Scientific research training resource data matching method and system based on knowledge graph

Abstract

The application provides a scientific research training resource data matching method and system based on a knowledge graph, relates to the technical field of knowledge graphs, and through the construction of a scientific research knowledge association structure containing entity nodes, association edges and attribute labels, the feature extraction of scientific research personnel data and training resource data is carried out, a personnel feature set and a resource feature set are generated, the association edge weight of the scientific research knowledge association structure is used for cross-dimension matching calculation, a matching correlation degree matrix is generated, and according to the matching correlation degree matrix and the dynamic updating rule of the scientific research knowledge association structure, a scientific research training resource matching result is generated, so that the accurate characterization and cross-dimension matching of the characteristics of scientific research personnel and training resources are realized, the demand change and resource update can be responded to in real time, personalized and accurate training resource recommendation is provided for scientific research personnel, and the effect and efficiency of scientific research training are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge graph, in particular to a scientific research training resource data matching method and system based on knowledge graph. BACKGROUND

[0002] In the field of scientific research, effective matching of scientific research training resources is crucial for improving the professional ability and research efficiency of scientific researchers. However, traditional methods of matching scientific research training resources often rely on manual screening and experience-based judgment, which is not only inefficient, but also difficult to fully and accurately grasp the actual needs of scientific researchers and the characteristics of training resources. With the continuous accumulation of scientific research data and the rapid development of knowledge graph technology, how to use knowledge graph to construct scientific research knowledge association structure and realize intelligent matching of scientific research training resources based on the structure has become a problem to be solved. Although existing resource matching methods attempt to introduce knowledge graph technology, they mostly only stay at the level of simple information retrieval and recommendation, lacking in-depth mining of the characteristics of scientific researchers and training resources and cross-dimensional matching calculation, resulting in inaccurate matching results and failing to meet the diverse training needs of scientific researchers. SUMMARY

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide a scientific research training resource data matching method based on knowledge graph, which comprises:

[0004] constructing a scientific research knowledge association structure, the scientific research knowledge association structure comprising entity nodes, association edges and attribute labels, the entity nodes including personnel nodes, resource nodes and knowledge nodes;

[0005] based on the scientific research knowledge association structure, performing feature extraction processing on scientific researcher data to generate a personnel feature set, the personnel feature set comprising academic background features, learning track features and demand preference features;

[0006] based on the scientific research knowledge association structure, performing feature extraction processing on training resource data to generate a resource feature set, the resource feature set comprising content coverage features, difficulty gradient features and adaptive population features;

[0007] performing cross-dimensional matching calculation on the personnel feature set and the resource feature set through the association edge weight of the scientific research knowledge association structure to generate a matching correlation degree matrix;

[0008] generating a scientific research training resource matching result according to the matching correlation degree matrix and a dynamic updating rule of the scientific research knowledge association structure.

[0009] In still another aspect, the embodiments of the present application also provide a scientific research training resource data matching system based on a knowledge graph, comprising a processor, a machine-readable storage medium, the machine-readable storage medium is connected with the processor, the machine-readable storage medium is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the machine-readable storage medium to realize the above method.

[0010] Based on the above aspects, the embodiments of the present application provide a comprehensive knowledge framework for feature extraction and matching of scientific research personnel and training resources by constructing a scientific research knowledge association structure containing entity nodes, association edges and attribute labels. Based on the scientific research knowledge association structure, feature extraction processing is performed on scientific research personnel data and training resource data to generate a personnel feature set and a resource feature set containing features such as academic background, learning track, demand preference, content coverage, difficulty gradient and adaptive population, thereby realizing accurate characterization of the characteristics of scientific research personnel and training resources. Through the association edge weight of the scientific research knowledge association structure, cross-dimension matching calculation is performed on the personnel feature set and the resource feature set to generate a matching correlation matrix, which fully considers the complex correlation between scientific research personnel and training resources, thereby improving the accuracy and effectiveness of the matching. Finally, according to the dynamic updating rule of the matching correlation matrix and the scientific research knowledge association structure, a scientific research training resource matching result is generated, which can respond to changes in the demand of scientific research personnel and the update of training resources in real time, provide personalized and accurate training resource recommendation for scientific research personnel, and effectively improve the effect and efficiency of scientific research training. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 is an execution flow diagram of the scientific research training resource data matching method based on a knowledge graph provided by the embodiments of the present application.

[0012] Figure 2 is a schematic diagram of exemplary hardware and software components of the scientific research training resource data matching system based on a knowledge graph provided by the embodiments of the present application. DETAILED DESCRIPTION

[0013] The present application will be specifically described below in conjunction with the drawings of the specification, Figure 1 is a flow diagram of the scientific research training resource data matching method based on a knowledge graph provided by an embodiment of the present application, and the scientific research training resource data matching method based on a knowledge graph will be described in detail below.

[0014] Step S110: constructing a scientific research knowledge association structure, the scientific research knowledge association structure contains entity nodes, association edges and attribute labels, the entity nodes include personnel nodes, resource nodes and knowledge nodes.

[0015] The embodiment aims to construct a scientific research knowledge association structure, which is the basic framework of the entire scientific research training resource data matching method. By clearly defining entity nodes, association edges and attribute labels, the complex relationships between personnel, resources and knowledge in the scientific research field can be presented. Specifically, the personnel node represents individuals participating in scientific research activities, the resource node corresponds to various training courses and other scientific research resources, and the knowledge node covers various knowledge points in the discipline. These entity nodes are connected to each other through association edges, and attribute labels further describe the characteristics of nodes and edges.

[0016] Step S111: Collecting multi-source data from academic paper databases, training course platforms and scientific research personnel management systems as a raw data set.

[0017] In order to construct an accurate and comprehensive scientific research knowledge association structure, data needs to be collected from multiple data sources. Academic paper databases are commonly used repositories for scientific research achievements, containing the latest research findings and academic viewpoints in various disciplines, which can reflect the cutting-edge dynamics and knowledge system of the discipline. Training course platforms record a wealth of training resources, including course content outlines, teaching methods, and target audiences, which help to understand the current status and needs of scientific research training. Scientific research personnel management systems store basic information, educational background, research experience and training records of scientific research personnel, which help to understand the academic background and development trajectory of scientific research personnel.

[0018] Therefore, data from the above three data sources is integrated to form a raw data set, which can provide rich materials for subsequent entity recognition and relationship extraction. For example, papers in the academic paper database may involve cross-disciplinary research, courses on the training course platform may be targeted at different levels of scientific research personnel, and data in the scientific research personnel management system can reflect the learning and research of scientific research personnel at different stages.

[0019] Step S112: Performing entity recognition processing on the raw data set to extract personnel entities, resource entities and knowledge entities. The personnel entity corresponds to a scientific research individual, the resource entity corresponds to a specific training course, and the knowledge entity corresponds to a knowledge point in the discipline.

[0020] After obtaining the original data set, entity recognition processing is needed to accurately identify personnel entities, resource entities, and knowledge entities from a large amount of data. For personnel entities, information in the scientific research personnel management system, such as names, positions, and affiliated institutions, is analyzed and corresponded to specific scientific research personnel. Resource entities are mainly extracted from the training course platform data, and specific training courses are determined based on course names, numbers, and content summaries. The extraction of knowledge entities is relatively complex, and in-depth analysis of the text content in the academic paper database and the training course platform is needed to identify the academic knowledge points involved, such as theoretical concepts, research methods, and technical means. For example, in an academic paper, a specific research method may be mentioned and identified as a knowledge entity; in the outline of a training course, each knowledge point listed will be extracted as a knowledge entity.

[0021] Step S113: Relationship extraction processing is performed on the original data set to establish entity association edges, including the mastery relationship between personnel entities and knowledge entities, the coverage relationship between resource entities and knowledge entities, and the historical learning relationship between personnel entities and resource entities.

[0022] After entity recognition, relationship extraction processing is performed to establish the association edges between entities. The mastery relationship between personnel entities and knowledge entities reflects the familiarity and mastery of scientific research personnel with different knowledge points. By analyzing the publication of academic papers, participation in research projects, and completed training courses of scientific research personnel, the mastery of each knowledge entity can be inferred. For example, if a scientific research personnel uses a specific research method in multiple papers, it can be considered that he has good mastery of the knowledge entity. The coverage relationship between resource entities and knowledge entities reflects the range of knowledge points covered by training courses. By analyzing the content outline of training courses, the knowledge entities involved in the courses are determined, and the coverage relationship between resource entities and knowledge entities is established. The historical learning relationship between personnel entities and resource entities records the training courses that scientific research personnel have participated in. Through the training records in the scientific research personnel management system, it can be determined which scientific research personnel have learned which training courses, and thus the historical learning relationship is established.

[0023] Step S114: Attribute labeling processing is performed on the entity nodes to set attribute labels, including the research direction of personnel entities, the teaching form of resource entities, and the knowledge level of knowledge entities.

[0024] To describe the characteristics of entity nodes in more detail, attribute labeling processing is needed. For personnel entities, research direction is an important attribute label that can reflect the main research areas and interests of researchers. By analyzing the topics of researchers' academic papers, the research projects they participate in, and their positions in research institutions, their research direction can be determined. The teaching form attribute label of resource entities can be divided into online teaching, offline teaching, and blended teaching, which helps researchers choose appropriate training courses based on their needs. The knowledge level attribute label of knowledge entities is used to distinguish the difficulty and complexity of knowledge points, such as basic knowledge points, intermediate knowledge points, and advanced knowledge points. By considering factors such as the position of knowledge entities in the discipline system, the depth of related research, and the extent of application, the knowledge level can be determined.

[0025] Step S115: Weight calibration processing is performed on the association edges, and the association edge weight value is calculated based on the co-occurrence frequency, which represents the closeness of the association between entities.

[0026] The weight value of the association edge is crucial for accurately reflecting the degree of association between entities. This step calculates the weight value of the association edge based on the co-occurrence frequency. Co-occurrence frequency refers to the number of times two entities appear simultaneously in the original data. By counting the co-occurrence frequency, the closeness of the association between entities can be quantified.

[0027] Step S1151: Count the co-occurrence times of personnel entities and knowledge entities in academic papers as the original weight value of the mastering relationship.

[0028] In academic papers, the co-occurrence of personnel entities and knowledge entities can reflect the attention and use of specific knowledge by researchers. By performing text analysis on the papers in the academic paper database, the number of times each personnel entity and each knowledge entity appears simultaneously is counted. For example, in multiple academic papers on artificial intelligence, a researcher frequently mentions the knowledge entity of deep learning algorithms, so the co-occurrence times of the personnel entity and the deep learning algorithm knowledge entity will be relatively high, and this co-occurrence time will be the original weight value of the mastering relationship between them.

[0029] Step S1152: Count the co-occurrence times of resource entities and knowledge entities in course syllabi as the original weight value of the covering relationship.

[0030] The course outline is a detailed plan of the content of a training course, which contains the knowledge entities involved in the course. By analyzing the course outlines on the training course platform, the number of times each resource entity (i.e., a specific training course) and each knowledge entity appear simultaneously is counted. For example, in a training course outline on big data analysis, the knowledge entity of data mining algorithm is mentioned multiple times, and the co-occurrence number of the resource entity and the data mining algorithm knowledge entity is taken as the original weight value of the coverage relationship between them.

[0031] Step S1153: The number of co-occurrences of personnel entities and resource entities in training records is counted as the original weight value of historical learning relationships.

[0032] The training records in the researcher management system record the participation of researchers in training courses. By counting these records, the number of times each personnel entity and each resource entity appears simultaneously is calculated. For example, a researcher has participated in multiple training courses on machine learning, and the co-occurrence number of the personnel entity and these machine learning training course resource entities is taken as the original weight value of the historical learning relationship between them.

[0033] Step S1154: The original weight values are normalized to generate standardized weight values in the range of 0-1.

[0034] Since the original weight values of different association edges may have different value ranges, in order to facilitate subsequent calculation and comparison, the original weight values need to be normalized. The purpose of normalization is to map the original weight values to the range of 0-1. The specific method can use common normalization algorithms, such as Min-Max normalization. For each association edge original weight value, subtract the minimum value in its set, and then divide by the difference between the maximum and minimum values in the set to get the standardized weight value. After this processing, all association edge weight values are between 0 and 1, and are comparable.

[0035] Step S1155: The standardized weight values are smoothed to generate final association edge weight values.

[0036] In order to avoid the influence of abnormal values in the standardized weight values on subsequent calculations, the standardized weight values need to be smoothed. Smoothing can use some smoothing algorithms, such as Gaussian smoothing or moving average smoothing. Taking moving average smoothing as an example, for each standardized weight value, take the average of its adjacent weight values as the smoothed weight value. Through smoothing, the weight values of the association edges are more stable and reasonable, and the final association edge weight values generated can more accurately reflect the association closeness between entities.

[0037] Step S120: Based on the scientific research knowledge association structure, feature extraction processing is performed on the scientific research personnel data to generate a personnel feature set, which includes academic background features, learning trajectory features, and demand preference features.

[0038] After constructing the scientific research knowledge association structure, feature extraction needs to be performed on the scientific research personnel data to fully understand the characteristics and needs of scientific research personnel. The academic background features in the personnel feature set can reflect the education background and knowledge reserve of scientific research personnel, the learning trajectory features can show the learning history and development trend of scientific research personnel, and the demand preference features can reflect the expectations and needs of scientific research personnel for training resources.

[0039] Step S121: Extract the education background information, completed training records, and training application text from the scientific research personnel data as input data.

[0040] Scientific research personnel data contains rich information, among which education background information, completed training records, and training application text are the key input data for feature extraction. Education background information includes the education level, graduate school, and major of scientific research personnel, which can reflect their basic knowledge reserve and professional direction. Completed training records record the training courses attended by scientific research personnel in the past, including course name, training time, and training effect, etc. By analyzing these records, we can understand the learning experience and knowledge expansion of scientific research personnel. Training application text is an important basis for scientific research personnel to express their training needs, which may contain information such as the knowledge field they want to learn and the training method they expect.

[0041] Step S122: Based on the mastery relationship between personnel entities and knowledge entities in the scientific research knowledge association structure, perform knowledge coverage analysis processing on the education background information, and extract the hierarchical distribution and quantity proportion of the knowledge entities mastered by scientific research personnel as academic background features.

[0042] Using the mastery relationship between personnel entities and knowledge entities in the scientific research knowledge association structure, the education background information is analyzed in depth. First, according to the education level and major of scientific research personnel, the range of knowledge entities they may contact and master is determined. Then, combined with the knowledge level attribute of knowledge entities, the distribution of knowledge entities mastered by scientific research personnel at different levels and the quantity proportion of knowledge entities at each level are counted. For example, for a scientific research personnel with a master's degree in computer science, his mastery of knowledge entities at different levels such as basic computer knowledge, intermediate algorithm design, and advanced artificial intelligence knowledge, and the quantity proportion of knowledge entities at each level constitute his academic background features.

[0043] Step S123: Based on the historical learning relationship between the personnel entity and the resource entity in the scientific research knowledge association structure, performing time series analysis processing on the completed training record, and extracting the difficulty change trend and knowledge coverage expansion path of the scientific research personnel learning resource as a learning track feature.

[0044] Performing time series analysis on the completed training record can reveal the learning development process of the scientific research personnel.

[0045] Step S1231: Arranging the completed training records of the scientific research personnel in chronological order to form a learning time series.

[0046] Arranging the completed training records of the scientific research personnel in chronological order of training time to form a learning time series. The learning time series can show the training courses attended by the scientific research personnel at different time points.

[0047] Step S1232: For each training resource in the learning time series, extracting the knowledge level span value of the corresponding difficulty gradient feature in the scientific research knowledge association structure.

[0048] For each training resource in the learning time series, according to the attribute information of the resource entity in the scientific research knowledge association structure, extract the knowledge level span value of the corresponding difficulty gradient feature. The knowledge level span value reflects the difficulty change range of the knowledge covered by the training course from basic to advanced. For example, a training course from basic programming knowledge to advanced algorithm design has a relatively large knowledge level span value.

[0049] Step S1233: Calculate the difference value of the knowledge level span values of adjacent training resources to generate a difficulty change rate sequence.

[0050] By calculating the difference value of the knowledge level span values of adjacent two training resources in the learning time series, the difficulty change rate sequence is obtained. The difficulty change rate sequence can reflect the speed and trend of the difficulty improvement of the scientific research personnel in the learning process. If the difference value of the knowledge level span values of adjacent training resources is large, it means that the scientific research personnel faces a large difficulty improvement in a short time; otherwise, it means that the difficulty improvement is relatively gentle.

[0051] Step S1234: Performing sliding window average processing on the difficulty change rate sequence to generate a difficulty change trend curve.

[0052] In order to more smoothly show the difficulty change trend, the difficulty change rate sequence is processed by sliding window averaging. A suitable window size is selected, and the difficulty change rate values in the window are averaged to obtain the average difficulty change rate corresponding to the center point of each window. These average difficulty change rates are connected to form a difficulty change trend curve. This difficulty change trend curve can intuitively show the change trend of the learning resource difficulty of the scientific researcher, which helps to judge the stability and development direction of the learning.

[0053] Step S1235: For each training resource in the learning time sequence, extract the knowledge association network of the corresponding content coverage feature in the scientific knowledge association structure.

[0054] In addition to the difficulty change trend, the knowledge coverage expansion path of the scientific researcher also needs to be analyzed. For each training resource in the learning time sequence, the knowledge association network of the corresponding content coverage feature is extracted from the scientific knowledge association structure. The knowledge association network shows the association relationship between the knowledge entities involved in the training course, including the order, causal relationship, etc.

[0055] Step S1236: Through node expansion analysis of the knowledge association network, extract the connection path of the knowledge coverage from the initial node to the new node in the learning process of the scientific researcher, and generate the knowledge coverage expansion path.

[0056] The knowledge association network of each training resource is analyzed by node expansion analysis. Starting from the knowledge entity initially mastered by the scientific researcher (i.e. the initial node), the newly added knowledge entity (i.e. the new node) in the learning process is gradually analyzed, and the connection path between them is determined. These connection paths constitute the knowledge coverage expansion path of the scientific researcher, reflecting the accumulation and expansion of the knowledge of the scientific researcher in the learning process.

[0057] Step S124: Based on the coverage relationship between the knowledge entities and resource entities of the scientific knowledge association structure, the training application text is processed by semantic analysis, and the knowledge entity to be learned and the expected resource form mentioned by the scientific researcher are extracted as demand preference features.

[0058] The training application text is processed by semantic analysis, and the coverage relationship between the knowledge entities and resource entities in the scientific knowledge association structure is used to accurately extract the knowledge entity to be learned and the expected resource form mentioned by the scientific researcher. First, the natural language processing of the training application text is performed to identify the keywords and semantic information therein. Then, the keywords are matched with the knowledge entities in the scientific knowledge association structure to determine the knowledge field that the scientific researcher wants to learn. At the same time, the description of the training method, course type, etc. in the text is analyzed to extract the expected resource form of the scientific researcher, such as online video course, offline lecture, etc.

[0059] Step S125: Combine the academic background features, learning trajectory features, and demand preference features into the personnel feature set.

[0060] The extracted academic background features, learning trajectory features, and demand preference features are combined to form the personnel feature set. The personnel feature set comprehensively describes the characteristics and needs of the scientific research personnel.

[0061] Step S130: Based on the scientific research knowledge association structure, perform feature extraction processing on the training resource data to generate a resource feature set, which includes content coverage features, difficulty gradient features, and adaptive population features.

[0062] To achieve precise matching of scientific research training resources and scientific research personnel, feature extraction of training resource data is also needed. The content coverage features in the resource feature set can reflect the knowledge range covered by the training resources, the difficulty gradient features can show the difficulty level and changes of the training resources, and the adaptive population features indicate the types of scientific research personnel suitable for the training resources.

[0063] Step S131: Extract course outline text, historical student data, and course evaluation records from the training resource data as input data.

[0064] Training resource data contains multiple aspects of information, among which course outline text, historical student data, and course evaluation records are important input data for feature extraction. Course outline text describes the content and structure of the training course in detail, which is the key basis for determining content coverage features. Historical student data records the information of scientific research personnel who have participated in the training course, including their academic background, learning effect, etc. By analyzing these data, the adaptive population features of the training resources can be understood. Course evaluation records reflect the satisfaction and feedback of students on the training course, which helps to assess the quality and applicability of the training resources.

[0065] Step S132: Based on the coverage relationship between resource entities and knowledge entities of the scientific research knowledge association structure, perform knowledge node mapping processing on the course outline text, and extract the hierarchical distribution and association network of course coverage knowledge entities as content coverage features.

[0066] The coverage relationship between the resource entity and the knowledge entity in the scientific research knowledge association structure is used to perform knowledge node mapping processing on the course outline text. The knowledge points in the course outline text are matched with the knowledge entities in the scientific research knowledge association structure to determine the range of knowledge entities covered by the course. Then, according to the knowledge level attribute of the knowledge entity, the distribution of the knowledge entities covered by the course at different levels is counted. At the same time, the association relationship between these knowledge entities is analyzed to construct a knowledge association network. For example, in a training course outline about bioinformatics, knowledge entities such as gene sequencing technology and biological data analysis algorithm are involved. Through mapping processing, the distribution of these knowledge entities at different levels and the association relationship between them can be determined. The above information constitutes the content coverage characteristics of the course.

[0067] Step S133: Based on the knowledge level attribute of the knowledge entity of the scientific research knowledge association structure, the difficulty increasing analysis processing is performed on the knowledge node order in the course outline text, and the knowledge level span from basic to advanced and the connection strength of adjacent levels of the course are extracted as the difficulty gradient characteristics.

[0068] The difficulty increasing analysis is performed on the knowledge node order in the course outline text to determine the difficulty gradient characteristics of the course.

[0069] Step S1331: The knowledge entities involved in the course outline text are identified, and their knowledge level attribute values in the scientific research knowledge association structure are obtained.

[0070] Firstly, the course outline text is analyzed to identify the knowledge entities involved. Then, the knowledge level attribute values of these knowledge entities are obtained from the scientific research knowledge association structure. For example, in a training course outline of mathematics, knowledge entities such as algebra, geometry, and calculus are involved. By querying the scientific research knowledge association structure, the knowledge level attribute values of these knowledge entities can be obtained.

[0071] Step S1332: The knowledge entities are arranged in the order of chapters of the course outline to form a knowledge node sequence.

[0072] The identified knowledge entities are arranged in the order of chapters of the course outline to form a knowledge node sequence. The knowledge node sequence can reflect the organization and logical order of the course content.

[0073] Step S1333: The difference between the knowledge level attribute values of the first node and the last node in the knowledge node sequence is calculated as the knowledge level span.

[0074] The knowledge level span of the course is obtained by calculating the difference between the knowledge level attribute values of the first node and the last node in the knowledge node sequence. The knowledge level span reflects the degree of difficulty increase of the course from the basic to the advanced. For example, for a course from basic programming knowledge to advanced algorithm design, the knowledge level span is relatively large.

[0075] Step S1334: Calculate the difference between the knowledge level attribute values of adjacent nodes in the knowledge node sequence to generate a level increase difference sequence.

[0076] The difference between the knowledge level attribute values of adjacent nodes in the knowledge node sequence is calculated, which can understand the increase of the course content in difficulty. Specifically, starting from the first node of the knowledge node sequence, the difference between the knowledge level attribute values of adjacent two nodes is calculated. For example, if the knowledge node sequence is node A, node B, node C, etc., the difference between the knowledge level attribute value of node B and the knowledge level attribute value of node A is calculated, and a difference value is obtained. Then, the difference between the knowledge level attribute value of node C and the knowledge level attribute value of node B is calculated, and another difference value is obtained. In this way, the level increase difference sequence is generated by arranging these difference values in order. Each difference value in the level increase difference sequence represents the difficulty increase amplitude of the course between adjacent knowledge nodes.

[0077] Step S1335: Calculate the proportion of the difference values less than or equal to the preset threshold in the level increase difference sequence as the connection strength of adjacent levels.

[0078] The preset threshold is a reference standard designed according to the course difficulty, which is used to measure whether the connection between adjacent knowledge levels is smooth. Each difference value in the level increase difference sequence is compared with the preset threshold, and the number of difference values less than or equal to the preset threshold is counted. Then, the proportion of the number of difference values less than or equal to the preset threshold to the total number of the level increase difference sequence is calculated, which is the connection strength of adjacent levels. If the proportion is high, it means that the difficulty increase between adjacent knowledge levels of the course is relatively smooth and the connection is close. On the contrary, if the proportion is low, it means that the difficulty jump between some adjacent knowledge levels of the course is large and the connection is not smooth.

[0079] Step S1336: Combine the knowledge level span and the connection strength into the difficulty gradient feature.

[0080] The calculated knowledge level span and the connection strength of adjacent levels are combined to form the difficulty gradient feature of the course. The knowledge level span reflects the overall difficulty increase of the course from the basic to the advanced, and the connection strength reflects the close degree of the connection between adjacent knowledge levels in the course. The two aspects together constitute the difficulty gradient feature, which can comprehensively describe the difficulty of the course and provide an important reference for researchers to choose a suitable training course.

[0081] Step S134: Based on the historical learning relationship between the personnel entity and the resource entity in the scientific research knowledge association structure, the statistical analysis processing is performed on the academic background information in the historical student data, and the knowledge coverage level distribution and the learning track feature of the course historical student are extracted as the adaptive crowd feature.

[0082] The historical student data is deeply mined by using the historical learning relationship between the personnel entity and the resource entity in the scientific research knowledge association structure. First, according to the education background, the completed training courses and other information of the historical student, the range of knowledge entities mastered by the historical student is determined. Then, combined with the knowledge level attribute of the knowledge entity, the knowledge coverage of the historical student at different knowledge levels is counted, and the knowledge coverage level distribution is obtained. At the same time, the learning track of the historical student is analyzed, including the order of their participation in the training courses, the difficulty change and the like, and the learning track feature is extracted. These knowledge coverage level distribution and learning track feature jointly constitute the adaptive crowd feature of the course, which can help to determine which type of scientific researchers the course is suitable for.

[0083] Step S135: The content coverage feature, the difficulty gradient feature and the adaptive crowd feature are combined as the resource feature set.

[0084] The extracted content coverage feature, difficulty gradient feature and adaptive crowd feature are combined to form a resource feature set. The resource feature set comprehensively describes the characteristics and application scope of the training resource.

[0085] Step S140: The personnel feature set and the resource feature set are calculated by cross-dimension matching through the association edge weight of the scientific research knowledge association structure, and a matching association degree matrix is generated.

[0086] After obtaining the personnel feature set and the resource feature set, the cross-dimension matching calculation is performed on the two sets through the association edge weight of the scientific research knowledge association structure, so as to determine the matching association degree between each scientific researcher and each training resource.

[0087] Step S141: The knowledge coverage level distribution of the academic background feature in the personnel feature set is extracted, and the intersection calculation is performed with the knowledge coverage level distribution of the content coverage feature in the resource feature set, and a knowledge overlap degree index is generated.

[0088] The knowledge coverage level distribution of the academic background features extracted from the personnel feature set shows the knowledge mastery of the researchers at different knowledge levels. Meanwhile, the knowledge coverage level distribution of the content coverage features extracted from the resource feature set reflects the content coverage of the training resources at different knowledge levels. The intersection of the two knowledge coverage level distributions is calculated, i.e., the range of knowledge entities that are commonly covered at the same knowledge level. By counting the number of commonly covered knowledge entities and weighting them with the associated edge weights, the knowledge overlap index is generated. The knowledge overlap index reflects the matching degree of the academic background of the researchers and the content of the training resources.

[0089] Step S142: Extract the difficulty change trend of the learning trajectory features in the personnel feature set, and perform trend consistency analysis on the knowledge level span of the difficulty gradient features in the resource feature set to generate the difficulty connection index.

[0090] The difficulty change trend of the learning trajectory features in the personnel feature set describes the difficulty improvement of the researchers in the learning process, while the knowledge level span of the difficulty gradient features in the resource feature set reflects the difficulty improvement degree of the training resources from basic to advanced. Trend consistency analysis is performed on the two, and whether the difficulty change trend of the researchers is consistent with the knowledge level span of the training resources is compared. For example, if the learning trajectory of the researchers shows that the difficulty improvement is relatively flat, and the knowledge level span of the training resources is large, their trend consistency may be low; on the contrary, if the trend is similar, the trend consistency is high. Through a series of analysis and weighting calculation, the difficulty connection index is generated, which reflects the matching degree of the learning difficulty change of the researchers and the difficulty of the training resources.

[0091] Step S143: Extract the to-be-learned knowledge entities of the demand preference features in the personnel feature set, and perform path matching processing on the knowledge association network of the content coverage features in the resource feature set to generate the demand satisfaction index.

[0092] The to-be-learned knowledge entities of the demand preference features are extracted from the personnel feature set, which represent the knowledge fields that the researchers want to learn. At the same time, the knowledge association network of the content coverage features is extracted from the resource feature set, which shows the association relationship between the knowledge entities covered by the training resources. Path matching processing is performed on the two to find out whether the knowledge association network of the training resources contains the to-be-learned knowledge entities of the researchers, and whether the connection path of these knowledge entities in the network is reasonable. By counting the number of matched knowledge entities and the path rationality, and weighting them with the associated edge weights, the demand satisfaction index is generated. The demand satisfaction index reflects whether the training resources can meet the learning needs of the researchers.

[0093] Step S144: Calculate the distribution similarity between the proportion of the number of knowledge coverage of the academic background feature in the personnel feature set and the hierarchical distribution of the knowledge coverage of the adapted population feature in the resource feature set to generate a population adaptation degree index.

[0094] The proportion of the number of knowledge coverage of the academic background feature in the personnel feature set reflects the proportion of knowledge mastery of scientific researchers at different knowledge levels, while the hierarchical distribution of the knowledge coverage of the adapted population feature in the resource feature set shows the knowledge coverage of scientific researchers suitable for the training resource at different knowledge levels. The distribution similarity between the two is calculated to compare whether their distribution patterns at different knowledge levels are similar. Some similarity calculation methods such as cosine similarity can be used, and the calculation results can be weighted by combining the associated edge weights to generate a population adaptation degree index. The population adaptation degree index reflects the matching degree of the academic background of scientific researchers and the adapted population of training resources.

[0095] Step S145: Based on the weight values of the associated edges in the scientific knowledge association structure, assign weight coefficients to the knowledge overlap degree index, the difficulty connection degree index, the demand satisfaction degree index, and the population adaptation degree index.

[0096] According to the weight values of the associated edges in the scientific knowledge association structure, assign corresponding weight coefficients to the knowledge overlap degree index, the difficulty connection degree index, the demand satisfaction degree index, and the population adaptation degree index. The weight values of the associated edges reflect the closeness of the association between entities, so when assigning weight coefficients, the importance of these associations to the matching result will be considered. For example, if the weight of the associated edge of the mastery relationship between the personnel entity and the knowledge entity is high, the weight coefficient of the knowledge overlap degree index may be relatively large.

[0097] Step S146: Multiply the knowledge overlap degree index, the difficulty connection degree index, the demand satisfaction degree index, and the population adaptation degree index by their corresponding weight coefficients and then sum them up to generate a matching association degree value between a single resource and a single personnel.

[0098] Multiply the knowledge overlap degree index, the difficulty connection degree index, the demand satisfaction degree index, and the population adaptation degree index by their corresponding weight coefficients, and then add these products to obtain a matching association degree value between a single resource and a single personnel. This matching association degree value considers multiple matching factors and can accurately reflect the matching degree between a certain scientific researcher and a certain training resource.

[0099] Step S147: Calculate the matching association degree value for all personnel and all resources to generate a matching association degree matrix containing the matching values of all personnel-resource pairs.

[0100] The matching correlation value calculation is performed on each scientific researcher and each training resource, and all the calculation results are arranged into a matching correlation matrix. The rows of the matching correlation matrix represent the scientific researchers, the columns represent the training resources, and each element in the matching correlation matrix is the matching correlation value of the corresponding scientific researcher and the training resource. The matching correlation matrix can comprehensively show the matching between all the scientific researchers and the training resources.

[0101] Step S150: generating a scientific training resource matching result according to the matching correlation matrix and a dynamic updating rule of the scientific knowledge correlation structure.

[0102] After obtaining the matching correlation matrix, the final scientific training resource matching result is generated in combination with the dynamic updating rule of the scientific knowledge correlation structure.

[0103] Step S151: performing descending order sorting processing on the matching correlation values in the matching correlation matrix to generate a resource sorting list corresponding to each scientific researcher.

[0104] The matching correlation values of each row (i.e., all the training resources corresponding to each scientific researcher) in the matching correlation matrix are sorted in descending order, the training resources are arranged in descending order of the matching correlation with the scientific researcher, and a resource sorting list corresponding to each scientific researcher is generated. The resource sorting list can intuitively show the matching order of each scientific researcher and different training resources.

[0105] Step S152: screening resources with a matching correlation value greater than a preset threshold from the resource sorting list to form a candidate resource set.

[0106] A preset threshold is set, the training resources with a matching correlation value greater than the threshold are screened from the resource sorting list of each scientific researcher, and these resources are combined to form a candidate resource set. The threshold is used to filter out resources with a low matching degree with the scientific researchers, and only resources with a high matching possibility are retained.

[0107] Step S153: based on the historical learning relationship between the personnel entity and the resource entity of the scientific knowledge correlation structure, the number of repeated recommendations of each resource in the candidate resource set is counted, and the number of repeated recommendations represents the frequency of selection of the resource by personnel with the same type of academic background.

[0108] The number of repeated recommendations of each resource in the candidate resource set is counted using the historical learning relationship between the personnel entity and the resource entity of the scientific knowledge correlation structure. Specifically, the selection of these resources by scientific researchers with similar academic backgrounds is analyzed, and the number of times each resource is selected is counted. The number of repeated recommendations reflects the popularity and applicability of the resource among personnel with the same type of academic background.

[0109] Step S154: Adjusting the matching correlation degree value of the resources in the candidate resource set according to the latest research dynamics of the knowledge entities in the scientific research knowledge association structure, wherein the adjustment process is based on the research popularity of the knowledge entities and the weight values of the resources covering the knowledge entities.

[0110] Considering the dynamic changes in the field of scientific research, it is necessary to adjust the matching correlation degree values of the resources in the candidate resource set according to the latest research dynamics of the knowledge entities in the scientific research knowledge association structure.

[0111] For example, step S1541: Extracting paper data published within a preset time span from an academic paper database, and counting the occurrence frequency of each knowledge entity in the paper as the research popularity value of the knowledge entity.

[0112] Selecting paper data within a preset time span from an academic paper database, and performing text analysis on these papers to count the occurrence frequency of each knowledge entity in the paper. This occurrence frequency serves as the research popularity value of the knowledge entity, reflecting the degree of attention to the knowledge entity in the current field of scientific research.

[0113] Step S1542: Normalizing the research popularity values to generate standardized popularity values within the range of 0-1.

[0114] In order to facilitate subsequent calculation and comparison, the research popularity values are normalized. A suitable normalization method is used to map the research popularity values to the range of 0-1 to obtain standardized popularity values. This ensures that the research popularity values of different knowledge entities are comparable.

[0115] Step S1543: Extracting the knowledge entities covered by each resource in the candidate resource set in the scientific research knowledge association structure and their corresponding coverage relationship weight values.

[0116] From the scientific research knowledge association structure, extract the knowledge entities covered by each resource in the candidate resource set, as well as the coverage relationship weight values between these resources and knowledge entities. These weight values reflect the coverage degree and importance of resources to knowledge entities.

[0117] Step S1544: Calculating the heat adjustment coefficient of each resource, which is the sum of the product of the standardized popularity values of the covered knowledge entities and the coverage relationship weight values.

[0118] For each resource, multiply the standardized popularity values of the covered knowledge entities by the corresponding coverage relationship weight values, and then add these products to obtain the heat adjustment coefficient of the resource. This heat adjustment coefficient takes into account the research popularity and coverage degree of the knowledge entities covered by the resource.

[0119] Step S1545: Multiply the original matching correlation value of each resource by the heat adjustment coefficient to generate an adjusted matching correlation value.

[0120] The original matching correlation value of each resource in the candidate resource set is multiplied by the corresponding heat adjustment coefficient to obtain the adjusted matching correlation value. In this way, the matching correlation value can better reflect the applicability of the resource in the current scientific research environment.

[0121] Step S155: Use the adjusted matching correlation value as the final sorting basis to generate a personalized resource matching list for each researcher.

[0122] Based on the adjusted matching correlation value, the candidate resource set for each researcher is reordered to generate a personalized resource matching list for each researcher. This personalized resource matching list takes into account the researcher's own characteristics, resource matching, and the latest research dynamics of knowledge entities, and can provide more accurate training resource recommendations for researchers.

[0123] Step S156: Combine all personalized resource matching lists into a scientific training resource matching result.

[0124] Combine the personalized resource matching list of each researcher to form the final scientific training resource matching result. This scientific training resource matching result can provide training resource recommendations that meet the needs and interests of researchers, promoting the learning and development of researchers.

[0125] In the entire data processing process, data collection, storage, and use are involved, which may involve privacy-sensitive data such as researchers' personal information, academic background, etc. In order to protect these privacy-sensitive data, a series of privacy protection and anti-leakage technical means are adopted. In the data collection stage, the principles of legality, legitimacy, and necessity are followed, only the necessary data related to scientific training resource matching are collected, and the explicit authorization of the data owner is obtained. In terms of data storage, encryption technology is used to encrypt the data storage to ensure the security of the data in the storage process. In the data use process, the data is anonymized to remove information that can directly identify the individual's identity, and access control technology is used to limit only authorized personnel to access and process the data. In addition, regular backups and monitoring of data are carried out to timely discover and handle possible security vulnerabilities, ensuring the security of privacy-sensitive data.

[0126] Figure 2An exemplary hardware and software components of the knowledge graph-based scientific research training resource data matching system 100 provided by some embodiments of the present application, which can implement the idea of the present application, are shown in the schematic diagram. For example, the processor 120 can be used in the knowledge graph-based scientific research training resource data matching system 100 and used to perform the functions in the present application.

[0127] The knowledge graph-based scientific research training resource data matching system 100 can be a general server or a special-purpose server, both of which can be used to implement the knowledge graph-based scientific research training resource data matching method of the present application. Although only one server is shown in the present application, for the sake of convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0128] For example, the knowledge graph-based scientific research training resource data matching system 100 can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, a ROM, or a RAM, or any combination thereof. Exemplarily, the knowledge graph-based scientific research training resource data matching system 100 can also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The knowledge graph-based scientific research training resource data matching system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0129] For the sake of illustration, only one processor is described in the knowledge graph-based scientific research training resource data matching system 100. However, it should be noted that the knowledge graph-based scientific research training resource data matching system 100 in the present application can also include multiple processors, so the steps performed by one processor described in the present application can also be jointly performed or individually performed by multiple processors. For example, if the processor of the knowledge graph-based scientific research training resource data matching system 100 performs steps A and B, it should be understood that steps A and B can also be jointly performed by two different processors or individually performed in one processor. For example, a first processor performs step A, a second processor performs step B, or a first processor and a second processor jointly perform steps A and B.

[0130] In addition, the present application also provides a readable storage medium, in which computer executable instructions are pre-set, when the processor executes the computer executable instructions, the knowledge graph-based scientific research training resource data matching method as above is implemented.

[0131] It should be noted that the foregoing description of embodiments of the application has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the application to the precise form described, and many modifications, variations, and alternatives are possible.

Claims

1. A knowledge graph-based scientific research training resource data matching method, characterized in that, The method comprises: constructing a scientific research knowledge association structure, the scientific research knowledge association structure comprising entity nodes, association edges and attribute labels, the entity nodes including personnel nodes, resource nodes and knowledge nodes; based on the scientific research knowledge association structure, performing feature extraction processing on scientific research personnel data to generate a personnel feature set, the personnel feature set comprising academic background features, learning track features and demand preference features; based on the scientific research knowledge association structure, performing feature extraction processing on training resource data to generate a resource feature set, the resource feature set comprising content coverage features, difficulty gradient features and adaptive population features; performing cross-dimension matching calculation on the personnel feature set and the resource feature set through the association edge weight of the scientific research knowledge association structure to generate a matching correlation degree matrix; generating a scientific research training resource matching result according to the matching correlation degree matrix and a dynamic updating rule of the scientific research knowledge association structure. 2.The knowledge graph-based scientific research training resource data matching method according to claim 1, characterized in that, The scientific research knowledge association structure comprises entity nodes, association edges and attribute labels, the entity nodes including personnel nodes, resource nodes and knowledge nodes, and comprises: collecting multi-source data of an academic paper database, a training course platform and a scientific research personnel management system as an original data set; performing entity recognition processing on the original data set to extract personnel entities, resource entities and knowledge entities, the personnel entities corresponding to scientific research personnel individuals, the resource entities corresponding to specific training courses, and the knowledge entities corresponding to discipline knowledge points; performing relationship extraction processing on the original data set to establish association edges between entities, the association edges including a mastery relationship between a personnel entity and a knowledge entity, a coverage relationship between a resource entity and a knowledge entity, and a historical learning relationship between a personnel entity and a resource entity; performing attribute labeling processing on the entity nodes to set attribute labels, the attribute labels including a research direction of a personnel entity, a teaching form of a resource entity and a knowledge level of a knowledge entity; performing weight calibration processing on the association edges to calculate association edge weight values based on co-occurrence frequency, the association edge weight values representing the closeness of the association between entities; organizing the entity nodes, association edges, attribute labels and weight values into the scientific research knowledge association structure. 3.The knowledge graph based scientific research training resource data matching method according to claim 2, characterized in that, The weight calibration processing on the association edges comprises: counting the co-occurrence times of a personnel entity and a knowledge entity in an academic paper as the original weight value of the mastery relationship; counting the co-occurrence times of a resource entity and a knowledge entity in a course outline as the original weight value of the coverage relationship; counting the co-occurrence times of a personnel entity and a resource entity in a training record as the original weight value of the historical learning relationship; performing normalization processing on the original weight values to generate standardized weight values in the range of 0-1; performing smoothing processing on the standardized weight values to generate final association edge weight values. 4.The knowledge graph based scientific research training resource data matching method according to claim 1, characterized in that, The feature extraction processing is performed on the scientific research personnel data based on the scientific research knowledge association structure to generate a personnel feature set, and the personnel feature set includes an academic background feature, a learning track feature, and a demand preference feature, including: The education background information, the completed training record, and the training application text are extracted from the scientific research personnel data as input data; The knowledge coverage analysis processing is performed on the education background information based on the mastering relationship between the personnel entity and the knowledge entity of the scientific research knowledge association structure, and the hierarchical distribution and the quantity proportion of the mastered knowledge entities of the scientific research personnel are extracted as the academic background feature; The time series analysis processing is performed on the completed training record based on the historical learning relationship between the personnel entity and the resource entity of the scientific research knowledge association structure, and the difficulty change trend and the knowledge coverage expansion path of the learning resource of the scientific research personnel are extracted as the learning track feature; The semantic analysis processing is performed on the training application text based on the coverage relationship between the knowledge entity and the resource entity of the scientific research knowledge association structure, and the mentioned to-be-learned knowledge entity and the expected resource form of the scientific research personnel are extracted as the demand preference feature; The academic background feature, the learning track feature, and the demand preference feature are combined as the personnel feature set. 5.The knowledge graph based scientific research training resource data matching method according to claim 4, characterized in that, The time series analysis processing is performed on the completed training record based on the historical learning relationship between the personnel entity and the resource entity of the scientific research knowledge association structure, and the difficulty change trend and the knowledge coverage expansion path of the learning resource of the scientific research personnel are extracted as the learning track feature, including: The completed training records of the scientific research personnel are arranged in chronological order to form a learning time sequence; The knowledge level span value of the difficulty gradient feature corresponding to each training resource in the learning time sequence in the scientific research knowledge association structure is extracted; The difference value of the knowledge level span values of adjacent training resources is calculated to generate a difficulty change rate sequence; The sliding window average processing is performed on the difficulty change rate sequence to generate a difficulty change trend curve; The knowledge association network of the content coverage feature corresponding to each training resource in the learning time sequence in the scientific research knowledge association structure is extracted; The connection path of the knowledge coverage from the initial node to the new node in the learning process of the scientific research personnel is extracted through the node expansion analysis of the knowledge association network to generate a knowledge coverage expansion path; The difficulty change trend curve and the knowledge coverage expansion path are combined as the learning track feature. 6.The knowledge graph based scientific research training resource data matching method according to claim 1, characterized in that, The feature extraction processing is performed on the training resource data based on the scientific research knowledge association structure to generate a resource feature set, and the set includes a content coverage feature, a difficulty gradient feature, and an adaptive crowd feature, including: The course outline text, the historical student data, and the course evaluation record are extracted from the training resource data as input data; The knowledge node mapping processing is performed on the course outline text based on the coverage relationship between the resource entity and the knowledge entity of the scientific research knowledge association structure, and the hierarchical distribution and the association network of the course coverage knowledge entity are extracted as the content coverage feature; Based on the knowledge level attribute of the knowledge entity in the scientific research knowledge association structure, the knowledge node sequence in the course outline text is analyzed and processed in difficulty increasing order, and the knowledge level span from basic to advanced and the connection strength of adjacent levels are extracted as difficulty gradient features; Based on the historical learning relationship between personnel entities and resource entities in the scientific research knowledge association structure, the academic background information in the historical student data is statistically analyzed and processed, and the knowledge coverage level distribution and learning track features of the historical students of the course are extracted as adaptation crowd features; The content coverage features, difficulty gradient features and adaptation crowd features are combined as the resource feature set.

7. The knowledge graph-based scientific research training resource data matching method according to claim 6, characterized in that, Based on the knowledge level attribute of the knowledge entity in the scientific research knowledge association structure, the knowledge node sequence in the course outline text is analyzed and processed in difficulty increasing order, and the knowledge level span from basic to advanced and the connection strength of adjacent levels are extracted as difficulty gradient features, including: Identify the knowledge entities involved in the course outline text, and obtain their knowledge level attribute values in the scientific research knowledge association structure; Arrange the knowledge entities in the order of chapters of the course outline to form a knowledge node sequence; Calculate the difference between the knowledge level attribute values of the first node and the last node in the knowledge node sequence as the knowledge level span; Calculate the difference between the knowledge level attribute values of adjacent nodes in the knowledge node sequence to generate a level increasing difference sequence; Statistically analyze the proportion of the level increasing difference sequence that is less than or equal to a preset threshold as the connection strength of adjacent levels; Combine the knowledge level span and the connection strength as the difficulty gradient features. 8.The knowledge graph based scientific research training resource data matching method according to claim 1, characterized in that, The association edge weight of the scientific research knowledge association structure is used to calculate the cross-dimension matching of the personnel feature set and the resource feature set to generate a matching association degree matrix, including: Extract the knowledge coverage level distribution of the academic background features in the personnel feature set, and calculate the intersection of the knowledge coverage level distribution of the content coverage features in the resource feature set to generate a knowledge overlap degree index; Extract the difficulty change trend of the learning track features in the personnel feature set, and perform trend consistency analysis on the knowledge level span of the difficulty gradient features in the resource feature set to generate a difficulty connection degree index; Extract the to-be-learned knowledge entity of the demand preference features in the personnel feature set, and perform path matching processing on the knowledge association network of the content coverage features in the resource feature set to generate a demand satisfaction degree index; Calculate the distribution similarity between the knowledge coverage quantity proportion of the academic background features in the personnel feature set and the knowledge coverage level distribution of the adaptation crowd features in the resource feature set to generate a crowd adaptation degree index; Based on the weight values of the association edges in the scientific research knowledge association structure, the knowledge overlap degree index, the difficulty connection degree index, the demand satisfaction degree index and the crowd adaptation degree index are assigned weight coefficients; The knowledge overlap index, the difficulty connection index, the demand satisfaction index, and the population adaptation index are multiplied by their corresponding weight coefficients and summed to generate a matching association value for a single resource and a single person; The matching association value is calculated for all personnel and all resources to generate a matching association matrix containing matching values for all personnel-resource pairs. 9.The knowledge graph based scientific research training resource data matching method according to claim 1, characterized in that, The matching association matrix and the dynamic updating rule of the scientific research knowledge association structure are used to generate a scientific research training resource matching result, including: The matching association values in the matching association matrix are sorted in descending order to generate a resource sorting list corresponding to each scientific researcher; Resources with matching association values greater than a preset threshold are selected from the resource sorting list to form a candidate resource set; Based on the historical learning relationship between the personnel entity and the resource entity of the scientific research knowledge association structure, the number of repeated recommendations of each resource in the candidate resource set is counted, and the number of repeated recommendations represents the frequency of selection of the resource by personnel with the same academic background; For the resources in the candidate resource set, the matching association values are adjusted based on the latest research dynamics of the knowledge entity of the scientific research knowledge association structure, and the adjustment process is based on the research popularity of the knowledge entity and the weight value of the resource covering the knowledge entity; The adjusted matching association values are used as the final sorting basis to generate a personalized resource matching list for each scientific researcher; All personalized resource matching lists are combined to generate a scientific research training resource matching result.

10. A knowledge graph-based scientific research training resource data matching system, characterized in that, A processor and a memory are included, the memory and the processor are connected, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the scientific research training resource data matching method based on a knowledge graph according to any one of claims 1-9.

Images