold;

[0129] 2. Convert the processed job description and skill requirement data into the input format of the Apriori algorithm, and determine the minimum support threshold and minimum confidence threshold;

[0130] 3. Treat a single feature as a candidate frequent item set, calculate its support, and filter out the frequent item sets that meet the conditions according to the minimum support threshold;

[0131] 4. Combine the filtered frequent item sets in pairs and calculate the support of the newly generated candidate frequent item sets;

[0132] 5. Filter out the frequent item sets that meet the conditions according to the minimum support threshold, and iterate until no new frequent item sets can be generated, and obtain all frequent item sets that meet the minimum support threshold conditions;

[0133] 6. Generate association rules based on frequent item sets, calculate the confidence of each association rule, and filter out strong association rules that meet the conditions according to the minimum confidence threshold.

[0134] In some embodiments, generating association rules based on frequent item sets, calculating the confidence of each association rule, and screening out strong association rules that meet the conditions according to a minimum confidence threshold comprises the following steps:

[0135] Traverse all frequent item sets and split each frequent item set into antecedents and consequents;

[0136] 1. For each combination of antecedent and consequent, calculate the confidence of the association rule;

[0137] 2. Filter out strong association rules that meet the conditions according to the minimum confidence threshold, and delete association rules whose confidence is less than the minimum confidence threshold;

[0138] 3. After generating association rules based on frequent item sets, strong association rules are screened out by calculating confidence and using the minimum confidence threshold;

[0139] 4. For the selected job-skill strong association rules, calculate the evaluation indicators of lift, total confidence and Kulc index, and sort them according to the evaluation indicators to identify meaningful job and job skill associations. Figure 3 , is an example schematic diagram of the relationship between a position and position skills provided in an embodiment of the present application. Figure 3 As shown, the correlation between positions and position skills is formed by taking the position of data analyst as an example.

[0140] Step S1023, job skill knowledge graph construction submodule:

[0141] Based on the relationship between positions and position skills, a graph structure is constructed, in which nodes represent entities such as positions and skills, and edges represent the relationship between them. Graph theory algorithms, such as graph traversal algorithms or shortest path algorithms, are used to construct a position skill knowledge graph of the relationship between positions and position skills.

[0142] Specifically, an initial vector is defined for each node in the relationship between the position and the position skill.

[0143] Specifically, each association relationship can be called a triple (head, relationship, tail) data, and each triple contains three nodes. For example, the node representing the occupation (such as "data analyst", "data scientist", "statistician") is called the head node, denoted by h; the node representing the skill (such as "statistics", "programming", "visualization") is called the tail node, denoted by t. The middle "relationship" is called the relationship node, denoted by r, which represents the relationship between the occupation and the skill (such as "data analyst" requires "statistics" and "programming").

[0144] In one example of the present invention, based on the triple data, the NetworkX tool is used to construct a job skill knowledge graph of the relationship between job positions and job skills.

[0145] Specifically, first, create an empty graph object. Next, add nodes representing occupations (such as "data analyst", "data scientist", "statistician") and nodes representing skills (such as "statistics", "programming", "visualization"). Then, connect these nodes by adding relationship edges to clarify the relationship between occupations and skills (such as "data analyst" requires "statistics" and "programming"). Finally, graph theory algorithms such as graph traversal or path search can be used to operate in the graph to build a job skill knowledge graph.

[0146] S103. User skills assessment.

[0147] The target position required skills analysis submodule uses natural language processing technology based on the user skill data set to analyze the skills required for the user's target position and generate a target position skill analysis report; the user skill matching submodule uses the cosine similarity algorithm based on the target position skill analysis report to perform skill matching, generate a skill matching report, and identify skills to be supplemented. It includes but is not limited to the following steps: step S1031, step S1032.

[0148] Step S1031, based on the job skill knowledge graph and the user's existing skill description corpus, identify and construct a set of user skill descriptions to be supplemented: extract nodes and edges related to skills from the job skill knowledge graph. Use natural language processing technology to perform text parsing on the description of the target job skill node, automatically extract keywords associated with the target job skill, and generate a target job skill keyword set. The job skill keyword set consists of a user preset keyword set and an automatically extracted keyword set, wherein the automatically extracted keyword set is generated by extracting keywords from the description content of the job skill node.

[0149] Specifically, natural language processing technology is used to perform text analysis on the description content of the target position skill node, automatically extract target position skill related keywords, and generate a target position skill keyword set;

[0150] The pre-trained word vector model is used to perform vector representation and addition processing on the target job skill keywords to obtain the semantic feature vector w of the target job skill keywords. k The target position skill keyword set is composed of a user preset keyword set and an automatically extracted keyword set, wherein the automatically extracted keyword set is generated by extracting keywords from the target position skill description text.

[0151] Step S1032: Identify and construct a set of user skill descriptions to be supplemented based on the semantic feature vector of the job skill keyword and the user's existing skill description corpus:

[0152] Among them, the user's existing skill description corpus consists of the description content of the user's existing skills;

[0153] According to the description content of the user's existing skills in the user's existing skill description corpus, natural language processing technology is used to perform text analysis on the description content of the user's existing skills, automatically extract keywords related to the user's existing skills, and generate a keyword set of the user's existing skills.

[0154] According to the user's existing skill keyword set, the pre-trained word vector model is used to perform vector representation and addition processing on the user's existing skill keywords to obtain the semantic feature vector w′ of the user's existing skill keywords k .

[0155] The user's existing skill keyword set consists of a user-preset keyword set and an automatically extracted keyword set, wherein the automatically extracted keyword set is generated by extracting keywords from the user's existing skill description text.

[0156] According to the semantic feature vector of the user's existing skill keywords and the semantic feature vector of the target position's skill keywords, the following formula is used to calculate the similarity between the semantic feature vector of the user's existing skill keywords and the semantic feature vector of the position's skill keywords. The similarity is the semantic feature vector w′ of the user's existing skill keywords. k and the target job skill keyword semantic feature vector w k The cosine similarity between .

[0157] s k = CosSimilarity(w′ k ,w k )

[0158] According to the similarity between the semantic feature vector of the user's existing skill keywords and the semantic feature vector of the job skill keywords, a user target job skill matching report is generated, showing the user's existing skills and similarities with higher matching degrees. At the same time, the user's existing skills and similarities with lower matching degrees are shown, which are the skills that the user needs to further improve in order to match the target job.

[0159] Specifically, skills with low similarity are skills that users need to supplement, and the lower the similarity, the more they need to be supplemented.

[0160] Next, a target position skill description set to be supplemented is constructed. The target position skill description set to be supplemented is a set of target position skills and their skill description texts after removing the skills and their skill description texts whose similarity with the user's existing skills is greater than a similarity threshold from the target position skill description set. The similarity threshold is a similarity threshold preset by the user.

[0161] Step S104: According to the user's skill description set to be supplemented and the course training objective information set in the first recommended course resource data set, identify course resources with a high degree of course training skill matching, and generate a second recommended course resource data set.

[0162] Natural language processing technology is used to perform text analysis on the target position's skill description content in the target position's skill description set, automatically extract keywords related to the target position's skill to be supplemented, and generate a target position's skill keyword set.

[0163] According to the skill keyword set to be supplemented for the position, the pre-trained word vector model is used to perform vector representation and addition processing on the skill keywords to be supplemented for the target position, and the semantic feature vector w of the skill keywords to be supplemented for the target position is obtained. m The target position skill to be supplemented keyword set is composed of a user preset keyword set and an automatically extracted keyword set, wherein the automatically extracted keyword set is generated by extracting keywords from the description content of the skill to be supplemented in the target position skill to be supplemented description set.

[0164] For the course training objective information dataset, after removing the course training objective information of the course resources not included in the first recommended course resource list set, natural language processing technology is used to perform text parsing on the description of the course training skills, automatically extract course training skill-related keywords, and generate a course training skill keyword set.

[0165] According to the course training skill keyword set, the pre-trained word vector model is used to perform vector representation and addition processing on the course training skill keywords to obtain the course training skill keyword semantic feature vector w′ m Among them, the course training skill keyword set is composed of a user preset keyword set and an automatically extracted keyword set, wherein the automatically extracted keyword set is generated by extracting keywords from the course training skill description text.

[0166] According to the semantic feature vector of the course training skill keywords in the first recommended course resource data set and the semantic feature vector of the target position's to-be-supplemented skill keywords, the following formula is used to calculate the similarity between the semantic feature vector of the course training skill keywords in the first recommended course resource data set and the semantic feature vector of the target position's to-be-supplemented skill keywords. The similarity is the semantic feature vector w′ of the course training skill keywords. m The semantic feature vector w of the target job's skills keywords m The cosine similarity between .

[0167] s m = CosSimilarity(w′ m ,w m )

[0168] Display course resources and similarities that have a high degree of match with course development skills, and build a second list of recommended course resources.

[0169] Specifically, according to the similarity between the semantic feature vector of the course training skill keyword and the semantic feature vector of the skill keyword to be supplemented in the target position, the course resources with a high degree of course training skill matching are identified, and a second recommended course resource data set is generated.

[0170] The second recommended course resource list is a set of course resources from the first recommended course resource list except for course resources whose similarity to the course training skills is less than a similarity threshold. The similarity threshold is a similarity threshold preset by the user.

[0171] S105. Construct an online course resource evaluation hierarchical model based on the course evaluation text dataset in the second recommended course resource dataset or by collecting expert opinions through the Delphi method.

[0172] Among them, the course evaluation text data set in the second recommended course resource data set is a set of other course resource evaluation text data after removing the course resources whose similarity with the course training skills is less than the similarity threshold in the first recommended course resource data set.

[0173] In one embodiment of the present invention, key main indicators and related sub-indicators for online course resource evaluation are collected from experts through the Delphi method, and the effectiveness and applicability of these key main indicators and related sub-indicators are evaluated, so as to construct an online course resource evaluation hierarchical model.

[0174] Specifically, in one embodiment of the present invention, to construct an online course resource evaluation hierarchical model through the Delphi method, it is first necessary to identify a group of experts, including educational psychologists, educational technology experts, and teacher education scholars, who have rich knowledge and experience in the field of online course teaching and have professional background and knowledge in online course resource evaluation.

[0175] Specifically, in one embodiment of the present invention, constructing an online course resource evaluation hierarchical model by the Delphi method includes the following steps:

[0176] 1) Provide each expert with an initial questionnaire that lists a series of factors (main indicators) for evaluating online course resources, such as teaching style, testing methods, learning resource model, problem-solving process, and their related sub-indicators. Ask the experts to evaluate these key main indicators and related sub-indicators and provide opinions, including applicability and evaluability.

[0177] 2) Summarize expert opinions and evaluation results, eliminate individual extreme opinions, retain the consensus among experts, and compile them into a feedback report.

[0178] 3) Provide the feedback report to the experts and ask them to re-evaluate the previous evaluation main (sub) indicators or propose new evaluation main (sub) indicators. Repeat this process until a consensus is reached among the experts to form a hierarchical structure of online course resource evaluation standards, namely, the online course resource evaluation hierarchical model.

[0179] Alternatively, in another embodiment of the present invention, based on the course evaluation text set in the second recommended course resource data set, the teaching evaluation topics (sub-indicators) are identified by the teaching evaluation topic modeling method, and the identified teaching evaluation topics are integrated through expert classification to form a hierarchical structure of online course resource evaluation standards, that is, an online course resource evaluation hierarchical model.

[0180] Among them, the course evaluation text data set in the second recommended course resource data set is a set of other course resource evaluation text data after removing the course resources whose similarity with the course training skills is less than the similarity threshold from the first recommended course resource data set.

[0181] Specifically, the teaching evaluation topic modeling method includes teaching evaluation data preprocessing, teaching evaluation feature selection, teaching evaluation topic modeling analysis, optimal teaching evaluation topic model selection, and construction of a course resource evaluation hierarchical model based on the interpretation and integration of the results of the optimal teaching evaluation topic model. The specific steps are described as follows:

[0182] 1) Teaching evaluation data preprocessing

[0183] Optionally, in one embodiment of the present invention, the text preprocessing process based on the English course evaluation text dataset includes six steps: tokenization and exclusion of special characters, normalization, removal of stop words, and lemma restoration.

[0184] Specifically, each sentence is first extracted from the English course resource evaluation text and decomposed into a word list, while special characters and punctuation marks are removed; secondly, uppercase letters are converted to lowercase letters; then, stop words that appear frequently but cannot represent the key content of the course resource evaluation text (for example, "me", "I", "or") are removed; finally, the words are restored to their basic form (for example, "courses" is converted to "course"), thus obtaining the preprocessed course evaluation text set.

[0185] Optionally, in another embodiment of the present invention, text preprocessing is performed based on the Chinese course resource evaluation text collection, including word segmentation, part-of-speech tagging, removal of specific part-of-speech words and removal of stop words.

[0186] 2) Selection of teaching evaluation features

[0187] Based on the preprocessed course evaluation text dataset, the TF-IDF (term frequency–inverse document frequency) weighted technology is used to evaluate the importance of any word to one of the course evaluation texts in the course evaluation text dataset, thereby quantifying the relative importance of each word in the course evaluation text dataset and realizing the selection of teaching evaluation features (key words).

[0188] The TF-IDF value of the i-th word in the course evaluation text dataset is determined by the following formula.

[0189]

[0190] All words are sorted in descending order according to their TF-IDF values, and according to the TF-IDF threshold a preset by the user, words with TF-IDF values ​​lower than the threshold are removed, and words with TF-IDF values ​​not lower than the threshold are retained to form a course evaluation text topic modeling input data set. The course evaluation text topic modeling input data set includes D course evaluation texts and W unique words. The W unique words are obtained by removing the words with TF-IDF values ​​lower than the threshold from the V unique words.

[0191] 3) Teaching Evaluation Topic Modeling Analysis

[0192] Based on the course evaluation text topic modeling input dataset, the Structural Topic Modeling (STM) algorithm is used to identify important evaluation factors and indicators in the course evaluation text. Given a course evaluation text topic modeling input dataset consisting of D course resource evaluation texts and W unique words, d∈{1,2,...,D} represents the dth course evaluation text, k∈{1,2,...,K} represents the kth evaluation topic, and document d consists of n words, n∈{1,2,...,N}. The main goal of STM modeling is to determine two distributions: evaluation text-evaluation topic distribution and evaluation topic-evaluation text word distribution, denoted as θ and β respectively. Among them, θ d represents the proportion of hidden topics in each document, and β d,k,v represents the evaluation topic - the distribution of words in the evaluation text, z d,n Represents the topic distribution of each word, w d,n represents a word selected from v∈{1,2,...,W}.

[0193] In this study, STM was implemented using the R package stm. By setting the range of the number of topics k, for example, k∈{1,2,…,K}, structural topic modeling was performed to obtain K alternative teaching evaluation topic models.

[0194] 4) Selection of the optimal teaching and evaluation topic model

[0195] Based on the modeling results of K alternative teaching evaluation topic models, the exclusivity and semantic coherence indicators are used to select the teaching evaluation topic model. i ,v j ) is the word v i and v j The number of times it appears in the course evaluation text d at the same time. According to the following formula, the semantic consistency value for the kth teaching evaluation topic is calculated, where M is the first M words with the highest correlation with the kth teaching evaluation topic.

[0196]

[0197] In one embodiment of the present invention, FREX (Frequency and Exclusivity) is used to quantify the degree of exclusivity of a given evaluation topic while considering word frequency balance. The following formula is used to calculate the weighted harmonic mean of the ranking of the vth word in the kth evaluation topic on the exclusivity and word frequency indicators. Wherein, ECDF represents the empirical cumulative distribution function, ω is the weight (usually set to 0.7 to give priority to exclusivity), k∈K represents the kth topic, v represents the word to be evaluated, and β represents the evaluation topic word distribution of the kth evaluation topic.

[0198]

[0199] In one embodiment of the present invention, the manyTopics function in the stm package of the R programming language is used to calculate the semantic consistency and exclusivity values ​​of each teaching evaluation topic in the teaching evaluation topic model. Then, the semantic consistency and exclusivity values ​​are averaged to obtain the semantic consistency and exclusivity values ​​of the teaching evaluation model.

[0200] The above calculation method is used to calculate the semantic consistency and exclusivity values ​​of K alternative teaching evaluation topic models, and a two-dimensional coordinate graph is drawn to intuitively compare the performance of each alternative teaching evaluation topic model in terms of semantic consistency and exclusivity indicators.

[0201] In one embodiment of the present invention, drawing Figure 4 The two-dimensional coordinate graph shown in Figure 1 shows that the x-axis represents the evaluation topic model with k (1, 2, ..., 30) evaluation topics, the left side of the y-axis represents the semantic consistency index value, and the right side of the y-axis represents the exclusivity index value. Figure 4 It can be observed that when k = 16, the teaching evaluation topic model simultaneously weighs the performance on semantic consistency and exclusivity indicators, so it is taken as the optimal teaching evaluation topic model.

[0202] 5) Constructing a hierarchical model for course resource evaluation based on the interpretation and integration of the results of the optimal teaching evaluation theme model

[0203] Based on the modeling results of the optimal teaching evaluation topic model, two experienced online education experts, combined with the semantic similarity between highly correlated words in each teaching evaluation topic, derived and annotated and integrated the labels of each teaching evaluation topic in the optimal model (i.e., the online course resource evaluation sub-indicators) to form a hierarchical structure of online course resource evaluation standards, namely, the online course resource evaluation hierarchical model.

[0204] Specifically, the embodiment of the present invention generates topic modeling results based on a collection of computer science, engineering, and programming course evaluation texts. According to the semantic similarity and correlation between highly associated words in each evaluation topic, after multiple rounds of in-depth discussions, each evaluation topic is labeled to form 16 evaluation topic labels.

[0205] Based on each teaching evaluation topic label, the embodiment of the present invention classifies and integrates each teaching evaluation topic label according to the semantic similarity and correlation between each teaching evaluation topic label to form a general theme (i.e., the main indicator of online course resource evaluation), thereby constructing an online course resource evaluation hierarchical model, which is composed of the online course resource evaluation main indicator and its associated online course resource evaluation sub-indicators.

[0206] Specifically, in one embodiment of the present invention, Figure 5 As shown, the embodiment of the present invention classifies and integrates 16 evaluation topic labels obtained by subject modeling based on the collection of computer science, engineering and programming course evaluation texts, forming 8 general topics (i.e., the main evaluation indicators of computer science, engineering and programming online course resources), thereby constructing a hierarchical model of computer science, engineering and programming online course resources evaluation, which is composed of the main evaluation indicators of computer science, engineering and programming online course resources and their associated online course resource evaluation sub-indicators. Among them, the main indicator of "course evaluation" is composed of two sub-indicators: "homework" and "exam". Among them, the main indicator of "course teaching" is composed of two sub-indicators: "degree of answering questions and solving doubts" and "clarity of explanation". The main indicator of "perceived effort" is composed of two sub-indicators: "perceived time of effort" and "perceived energy of effort". The main indicator of "pleasure" is composed of two sub-indicators: "pleasure degree" and "satisfaction degree". The main indicator of "teacher" is composed of two sub-indicators: "teacher image" and "teaching style". The main indicator of "interactivity" is composed of two sub-indicators: "teacher-student interaction" and "student-student interaction". The main indicator of "technicality" consists of two sub-indicators: "complexity" and "flexibility". The main indicator of "usefulness" consists of two sub-indicators: "effectiveness" and "professional development".

[0207] S106. According to the course resource evaluation hierarchical model structure, the evaluation indicators are weighted through topic modeling analysis or expert scoring method, and the course resources in the second recommended course resource data set are comprehensively evaluated and scored to generate a third recommended course resource data set.

[0208] For the constructed online course resource evaluation hierarchical model, it is necessary to measure the relative importance between the main evaluation indicators and evaluation sub-indicators in the evaluation model to identify and evaluate the relative weights of the various main evaluation indicators and evaluation sub-indicators that affect the online course resource evaluation. Then, based on the constructed online course resource evaluation hierarchical model and the determined relative weights of the various main evaluation indicators and evaluation sub-indicators, the hierarchical analysis method is used to rank the various course resources in the second recommended course resource list to determine the overall relative advantage ranking of each course resource and the relative advantage ranking of each evaluation main indicator and evaluation sub-indicator of each course resource.

[0209] Based on the determined overall relative advantage ranking of each course resource and the relative advantage ranking of each course resource on each evaluation main indicator and evaluation sub-indicator, the user selects several course resources with high relative advantage ranking in the overall relative advantage ranking or on each evaluation main indicator or evaluation sub-indicator according to the custom threshold as the third course resource recommendation list. Through the analysis of the overall relative advantage ranking of each course resource and the relative advantage ranking of each course resource on each evaluation main indicator and evaluation sub-indicator, the scope of recommended course resources can be further narrowed to achieve more refined and personalized course resource recommendations.

[0210] In one embodiment of the present invention, the relative weights of each evaluation main index and evaluation sub-index in the online course resource evaluation hierarchical model are calculated by using the association matrix data of the evaluation text and the evaluation topic in the optimal evaluation topic model result. Specifically, two steps are included: weighting the course resource evaluation index and obtaining the third recommended course resource data set based on the online course resource evaluation hierarchical model.

[0211] 1) Course resource evaluation indicators empowerment

[0212] According to the θ matrix estimated by the teaching evaluation topic modeling, the following formula is used to calculate the popularity of the kth teaching evaluation topic in the course evaluation text set associated with the course resources in the second recommended course resource dataset, denoted as P k , where θ d,k (d=1,...,D,k=1,...,K) is the correlation degree between the dth course evaluation text and the kth teaching evaluation topic, D is the number of teaching evaluation texts in the course evaluation text set, and K is the number of topics fitted by the teaching evaluation topic modeling.

[0213]

[0214] Next, for P k (k=1,...,K) is normalized to obtain P′ k (k=1,...,K) is used as the relative weight of each evaluation sub-indicator (i.e., evaluation topic) in the constructed online course resource evaluation hierarchical model. The weights of the evaluation sub-indicators corresponding to each evaluation main indicator are summed up by the following formula to obtain the relative weight of each evaluation main indicator.

[0215]

[0216] In one embodiment of the present invention, according to the course resource evaluation hierarchical model structure, by using the Likert seven-level scale for level comparison, based on the main indicators and related sub-indicators in the online course resource evaluation hierarchical model, according to the relative importance scaling method, relevant experts compare the indicators in the pairwise comparison matrix of the same level respectively, and give their relative importance judgment values ​​to form a comparison judgment matrix, and then calculate the relative weights of each main evaluation indicator and evaluation sub-indicator in the online course resource evaluation hierarchical model.

[0217] Specifically, in one embodiment of the present invention, according to the course resource evaluation hierarchical model structure, the expert scoring method is adopted, and the opinions of 10 relevant experts are anonymously solicited, wherein the 10 relevant experts all have online course teaching experience. Then, the 10 relevant experts use the Likert seven-level scale to compare the levels, based on the 8 main indicators of the first level and the 16 sub-indicators of the second level in the constructed course resource evaluation hierarchical model, according to the scaling method in the relative importance scale, the experts compare the indicators in the pairwise comparison matrix of the same level respectively, and give their relative importance judgment values, and after all indicators are judged pairwise, a comparison judgment matrix is ​​formed. In one embodiment of the present invention, the scaling method used includes 7 levels, wherein "1" indicates that two elements are compared, which have equal importance; "3" indicates that two elements are compared, the former is moderately important than the latter; "5" indicates that two elements are compared, the former is strongly important than the latter; "7" indicates that two elements are compared, the former is extremely important than the latter; "2", "4", "6" represent the intermediate values ​​of the above adjacent comparisons.

[0218] After collecting the scoring data of all experts, the average value of each comparison level is calculated to obtain a specific comparison matrix. The matrix data is calculated using the sum-product method, and the obtained eigenvector is the weight of the corresponding level. Considering the subjectivity of experts in making judgments, all comparison matrices need to be verified for consistency to ensure the usability of the scale. After calculating the relative weights of the indicators, the following formula and random consistency RI (Random Index) are used to perform consistency checks on the judgment matrix. Among them, λ max is the maximum eigenvalue, and n is the order of the comparison matrix.

[0219]

[0220] 2) Obtain the third recommended course resource dataset based on the online course resource evaluation hierarchical model

[0221] Based on the online course resource evaluation hierarchical model and the relative weights of each main evaluation indicator and evaluation sub-indicator in the online course resource evaluation hierarchical model, the course resources in the second recommended course resource data set are comprehensively evaluated to obtain a comprehensive evaluation score, and a third recommended course resource data set is generated.

[0222] Specifically, based on the course evaluation text set associated with the course resources in the second recommended course resource data set and the course evaluation grade data (1, 2, ..., 5; 1 is the lowest score) of the users associated with the course resources in the second recommended course resource data set, and the evaluation text-topic distribution matrix data estimated by the optimal evaluation topic model result, the score value of the c-th course learning user on the course resource in the k-th evaluation topic (i.e., evaluation sub-indicator) is calculated according to the following formula. Where, r c is the evaluation level of course resources given by course learning users; p cm It is the numerical value of the correlation degree between the course evaluation text of the c-th course learning user and the k-th teaching evaluation topic. The numerical value of the correlation degree between the course evaluation text of the c-th course learning user and the k-th teaching evaluation topic is directly obtained from the teaching evaluation text-topic distribution matrix data in the result estimation of the optimal teaching evaluation topic model.

[0223]

[0224] The following formula calculates the score value of the c-th course learning user for the main indicator of online course resource evaluation, which is obtained by summing the scores of the c-th course learning user for the associated evaluation sub-indicators.

[0225]

[0226] According to the score values ​​of the main indicators of online course resource evaluation given by the course learning users and the relative weights of the main evaluation indicators in the online course resource evaluation hierarchical model, the comprehensive evaluation scores of the course resources in the second recommended course resource data set are calculated by the weighted summation method, and then the comprehensive scores of the target courses are arranged in descending order to obtain the third recommended course resource data set.

[0227] The third recommended course resource data set is a set of other course resources in the second recommended course resource data set except for the course resources with comprehensive evaluation scores less than the similarity threshold.

[0228] S107 . According to the user's learning history information set and the course information and knowledge point information set in the third recommended course resource data set, identify course resources that cultivate high-priority and high-importance knowledge points, and generate a fourth recommended course resource data set.

[0229] 1) Determine the priority of knowledge points for users

[0230] According to the course information and knowledge point information set in the third recommended course resource data set and the key knowledge point information set of the user's learning history course resources, the priority of each knowledge point for the target user is determined. Specifically, the following steps 1-3 are included:

[0231] Step 1: Calculate the second similarity based on the feature vector of the course resource knowledge point information;

[0232] Specifically, a course resource key knowledge point information set is extracted from the course resource directory content data in the third recommended course resource data set, wherein the key knowledge point information set is composed of a user-preset key knowledge point information set and an automatically extracted key knowledge point information set.

[0233] The automatically extracted key knowledge point information set is obtained by automatically extracting key knowledge point information after text preprocessing the content of the course resource directory in the third recommended course resource dataset.

[0234] Among them, text preprocessing includes word segmentation, stop word removal, noise filtering and word ordering.

[0235] Specifically, in an embodiment of the present invention, the JE-analysis word segmentation tool is used to segment the content data of the course resource directory in the third recommended course resource data set, and extract key knowledge point information points to obtain a course resource key knowledge point information set, and then the pre-trained model is used to extract the course resource key knowledge point information feature vector w in the third recommended course resource list. m .

[0236] Secondly, the user's learned knowledge point information set is extracted according to the course teaching resource directory content corresponding to the user's learned knowledge point information in the user's learning history information set, wherein the knowledge point information set is composed of the user's preset knowledge point information set and the automatically extracted knowledge point information set.

[0237] The automatically extracted knowledge point information set is obtained by automatically extracting key knowledge point information after text preprocessing of the content of the course teaching resource directory corresponding to the user's learned knowledge point information in the user's learning history information set.

[0238] Among them, text preprocessing includes word segmentation, stop word removal, noise filtering and word ordering.

[0239] Specifically, in an embodiment of the present invention, the JE-analysis word segmentation tool is used to segment the content data of the course teaching resource directory corresponding to the user's learned knowledge point information in the user's learning history information set, and extract the knowledge point information points to obtain the user's learned knowledge point information set, and then extract the user's learned knowledge point information corresponding to the course teaching resource course resource knowledge point information feature vector w' through the pre-training model m .

[0240] According to the first similarity between the feature vector of key knowledge point information of course resources in the third recommended course resource list and the feature vector of knowledge point information of course teaching resources corresponding to the knowledge point information learned by the user:

[0241] Among them, the cosine similarity is the feature vector of key knowledge point information of the course resources in the third recommended course resource list, w m The feature vector w' of the course teaching resource knowledge point information corresponding to the user's learned knowledge point information m The cosine similarity between:

[0242] S m =CosSimilarity(w' m ,w m )

[0243] Step 2, calculating the second similarity based on the keyword information feature vector of the teaching content of the teaching resource;

[0244] Specifically, a course resource teaching content keyword information set is extracted from the teaching content data of the course resources in the third recommended course resource data set (including course teaching materials text, course teaching video content text, and course assessment content text), wherein the course resource teaching content keyword information set is composed of a user-preset course resource teaching content keyword information set and an automatically extracted course resource teaching content keyword information set.

[0245] The automatically extracted course resource teaching content keyword information set is obtained by automatically extracting the course resource teaching content keyword information after text preprocessing the course resource teaching content data in the third recommended course resource data set.

[0246] Among them, text preprocessing includes word segmentation, stop word removal, noise filtering and word ordering.

[0247] Specifically, in an embodiment of the present invention, the JE-analysis word segmentation tool is used to segment the teaching content data of the course resources in the third recommended course resource data set, and the keyword information of the course resource teaching content is extracted to obtain a course resource teaching content keyword information set, and then the pre-trained model is used to extract the course resource teaching content keyword information feature vector w in the third recommended course resource list.k .

[0248] Secondly, based on the teaching content data of the course resources corresponding to the user's learned knowledge points in the user's learning history information set (including course teaching materials text, course teaching video content text, and course assessment content text), the user's learned content keyword information set is extracted, where the user's learned course resource keyword information set is composed of the user's learned content keyword information and the automatically extracted content keyword information set.

[0249] The automatically extracted content keyword information set is obtained by automatically extracting the content keyword information after text preprocessing of the teaching content data of the course resources corresponding to the user's learned knowledge point information in the user's learning history information set.

[0250] Among them, text preprocessing includes word segmentation, stop word removal, noise filtering and word ordering.

[0251] Specifically, in an embodiment of the present invention, the JE-analysis word segmentation tool is used to segment the teaching content data of the course resources corresponding to the user's learned knowledge point information in the user's learning history information set, and extract the user's learned content keyword information to obtain the user's learned content keyword information set, and then extract the user's learned content keyword information vector w' through the pre-training model k .

[0252] According to the first similarity between the keyword information feature vector of the course resource teaching content in the third recommended course resource list and the keyword information vector of the user's learned content:

[0253] Among them, the cosine similarity is the feature vector of the course resource keyword information in the third recommended course resource list, w k and the user's learned content keyword information vector w' k The cosine similarity between:

[0254] S k =CosSimilarity(w' k ,w k )

[0255] Step 3, calculating a first comprehensive similarity according to the first similarity and the second similarity;

[0256] Specifically, in the embodiment of the present invention, based on the weight coefficient a of the course meta-information feature vector and the weight coefficient b of the keyword semantic feature vector, the comprehensive similarity is calculated according to the first similarity and the second similarity.

[0257]

[0258] According to the comprehensive similarity, the priority of the knowledge point for the user is determined. The judgment basis is: the higher the comprehensive similarity, the lower the priority. Therefore, the priority of the knowledge point for the user is calculated as: p = 1-s. Arrange the priority of each knowledge point for the user in descending order, and construct the priority vector of the knowledge point for the user;

[0259] 2) Determine the importance of the knowledge point to the course

[0260] First, the importance of each chapter in the target course is calculated based on the proportion of each chapter in the course teaching resources. The proportion of course teaching resources includes the proportion of each chapter in the number of pages of the target course teaching materials, the proportion of each chapter in the number of hours of the target course teaching video, and the proportion of each chapter in the score of the target course assessment content.

[0261] Specifically, the importance of each chapter in the target course is calculated according to the following formula. c,i is the importance of each chapter i in the target course c, mf c,i is the page number of the teaching material for chapter i in the cth target course, is the total number of pages of the teaching materials for the cth target course, vf c,i The length of the video for the lecture of the i-th chapter in the c-th target course, The total duration of the video for the cth target course, tf c,i is the score of the assessment content of chapter i in target course c, It is the total score of the assessment content of the cth target course.

[0262]

[0263] Next, the importance of each knowledge point in each chapter is calculated according to the number of times each knowledge point appears in the teaching material information of each chapter. The number of times each knowledge point appears in the teaching material information of each chapter includes the frequency of each knowledge point appearing in the teaching material text of each chapter in each target course, and the frequency of each knowledge point appearing in the teaching explanation video text of each chapter in each target course.

[0264] Specifically, the importance of each knowledge point in each chapter of the target course is calculated according to the following formula. c,k,i is the importance of the kth knowledge point in the i-th chapter of the target course c, mf k,i is the frequency of occurrence of the kth knowledge point in the text of the i-th chapter of the target course c. vf is the total frequency of all knowledge points in the textbook text of chapter i in target course c.k,i The frequency of occurrence of the kth knowledge point in the video text of the lecture of the i-th chapter in the target course c. The total frequency of all knowledge points in the video text of the lecture of chapter i in target course c.

[0265]

[0266] Based on the importance of each chapter in the course resources and the importance of each knowledge point information in each chapter, the importance of each knowledge point in the course resources is calculated.

[0267] Specifically, the importance CK of knowledge point k in the cth target course is calculated according to the following formula: c,k The importance of the kth knowledge point in the target course c is that the larger the proportion, the more important it is.

[0268] CK c,k =IK c,k,i ×IC c,i

[0269] Arrange each target course in descending order according to the importance of each knowledge point in the course resources, and obtain the importance vector of the knowledge points in the candidate courses.

[0270] Normalize the importance vector of the candidate course resource knowledge points and the priority vector of the knowledge points for the user and then multiply them to obtain the comprehensive priority vector of each knowledge point in the course resource;

[0271] The comprehensive priority vectors of learning each knowledge point in the course resources are summed up to obtain the comprehensive priority of learning all knowledge points in the course resources;

[0272] Arrange each course resource in the third recommended course resource list in descending order according to the comprehensive priority of learning all knowledge points of the course resources to obtain a fourth recommended course resource list.

[0273] The fourth recommended course resource data set is a set of other course resources in the third recommended course resource data set except for course resources whose comprehensive priority of learning all knowledge points of course resources is less than the comprehensive priority threshold.

[0274] In some embodiments, after generating a learning recommendation report for career development according to the first recommended course resource list, the second recommended course resource list, the third recommended course resource list, and the fourth recommended course resource list, the learning recommendation method for career development further includes:

[0275] Present a report of recommended learning for career development.

[0276] The above results will be fed back to the online course platform or corporate training platform so that they can adjust the course resource recommendation strategy in a timely manner, pay attention to the matching degree between the user's own skills and the specific market job requirements, and guide and promote users to improve the matching degree between their own skills and the specific market job requirements.

[0277] In summary, the learning recommendation method, system, device and storage medium for career development proposed in the embodiments of the present invention. By utilizing big data analysis technology, machine learning technology, and comprehensively considering the dynamic changes in the job market's demand for job skills, online course resource design and implementation factors, and the user's current knowledge and skills, a complex course recommendation method suitable for career development is developed. In addition, the learning recommendation method proposed in the present invention will integrate real-time labor market analysis to ensure consistency with the latest job market needs. Furthermore, it will evaluate the importance and priority of knowledge points for each user, thereby achieving fine-grained course resource recommendations. The present invention aims to improve the accuracy and relevance of course resource recommendations, promote more effective career planning and skill acquisition, and at the same time cultivate a more adaptable workforce by being consistent with job market needs.

[0278] The embodiment of the present invention further provides a learning recommendation device for career development, comprising:

[0279] The first module is used to obtain a first recommended course resource list according to user preset indicators and construct a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set;

[0280] The second module is used to analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and construct a position skill knowledge graph;

[0281] The third module is used to identify and construct a set of user skill descriptions to be supplemented based on the job skill knowledge graph and the user's existing skill description corpus;

[0282] The fourth module is used to identify course resources with high course training skill matching according to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, and generate a second recommended course resource data set;

[0283] A fifth module is used to construct a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or expert opinions collected through the Delphi method;

[0284] The sixth module is used to weight the evaluation indicators through topic modeling analysis or expert scoring method according to the model structure of the course resource evaluation hierarchical model, and to comprehensively evaluate and score the course resources in the second recommended course resource data set to generate a third recommended course resource data set;

[0285] The seventh module is used to identify course resources that cultivate high-priority and high-importance knowledge points based on the user's learning history information set and the third recommended course resource data set, and generate a fourth recommended course resource data set.

[0286] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0287] The embodiment of the present invention further provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above-mentioned learning recommendation method for career development when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0288] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0289] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned learning recommendation method for career development when executed by a processor.

[0290] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0291] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0292] It should be noted that in various specific embodiments of the present invention, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present invention needs to obtain the user's sensitive personal information, it will obtain the user's separate permission or consent through a pop-up window or jump to a confirmation page, and after clearly obtaining the user's separate permission or consent, it will obtain the necessary user-related data for the normal operation of the embodiment of the present invention.

[0293] The embodiments described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art can appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0294] Those skilled in the art will appreciate that the technical solutions shown in the figures do not limit the embodiments of the present invention and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0295] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0296] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0297] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0298] It should be understood that in the present invention, "at least one (item)" refers to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can represent: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0299] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0300] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0301] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0302] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including multiple instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store programs.

[0303] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the embodiments of the present invention is not limited thereby. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present invention shall be within the scope of the rights of the embodiments of the present invention.

Claims

1. A learning recommendation method for career development, characterized in that: The following steps are involved: Obtaining a first recommended course resource list according to user preset indicators, and constructing a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set; Analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and build a position skill knowledge graph; According to the job skill knowledge graph and the user's existing skill description corpus, identify and construct a set of user skill descriptions to be supplemented; According to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, identifying course resources with a high degree of course training skill matching, and generating a second recommended course resource data set; Constructing a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or collecting expert opinions through the Delphi method; According to the model structure of the course resource evaluation hierarchical model, the evaluation indicators are weighted by subject modeling analysis or expert scoring method, and the course resources in the second recommended course resource data set are comprehensively evaluated and scored to generate a third recommended course resource data set; According to the user's learning history information set and the third recommended course resource data set, course resources that cultivate high-priority and high-importance knowledge points are identified to generate a fourth recommended course resource data set.

2. A learning recommendation method for career development according to claim 1, characterized in that: The method of obtaining a first recommended course resource list according to user preset indicators and constructing a first recommended course resource data set includes the following steps: Acquire the first recommended course resource list from the course resource database according to a first preset rule; wherein the first preset rule includes a matching condition for screening different attributes in the course resource corpus metadata of the course resource database; Constructing the first recommended course resource data set according to the course resource meta information, course training objectives, course resource description and catalog content, and course evaluation information in the first recommended course resource list; Among them, the course resource meta-information is used to describe the basic information of the course resources; the course training objectives are used to describe the training skill information of the course resources; the course resource description and directory content are used to describe the specific information contained in the course resource material content document; the course resource material content document is used for the keyword information of the course resource description and directory content; the course evaluation information is used to describe the user's evaluation level of the course, the course evaluation user and the course evaluation text information.

3. A learning recommendation method for career development according to claim 1, characterized in that: The method of performing position and skill association analysis based on the position information corpus, identifying position skills with strong association, and constructing a position skill knowledge graph includes the following steps: Obtain a job information corpus, and perform text preprocessing on the job information in the job information corpus to obtain preprocessed job information; the text preprocessing includes: using the jieba word segmentation library to perform word segmentation processing on the text data of occupation-related information, dividing the continuous text sequence into words or phrases, and obtaining a word sequence; cleaning the word sequence to obtain a cleaned word sequence; extracting text feature data based on the job information according to the word sequence, and obtaining a text encoding feature vector of the job information through a pre-trained BERT model; based on semantic analysis means, find out job skills in combination with contextual relationships, form a job skill set, and convert each job in the job skill set into a job skill vector based on the BERT model; According to the position vector and the position skill vector, an association matrix between data items is established, the association between data items is calculated according to co-occurrence or correlation, frequent item sets are discovered using an association rule mining algorithm, association rules are generated based on the frequent item sets, and the association relationship between each position and position skill is generated; According to the relationship between each position and position skills, a graph structure is constructed, and then a position skill knowledge graph of the relationship between positions and position skills is constructed.

4. A learning recommendation method for career development according to claim 3, characterized in that: The method of establishing an association matrix between data items based on the position vector and the position skill vector, calculating the association between data items based on co-occurrence or correlation, discovering frequent item sets using an association rule mining algorithm, generating association rules based on the frequent item sets, and generating the association relationship between each position and position skill includes the following steps: The BERT model obtained from the job description and skill requirement text embedding representation data is used to build a correlation matrix between data items; Construct a correlation matrix by calculating the co-occurrence or correlation between the data represented by each embedding; Use the Apriori algorithm to mine frequent item sets on job descriptions and skill requirements data items, discover frequent item sets, generate association rules based on frequent item sets, and generate the association relationship between each position and job skills; The process of mining frequent itemsets includes the following steps: Extract features related to association rule mining from the collected job description and skill requirement data, and set the minimum support threshold and minimum confidence threshold; Convert the processed job description and skill requirement data into the input format of the Apriori algorithm, and determine the minimum support threshold and the minimum confidence threshold; Treat a single feature as a candidate frequent item set, calculate its support, and filter out the frequent item sets that meet the conditions according to the minimum support threshold; Combine the filtered frequent item sets in pairs and calculate the support of the newly generated candidate frequent item sets; According to the minimum support threshold, filter out the frequent item sets that meet the conditions, and iterate until no new frequent item sets can be generated, and obtain all frequent item sets that meet the minimum support threshold conditions; Generate association rules based on frequent item sets, calculate the confidence of each association rule, and filter out strong association rules that meet the conditions according to the minimum confidence threshold; The method of generating association rules based on frequent item sets, calculating the confidence of each association rule, and selecting strong association rules that meet the conditions according to the minimum confidence threshold comprises the following steps: Traverse all frequent item sets and split each frequent item set into antecedents and consequents; For each combination of antecedent and consequent, calculate the confidence of the association rule; According to the minimum confidence threshold, strong association rules that meet the conditions are screened out, and association rules with confidence less than the minimum confidence threshold are deleted; After generating association rules based on frequent item sets, strong association rules are screened out by calculating confidence and using the minimum confidence threshold; For the screened position-skill strong association rules, the evaluation indicators of lift, total confidence and Kulc index are calculated, and they are sorted according to the evaluation indicators to identify meaningful positions and position-skill associations.

5. A learning recommendation method for career development according to claim 1, characterized in that: The step of identifying and constructing a set of user skill descriptions to be supplemented based on the job skill knowledge graph and the user's existing skill description corpus includes the following steps: Extracting nodes and edges related to skills from the job skill knowledge graph, using natural language processing technology to perform text parsing on the description of the target job skill node, automatically extracting keywords related to the target job skill, and generating a target job skill keyword set; wherein the target job skill keyword set is composed of a user-preset keyword set and an automatically extracted keyword set; the automatically extracted keyword set is generated by extracting keywords from the description content of the target job skill node; According to the description content of the user's existing skills in the user's existing skill description corpus, natural language processing technology is used to perform text analysis on the description content of the user's existing skills, automatically extract keywords related to the user's existing skills, and generate a keyword set of the user's existing skills; According to the user's existing skill keyword set, a pre-trained word vector model is used to perform vector representation and addition processing on the user's existing skill keywords to obtain the semantic feature vector of the user's existing skill keywords; Calculating the similarity between the semantic feature vector of the user's existing skill keywords and the semantic feature vector of the target position's skill keywords; Generate a user target position skill matching report based on the similarity, and display the existing skills and similarities of users with higher matching degrees; Construct a set of skill descriptions to be supplemented for the target position; wherein the set of skill descriptions to be supplemented for the target position is a set consisting of other target position skills and their skill description texts after removing the skills and their skill description texts whose similarity with the user's existing skills is greater than a similarity threshold from the target position skill description set.

6. A learning recommendation method for career development according to claim 1, characterized in that: The step of identifying course resources with a high degree of course training skill matching according to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, and generating a second recommended course resource data set, comprises the following steps: Use natural language processing technology to perform text analysis on the target position's skill description content in the user's skill description set to be supplemented, automatically extract keywords related to the target position's skill to be supplemented, and generate a target position's skill keyword set to be supplemented; According to the target position skill keyword set to be supplemented, a pre-trained word vector model is used to perform vector representation and addition processing on the target position skill keywords to be supplemented therein, so as to obtain a semantic feature vector of the target position skill keywords to be supplemented; For the course training goal data set, after removing the course training goal information of the course resources not included in the first recommended course resource list set, natural language processing technology is used to perform text parsing on the description of the course training skills, and course training skill related keywords are automatically extracted to generate a course training skill keyword set; According to the course training skill keyword set, the pre-trained word vector model is used to perform vector representation and addition processing on the course training skill keywords therein to obtain the semantic feature vector of the course training skill keywords; Calculate the similarity between the semantic feature vectors of the course training skill keywords in the first recommended course resource dataset and the semantic feature vectors of the target position's skill keywords to be supplemented, based on the semantic feature vectors of the course training skill keywords in the first recommended course resource dataset and the semantic feature vectors of the target position's skill keywords to be supplemented; Display course resources and similarities with high course training skill matching, and build a second recommended course resource dataset.

7. A learning recommendation method for career development according to claim 1, characterized in that: The step of weighting evaluation indicators by subject modeling analysis or expert scoring method according to the model structure of the course resource evaluation hierarchical model and comprehensively evaluating and scoring the course resources in the second recommended course resource data set to generate a third recommended course resource data set includes the following steps: Based on the constructed course resource evaluation hierarchical model and the determined relative weights of each evaluation main index and evaluation sub-index, the hierarchical analysis method is used to rank each course resource in the second recommended course resource data set to determine the overall relative advantage ranking of each course resource and the relative advantage ranking of each evaluation main index and evaluation sub-index of each course resource; Based on the determined overall relative advantage ranking of each course resource and the relative advantage ranking of each course resource in terms of each main evaluation indicator and evaluation sub-indicator, according to the custom threshold, several course resources with high relative advantage ranking in the overall relative advantage ranking or in terms of each main evaluation indicator or evaluation sub-indicator are selected as the third recommended course resource data set.

8. A learning recommendation method for career development according to claim 1, characterized in that: The step of identifying course resources that cultivate high-priority and high-importance knowledge points based on the user's learning history information set and the third recommended course resource data set, and generating a fourth recommended course resource data set, comprises the following steps: Determine the priority of the knowledge point for the user and determine the priority vector; Determine the importance of the knowledge point to the course and determine the importance vector; Determine the comprehensive priority vector of each knowledge point in the course resource according to the priority vector and the importance vector; The comprehensive priority vectors of learning each knowledge point in the course resources are summed up to obtain the comprehensive priority of learning all knowledge points in the course resources; The course resources in the third recommended course resource data set are arranged in descending order according to the comprehensive priority of learning all knowledge points of the course resources to obtain a fourth recommended course resource data set.

9. A learning recommendation device for career development, characterized in that: include: The first module is used to obtain a first recommended course resource list according to user preset indicators and construct a first recommended course resource data set, wherein the first recommended course resource data set includes a course resource meta information data set, a course training goal data set, a course resource description and catalog content data set, and a course evaluation information data set; The second module is used to analyze the correlation between positions and skills based on the position information corpus, identify the position skills with strong correlation, and construct a position skill knowledge graph; The third module is used to identify and construct a set of user skill descriptions to be supplemented based on the job skill knowledge graph and the user's existing skill description corpus; The fourth module is used to identify course resources with high course training skill matching according to the user's skill description set to be supplemented and the course training target data set in the first recommended course resource data set, and generate a second recommended course resource data set; A fifth module is used to construct a course resource evaluation hierarchical model based on the course evaluation information data set in the second recommended course resource data set or expert opinions collected through the Delphi method; The sixth module is used to weight the evaluation indicators through topic modeling analysis or expert scoring method according to the model structure of the course resource evaluation hierarchical model, and to comprehensively evaluate and score the course resources in the second recommended course resource data set to generate a third recommended course resource data set; The seventh module is used to identify course resources that cultivate high-priority and high-importance knowledge points based on the user's learning history information set and the third recommended course resource data set, and generate a fourth recommended course resource data set.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 1 to 8.

Citation Information

Cited By

  • Automatic teaching evaluation text generation method

    CN119203975A

  • Micro course generation method based on OCR (Optical Character Recognition) and server

    CN119580289A

  • Network teaching information automatic extraction system and method based on cloud computing

    CN120067324A

  • Automatic Extraction System and Method for Network Teaching Information Based on Cloud Computing

    CN120067324B

  • Artificial intelligence-based adaptive learning path training management system and method thereof

    CN120146801A