A standard information management method and system based on a knowledge graph

Abstract

The present invention discloses a standard information management method and system based on a knowledge graph, relating to the technical field of standard information management. The method includes the following steps: obtaining a schema layer of the knowledge graph to be constructed; generating a knowledge graph; obtaining standard information data; and iterative optimization of the knowledge graph. The present invention collects information data to be managed within a preset time period and performs preprocessing to obtain the schema layer of the knowledge graph to be constructed, then conceptually abstracts the schema layer of the knowledge graph to be constructed and generates the corresponding knowledge graph in combination with the corresponding ontology structure and knowledge relevance index, and finally updates the knowledge relationship information in the knowledge graph to obtain standard information data. At the same time, the constructed knowledge graph is iteratively optimized in combination with the feedback results of users and preset standard information indicators, achieving the effect of improving the accuracy of obtaining standard information data and solving the problem of low accuracy of obtaining standard information data in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of standard information management, and particularly to a method and system for standard information management based on a knowledge graph. Background Art

[0002] With the rapid development of the Internet and information technology, the amount of data has shown an explosive growth. This data includes structured and unstructured data, covering all aspects of human society, such as entities like people, organizations, locations, and the relationships between them, which also provides a rich data foundation for the construction of a knowledge graph. Secondly, as a technology based on the semantic web, the knowledge graph technology can integrate a large amount of structured and unstructured information to construct a graph, thereby helping us better understand the relevance and correlation of information. However, traditional information management methods have been difficult to meet the effective management and utilization of massive information. Therefore, in order to promote knowledge sharing and collaborative innovation, a method and system for standard information management based on a knowledge graph have emerged, realizing comprehensive, efficient, and intelligent management of standard information, and providing more high-quality and personalized services for enterprises and organizations.

[0003] Existing standard information management methods extract entity information and relationship information from text data and image data automatically, and at the same time integrate the entity information and relationship information to construct a knowledge graph, and then associate entities and relationships by setting primary keys and foreign keys in the knowledge graph to realize the storage and management of standard information.

[0004] For example, a method and system for intelligent operation and maintenance management based on a knowledge graph disclosed in the invention patent announcement with the announcement number: CN116611813B includes: obtaining first information and performing data conversion processing to obtain a knowledge graph; performing fusion based on the entity relationships in the knowledge graph and combining the operation and maintenance information in the text data for encoding processing to obtain an encoded representation of the knowledge graph; constructing a prediction model according to a preset operation and maintenance strategy and text data, and performing decoding processing on the encoded representation according to the prediction model to obtain an operation and maintenance strategy sequence; automatically executing operation and maintenance operations according to the operation and maintenance strategy sequence and preset rules in the operation and maintenance tasks.

[0005] For example, a landslide prediction method and system for a target research area based on a knowledge graph disclosed in the invention patent announcement with the announcement number: CN116611546B includes: importing target data into the knowledge graph and, in the case where the target research area is a sample-scarce area, a candidate prediction model based on candidate areas; performing landslide prediction on the target research area according to the candidate prediction model.

[0006] However, in the process of implementing the technical solutions of the present invention in the embodiments of the present application, it is found that the above technologies have at least the following technical problems:

[0007] In the prior art, due to the continuous update of the data volume during the extraction process of entity information and relationship information, the complexity of text data and image data increases, resulting in a problem of low accuracy in obtaining standard information data. Summary of the Invention

[0008] By providing a method and system for standard information management based on a knowledge graph, embodiments of the present application solve the problem of low accuracy in obtaining standard information data in the prior art and achieve an improvement in the accuracy of obtaining standard information data.

[0009] Embodiments of the present application provide a method for standard information management based on a knowledge graph, including the following steps: S1, collecting information data to be managed within a preset time period and performing preprocessing, and at the same time performing knowledge extraction and knowledge processing on the preprocessed information data to obtain a schema layer of the knowledge graph to be constructed; S2, performing concept abstraction on the schema layer of the knowledge graph to be constructed and generating a corresponding knowledge graph in combination with the corresponding ontology structure and knowledge correlation index, where the knowledge correlation index is used to measure the degree of association between the schema layer of the knowledge graph to be constructed and the knowledge graph within a preset time period; S3, performing statistical analysis on the knowledge graph and obtaining data similarity according to the results of the statistical analysis, and at the same time updating the knowledge relationship information in the knowledge graph according to the data similarity to obtain standard information data, where the data similarity is used to measure the similarity degree between information data in the knowledge graph; S4, performing interactive management on the standard information data and iteratively optimizing the constructed knowledge graph in combination with the feedback results of the user and preset standard information indicators.

[0010] Further, the specific steps for obtaining the node text matching degree are as follows: S21, performing word segmentation processing on the information data within a preset time period and the node text in the schema layer structure, and at the same time obtaining a string matching degree according to the results of the word segmentation processing, where the string matching degree is used to measure the matching degree between the information data and the phrase string in the schema layer structure; S22, identifying entity semantic information in the information data and the node text of the schema layer structure, and obtaining semantic similarity based on the cosine similarity of the vector space; S23, extracting attribute values corresponding to the target node text attribute from the information data and the schema layer structure and performing attribute verification, and at the same time determining whether the result of the attribute verification meets the preset requirements of the target node text. If it meets, stop the attribute verification and input the attribute value into the schema layer, otherwise execute S24; S24, re-performing attribute verification on the attribute value that does not meet the preset requirements of the target node text to obtain an attribute verification value and performing data conversion according to the requirements of the schema layer structure, and at the same time obtaining a preset attribute verification correction factor from the database according to the preset attribute verification standard, and combining the string matching degree and the semantic similarity to obtain the node text matching degree.

[0011] Further, the node text matching degree is calculated by the following formula:

[0012]

[0013] In the formula, g is the number of the preset time period, g = 1, 2,..., G, G is the total number of the preset time periods, YI g represents the node text matching degree of the information data in the g-th preset time period, e is the natural constant, A g represents the string matching degree of the information data in the g-th preset time period, A 0 represents the string reference matching degree, B g represents the semantic similarity degree of the information data in the g-th preset time period, B 0 represents the semantic reference similarity degree, C g represents the attribute verification value of the information data in the g-th preset time period, C 0 represents the attribute reference verification value, and α represents the attribute verification correction factor.

[0014] Further, the specific steps for obtaining the data similarity include: converting the entity information and relationship information in the knowledge graph into vectors and combining the corresponding spatial angles to obtain the entity vector value and the relationship vector value, and simultaneously monitoring the spatial position changes of the entity information and relationship information in the knowledge graph during the vector conversion process and obtaining the preset cosine similarity correction factor from the database; monitoring the changes of the information data during the construction of the knowledge graph in real time, and simultaneously obtaining the structure coefficient according to the result of the position mapping and combining the preset cosine similarity correction factor to obtain the data similarity, and the data similarity is calculated by the following formula:

[0015]

[0016] In the formula, g is the number of the preset time period, g = 1, 2,..., G, G is the total number of the preset time periods, TN g represents the data similarity of the knowledge graph in the g-th preset time period, δ represents the cosine similarity correction factor, δ ≥ 1, e is the natural constant, H g represents the entity vector of the entity information in the g-th preset time period, K g represents the relationship vector of the relationship information in the g-th preset time period, W g represents the structure coefficient of the structured data in the knowledge graph in the g-th preset time period, W 0 represents the structure reference coefficient.

[0017] Further, the specific process for obtaining the standard information data is as follows: extract structured data from the knowledge graph according to data similarity and in combination with relational database query language, and perform frequency statistics; encode and classify the structured data after frequency analysis according to preset standard information indicators, and visually display the frequency distribution of the structured data in the form of a chart; generate standard information data based on the results of encoding and classification within a preset time period and conduct verification, and at the same time, update the entity information and relationship information in the knowledge graph in real time according to the verification results.

[0018] The embodiment of the present application provides a standard information management system based on a knowledge graph, including: a schema layer acquisition module for the knowledge graph to be constructed, a knowledge graph generation module, a standard information data acquisition module, and a knowledge graph iterative optimization module; wherein, the schema layer acquisition module for the knowledge graph to be constructed is used to collect information data to be managed within a preset time period and perform preprocessing, and at the same time, perform knowledge extraction and knowledge processing on the preprocessed information data to obtain the schema layer of the knowledge graph to be constructed; the knowledge graph generation module is used to perform concept abstraction on the schema layer of the knowledge graph to be constructed and generate the corresponding knowledge graph in combination with the corresponding ontology structure and knowledge correlation index, and the knowledge correlation index is used to measure the degree of association between the schema layer of the knowledge graph to be constructed and the knowledge graph within a preset time period; the standard information data acquisition module is used to perform statistical analysis on the knowledge graph and obtain data similarity according to the results of the statistical analysis, and at the same time, update the knowledge relationship information in the knowledge graph according to the data similarity to obtain standard information data, and the data similarity is used to measure the similarity degree between information data in the knowledge graph; the knowledge graph iterative optimization module is used to perform interactive management on the standard information data and perform iterative optimization on the constructed knowledge graph in combination with the feedback results of the user and preset standard information indicators.

[0019] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0020] 1. By collecting information data to be managed within a preset time period and performing preprocessing to obtain the schema layer of the knowledge graph to be constructed, then generating the corresponding knowledge graph in combination with the corresponding ontology structure and knowledge correlation index, and finally updating the knowledge relationship information in the knowledge graph to obtain standard information data and performing iterative optimization on the constructed knowledge graph, the accuracy of the schema layer of the knowledge graph to be constructed is improved, and further the accuracy of obtaining standard information data is improved, effectively solving the problem of low accuracy of obtaining standard information data in the prior art.

[0021] 2. By performing position mapping on the preprocessed information data according to the ontology structure of the schema layer of the knowledge graph to be constructed, generating the structural framework of the knowledge graph to be constructed based on the result of the position mapping, then obtaining the knowledge correlation index according to the co-occurrence frequency between entity information in the ontology structure, and simultaneously validating the generated structural framework and improving the attribute relationship between information data to generate the knowledge graph, the more accurate acquisition of the knowledge correlation index is realized, and further the accuracy and reliability of the knowledge graph are improved.

[0022] 3. By combining the relational database query language to extract structured data from the knowledge graph and perform frequency statistics, encoding and classifying the structured data after frequency analysis and visually displaying the frequency distribution of the structured data in the form of a chart, then generating standard information data based on the results of the encoding and classification within a preset time period and validating it, and finally updating the entity information and relationship information in the knowledge graph in real time according to the validation results, the improvement of the update timeliness of the knowledge graph is realized, and further the more accurate acquisition of the standard information data is realized. Brief Description of the Drawings

[0023] Figure 1 It is a flowchart of a standard information management method based on a knowledge graph provided by an embodiment of the present application;

[0024] Figure 2 It is a flowchart for obtaining the node text matching degree provided by an embodiment of the present application;

[0025] Figure 3 It is a statistical chart of the change of the knowledge correlation index provided by an embodiment of the present application;

[0026] Figure 4 It is a flowchart for obtaining the standard information data provided by an embodiment of the present application;

[0027] Figure 5 It is a schematic structural diagram of a standard information management system based on a knowledge graph provided by an embodiment of the present application. Detailed Embodiment

[0028] Embodiments of the present application provide a method and system for standard information management based on a knowledge graph, which solve the problem of low accuracy in obtaining standard information data in the prior art. By collecting structured data and unstructured data to be managed within a preset time period and performing preprocessing, knowledge extraction and knowledge processing are carried out simultaneously to obtain the schema layer of the knowledge graph to be constructed, and the information data after knowledge processing is integrated with the constructed schema layer into a schema framework. Then, a corresponding knowledge graph is generated in combination with the corresponding ontology structure and knowledge relevance index. Next, statistical analysis is performed on the knowledge graph, and data similarity is obtained according to the results of the statistical analysis. At the same time, the knowledge relationship information in the knowledge graph is updated to obtain standard information data, and the standard information data in the knowledge graph is displayed in real time according to the requests of the accessing users. Finally, the constructed knowledge graph is updated in real time in combination with the feedback results of the users and the preset standard information indicators until the preset standard information indicators are met, achieving an improvement in the accuracy of obtaining standard information data.

[0029] The technical solution in the embodiments of the present application for solving the problem of low accuracy in obtaining the above standard information data has the following general idea:

[0030] By collecting information data to be managed within a preset time period and performing preprocessing to obtain the schema layer of the knowledge graph to be constructed, then conceptually abstracting the schema layer of the knowledge graph to be constructed and generating a corresponding knowledge graph in combination with the corresponding ontology structure and knowledge relevance index, and finally updating the knowledge relationship information in the knowledge graph to obtain standard information data, and iteratively optimizing the constructed knowledge graph in combination with the feedback results of the users and the preset standard information indicators, the effect of improving the accuracy of obtaining standard information data is achieved.

[0031] To better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0032] As Figure 1As shown in the figure, it is a flowchart of a standard information management method based on a knowledge graph provided by an embodiment of the present application. The method includes the following steps: S1, collect the information data to be managed within a preset time period and perform preprocessing. At the same time, perform knowledge extraction and knowledge processing on the preprocessed information data to obtain the schema layer of the knowledge graph to be constructed. The information data includes structured data and unstructured data. The preprocessing is used to remove duplicate data and noise data in the information data. The knowledge extraction includes entity recognition and relationship extraction. The knowledge processing is used to perform knowledge classification and knowledge reasoning on the preprocessed information data. The schema layer of the knowledge graph to be constructed is used to instantiate the preprocessed information data according to a preset knowledge graph schema to form a semantic network of a graph structure; S2, perform concept abstraction on the schema layer of the knowledge graph to be constructed and generate a corresponding knowledge graph in combination with the corresponding ontology structure and knowledge relevance index. The probability abstraction is used to integrate the information data after knowledge processing with the constructed schema layer into a schema framework. The knowledge relevance index is used to measure the degree of association between the schema layer of the knowledge graph to be constructed and the knowledge graph within a preset time period. The knowledge graph is used to visualize the complexity change relationship between information data within a preset time period; S3, perform statistical analysis on the knowledge graph and obtain data similarity according to the results of the statistical analysis. At the same time, update the knowledge relationship information in the knowledge graph according to the data similarity to obtain standard information data. The data similarity is used to measure the similarity degree between information data in the knowledge graph. The statistical analysis represents the occurrence frequency of structured data in the statistical information data in the knowledge graph. The standard information data is used to update the knowledge relationship information in the knowledge graph within a preset time period. The knowledge relationship information represents the similar knowledge information between the information data and the standard information data within a preset time period; S4, perform interactive management on the standard information data and perform iterative optimization on the constructed knowledge graph in combination with the feedback results of the user and preset standard information indicators. The interactive management means to display the standard information data in the knowledge graph in real time according to the request of the accessing user. The preset standard information indicators represent the reference standards set according to the performance of the knowledge graph within a preset time period. The iterative optimization means to update the standard information data in the knowledge graph in real time according to the feedback results of the accessing user within a preset time period until the preset standard information indicators are met.

[0033] In this embodiment, structured data usually represents tabular data in information data, while unstructured data usually represents text, picture, and video data. In practical applications, natural language processing technology is usually used to extract entity, relationship, and event information from text data. For picture and video data, image recognition technology is usually used to extract knowledge relationship information. The specific construction steps of the ontology structure are as follows: start constructing the top-level ontology from the topmost concept, and then refine the concepts and relationships to form a well-structured concept hierarchy tree; extract entities, attributes, and relationships from open-linked data sources, and gradually abstract them upward into concepts to form a schema layer; extract entities and relationships from data based on a bottom-up method, then construct a concept hierarchy structure based on a top-down method, and finally integrate the two to optimize the schema layer; based on the construction of the ontology structure, concept abstraction is performed on the extracted entities, relationships, and attributes, which usually involves classifying and generalizing entities, defining and standardizing relationships, and unifying and standardizing attributes, helping to form a clear, consistent, and schema layer that can reflect the relevance between entities, and can more efficiently reflect the knowledge relationship field between information data, achieving an improvement in the accuracy of obtaining standard information data.

[0034] Furthermore, the specific acquisition process of the schema layer of the knowledge graph to be constructed is as follows: perform alignment processing on the information data after knowledge processing according to the ontology library of the information data to be managed within a preset time period to obtain a graph structure. The ontology library is used to structurally display the preprocessed information data. The alignment processing includes entity alignment and relationship alignment. The graph structure is used to visualize the relevance of the knowledge distribution in the information data after alignment processing; define the schema layer structure based on the obtained graph structure and in combination with the preset standard information management target, and at the same time store the defined schema layer structure in the ontology library. The schema layer structure is used to integrate the information data in the ontology library with the graph structure and set the corresponding storage format and query language; obtain the node text matching degree according to the relevance between the schema layer structures within a preset time period, and fill the information data after alignment processing with the nodes in the defined schema layer structure to obtain the schema layer of the knowledge graph to be constructed. The node text matching degree is used to measure the matching degree between the node text in the information data after alignment processing and the node text in the schema layer structure. The filling correspondence is used to fill the attribute values corresponding to the information data within a preset time period with respect to the schema layer structure.

[0035] In this embodiment, entity alignment is generally a process of identifying and merging similar entities from different data sources and different expressions. Relationship alignment is usually to ensure the consistency of the relationships between the same entities described in different data sources, which generally involves the mapping, transformation, and merging of relationship information. By visualizing the graph structure in the form of a chart, it is possible to clearly see the flow and distribution of knowledge relationship information between different entity information, as well as effective data management and query, so as to understand their relevance. Corresponding the nodes in the information data after alignment processing with the nodes in the pattern layer structure. Specifically, traverse each node in the information data, find the node in the pattern layer structure with the highest matching degree according to the mapping index, and then fill the node in the information data into the corresponding node in the pattern layer structure. The filling and correspondence of the attribute values are realized by extracting the corresponding attribute values from the information data and assigning them to the attributes of the corresponding nodes in the pattern layer structure, thus realizing a more accurate construction of the pattern layer.

[0036] Further, as Figure 2As shown in the figure, it is a flowchart for obtaining the node text matching degree provided by the embodiment of the present application. The specific steps for obtaining the node text matching degree are as follows: S21, perform word segmentation on the information data within a preset time period and the node text in the pattern layer structure, and at the same time obtain the string matching degree according to the results of the word segmentation. The word segmentation is used to remove the stop words in the node text and split the node text after removing the stop words into phrase strings. The string matching degree is used to measure the matching degree between the information data and the phrase strings in the pattern layer structure; S22, identify the entity semantic information in the information data and the node text of the pattern layer structure, and at the same time obtain the semantic similarity based on the cosine similarity of the vector space. The entity semantic information represents querying the corresponding semantic answer information in the node text according to the parsed access requirements of the access user. The semantic similarity is used to measure the similarity degree between the entity semantic information in the information data and the entity semantic information in the pattern layer structure; S23, extract the attribute values corresponding to the target node text attributes from the information data and the pattern layer structure and perform attribute verification, and at the same time determine whether the result of the attribute verification meets the preset requirements of the target node text. If it meets, stop the attribute verification and input the attribute value into the pattern layer. Otherwise, execute S24. The attribute verification includes data type verification and data format verification; S24, re-perform attribute verification on the attribute values that do not meet the preset requirements of the target node text to obtain the attribute verification values and perform data conversion according to the requirements of the pattern layer structure. At the same time, obtain the preset attribute verification correction factor from the database according to the preset attribute verification standard, and combine the string matching degree and the semantic similarity to obtain the node text matching degree. The attribute verification value is used to measure the accuracy index of the attribute verification. The data conversion is used to re-perform attribute verification on the attribute values that do not meet the preset requirements of the target node text until they meet the preset requirements of the target node text. The attribute verification correction factor is used to correct the error rate of the attribute values during the attribute verification process.

[0037] In this embodiment, the cosine similarity measures the cosine value of the angle between two vectors. The closer its value is to 1, the higher the similarity between the two vectors. For the entity semantic information with low similarity, alignment processing is performed, which usually includes linking the entities in the information data to the corresponding entities in the pattern layer structure, or filling the entity attribute information in the pattern layer structure into the information data; checking whether the data type of the attribute value matches the data type required by the target node text. For example, if the target node text requires an integer type, but the extracted attribute value is a string type, corresponding conversion or error reporting processing is required. For the attribute values that do not meet the format requirements, format conversion or error reporting processing is required; it should be noted that during the attribute verification process, the situation of incorrect attribute values is usually encountered. To correct these errors, an attribute verification correction factor can be introduced to correct the error rate of the attribute value during the attribute verification process. In practical applications, this correction factor can be a weight value or a proportional value, used to adjust the attribute value or determine whether further processing is required, achieving an improvement in the efficiency of obtaining the node text matching degree.

[0038] Furthermore, the node text matching degree is calculated by the following formula:

[0039]

[0040] In the formula, g is the number of the preset time period, g = 1, 2,..., G, and G is the total number of the preset time periods, YI g represents the node text matching degree of the information data in the g-th preset time period, e is the natural constant, A g represents the string matching degree of the information data in the g-th preset time period, A 0 represents the string reference matching degree, B g represents the semantic similarity of the information data in the g-th preset time period, B 0 represents the semantic reference similarity, C g represents the attribute verification value of the information data in the g-th preset time period. The attribute verification value represents the matching degree between the node text and the preset target node text. For example, the similarity score between a certain text paragraph in the information data and the target node text in the pattern layer, C 0 represents the attribute reference verification value, and α represents the attribute verification correction factor.

[0041] In this embodiment, for simplicity of analysis, wherein, A1 g represents the string matching degree coefficient, B1 g represents the semantic similarity coefficient, C1 g represents the attribute verification coefficient. The change statistical table of the node text matching degree is shown in Table 1:

[0042] Statistical Table of Changes in Node Text Matching Degree in Table 1

[0043]

[0044] It should be understood that the string reference matching degree and semantic reference similarity are usually obtained by taking the average after being statistically calculated through existing historical well - operating information data. The attribute reference verification value is preset in the database of this system, and its specific data can be obtained by performing weighted summation according to the historical attribute verification values in the database; in a specific embodiment, by fitting the verification relationship between the attribute verification value and the node text matching degree during the attribute verification process based on a large amount of historical information data to obtain a fitting curve, and then real - time monitoring the change of data attributes during the attribute verification process and inputting the attribute verification value into the fitting curve to obtain the corresponding attribute verification correction factor. It should be noted that the attribute verification correction factor is usually a value greater than 1, while the attribute verification value is usually a value less than 1. The attribute verification value represents the verification result of data attributes (fields and features) during the attribute verification process in information data and the pattern - layer structure. When this occurs, the attribute verification correction factor is also 0, indicating that there is no attribute value error during the attribute verification process. When and this is the case, the node text matching degree is equal to 2.1. At this time, the matching degree between the node text in the information data after alignment processing and the node text in the pattern - layer structure is the highest, achieving a more accurate acquisition of the node text matching degree, further realizing an improvement in the acquisition accuracy of standard information data, and effectively solving the problem of low acquisition accuracy of standard information data in the prior art.

[0045] Furthermore, the specific steps for generating the knowledge graph include: performing position mapping on the pre - processed information data according to the ontology structure of the pattern layer of the to - be - constructed knowledge graph, and simultaneously generating the structural framework of the to - be - constructed knowledge graph based on the result of the position mapping. The position mapping is used to map the entity information and relationship information in the information data to the corresponding positions in the pattern layer of the to - be - constructed knowledge graph; obtaining the knowledge correlation index based on the co - occurrence frequency between entity information in the ontology structure and performing information verification on the generated structural framework, and simultaneously improving the attribute relationship between information data according to the result of the information verification to generate the knowledge graph. The information verification is used to remove redundant information in the information data and the structural framework, and the attribute relationship is used to visualize the connection relationship between entity information and relationship information in the information data.

[0046] In this embodiment, the specific steps of position mapping include: classifying the extracted entities according to the entity categories in the ontology structure and mapping each entity to the corresponding entity type; classifying the extracted relationships and mapping each relationship to the corresponding relationship type according to the relationship types in the ontology structure; mapping the attribute information of the entities to the corresponding attribute fields in the ontology structure; the specific steps of information verification for the generated structure framework include: verifying whether the relationships between entities in the structure framework conform to the actual knowledge, including the direction and type of the relationships; checking whether the relevance between entities is reasonable according to the knowledge relevance index, and whether there are missing or incorrect relationships; according to the results of information verification, improving the attribute relationships between information data, and the specific steps include: for the missing relationships found in the verification process, adding the corresponding relationships according to the data characteristics and business requirements; for the incorrect relationships found in the verification process, correcting or deleting them; for the inaccurate or imperfect attribute information found in the verification process, refining or supplementing it according to the data characteristics and business requirements; through the processing and improvement of the above steps, the finally generated knowledge graph can comprehensively and accurately express the domain knowledge, achieving the improvement of the accuracy and reliability of the knowledge graph.

[0047] Furthermore, the knowledge relevance index is obtained through the following method: obtaining the concept coverage rate based on the number of knowledge concepts in the knowledge graph and the total number of knowledge probabilities in the schema layer of the knowledge graph to be constructed, and at the same time obtaining the entity matching degree according to the alignment situation between the schema layer of the knowledge graph to be constructed and the entity information in the knowledge graph. The concept coverage rate is used to measure the proportion of the number of knowledge concepts in the knowledge graph in the total number of knowledge concepts in the schema layer of the knowledge graph to be constructed, and the entity matching degree is used to measure the similarity between the entity information in the schema layer of the knowledge graph to be constructed and the entity information in the knowledge graph; performing a consistency score on the relationship information according to the coincidence degree between the schema layer of the knowledge graph to be constructed and the knowledge graph in the preset relationship information, and at the same time setting the corresponding proportion weight based on the result of the consistency score to obtain the relationship consistency, and combining the concept coverage rate and the entity matching degree to obtain the knowledge relevance index. The consistency score is an index used to quantify the coincidence degree between the schema layer of the knowledge graph to be constructed and the knowledge graph in the preset relationship information, and the relationship consistency is used to measure the coincidence degree between the relationship information in the schema layer and the knowledge graph and the preset relationship information.

[0048] In this embodiment, the knowledge relevance index is calculated through the following formula:

[0049]

[0050] In the formula, g is the number of the preset time period, g = 1, 2,..., G, and G is the total number of the preset time periods, ZH grepresents the knowledge relevance index of the knowledge graph to be constructed within the g-th preset time period, where e is the natural constant, and M g represents the entity matching degree of the entity information in the knowledge graph to be constructed within the g-th preset time period, and M 0 represents the entity reference matching degree, and N g represents the relationship consistency of the relationship information in the knowledge graph to be constructed within the g-th preset time period, and N 0 represents the relationship reference consistency, and U g represents the concept coverage rate of the knowledge concepts in the knowledge graph to be constructed within the g-th preset time period. Among them, the reference entity matching degree and the relationship reference consistency are usually obtained by taking the average after statistical analysis of the existing entity matching degree data and relationship consistency data. In practical applications, the reference values in the formula need to be re-obtained or corrected according to the actual management requirements of the standard information.

[0051] Specifically, when holds, the change statistical graph of the knowledge relevance index is as shown in Figure 3 . Among them, the entity matching degree coefficient represents the ratio of the entity matching degree to the entity reference matching degree (i.e., ), and the relationship consistency coefficient represents the ratio of the relationship consistency to the relationship reference consistency (i.e., ). It should be understood that when the entity matching degree coefficient acts alone, the corresponding knowledge relevance index does not change significantly (when the entity matching degree coefficient is equal to 1, the knowledge relevance index is 0.82). When the relationship consistency coefficient acts alone, the corresponding knowledge relevance index increases along a straight line at a 45° angle (i.e., y = x). When holds, the knowledge relevance index is 5.17. At this time, the association degree between the pattern layer of the knowledge graph to be constructed and the knowledge graph within the preset time period is the highest, achieving an improvement in the accuracy and reliability of the knowledge relevance index, further realizing an improvement in the accuracy of obtaining standard information data, and effectively solving the problem of low accuracy in obtaining standard information data in the prior art.

[0052] Further, the specific steps for obtaining the data similarity include: converting the entity information and relationship information in the knowledge graph into vectors, combining the corresponding spatial angles to obtain the entity vector values and relationship vector values, and simultaneously monitoring in real time the changes in the spatial positions of the entity information and relationship information in the knowledge graph during the vector conversion process and obtaining a preset cosine similarity correction factor from the database. The vector conversion is used to convert the numerical data in the entity information and relationship information in the knowledge graph into vector representations. The spatial angle represents the vector angle corresponding to the entity information and relationship information in the knowledge graph. The cosine similarity correction factor is used to correct the matching degree of the entity information and relationship information vectors in the knowledge graph relative to the knowledge graph. Monitor in real time the changes in the information data during the construction of the knowledge graph, and at the same time obtain the structural coefficient according to the result of the position mapping and combine it with the preset cosine similarity correction factor to obtain the data similarity. The structural coefficient is used to measure the consistency index of the structured data in the information data in the knowledge graph. The data similarity is calculated by the following formula:

[0053]

[0054] In the formula, g is the number of the preset time period, g = 1, 2,..., G, and G is the total number of the preset time periods, TN g represents the data similarity of the knowledge graph in the g-th preset time period, δ represents the cosine similarity correction factor, δ≥1, e is the natural constant, H g represents the entity vector of the entity information in the g-th preset time period, K g represents the relationship vector of the relationship information in the g-th preset time period, W g represents the structural coefficient of the structured data in the knowledge graph in the g-th preset time period, W 0 represents the structural reference coefficient.

[0055] In this embodiment, the cosine similarity is a commonly used method for measuring the similarity between two vectors. However, in some cases, directly calculating the cosine similarity is usually not accurate enough, and a correction factor (i.e., the cosine similarity correction factor) needs to be introduced, such as listening to the update events of the graph, regularly scanning the graph to detect changes, etc.; the structural coefficient usually refers to the influence of the organizational structure of entities and relationships in the knowledge graph on the similarity calculation. For example, the distance between two entities in the graph and the types of relationships they share will affect their similarity. The specific implementation process usually varies according to the requirements of the application scenario; θ represents the spatial angle between the entity vector and the relationship vector, then θ ∈ [0°, 90°). In a specific embodiment, the cosine similarity correction factor in the database can expand the influence degree of the spatial angle between the entity vector and the relationship vector. A mapping table can be formed between the spatial angle between the entity vector and the relationship vector and the cosine similarity correction factor according to the required sensitivity. The cosine value corresponding to the real-time spatial angle between the entity vector and the relationship vector is input into the mapping table to obtain the corresponding cosine similarity correction factor. It should be noted that the cosine similarity correction factor is usually a value greater than or equal to 1 and increases as cosθ increases. When cosθ = 1, the cosine similarity correction factor is equal to 1, that is, when and When, the data similarity is equal to 2.25, indicating that the similarity degree between information data in the knowledge graph is the highest, and the accuracy of data similarity is improved.

[0056] Furthermore, as Figure 4 shown, it is a flowchart for obtaining standard information data provided by an embodiment of the present application. The specific process for obtaining standard information data is as follows: Extract structured data from the knowledge graph according to the data similarity and in combination with relational database query language, and perform frequency statistics. Frequency statistics represent the frequency of occurrence of structured data in information data within a preset time period in the knowledge graph; Code and classify the structured data after frequency analysis according to preset standard information indicators, and visually display the frequency distribution of the structured data in the form of a chart; Generate standard information data in combination with the results of coding and classification within a preset time period and perform verification. At the same time, perform real-time updates on the entity information and relationship information in the knowledge graph according to the verification results. Real-time updates include supplementing missing information and adding new data.

[0057] In this embodiment, before performing frequency statistics, statistical analysis of the structured data is also required, which specifically includes: directly counting the total number of entity information and relationship information in the knowledge graph and recording the results; Dividing the total number of relationship information by the total number of entity information to obtain the number of relationship information owned by each entity information on average; Grouping entity information according to entity information types, and counting the number of each group and its proportion (i.e., the average knowledge relationship number); Grouping relationship information according to relationship information types, and counting the number of each group and its proportion; At the same time, display the results of statistical analysis in a clear and easy-to-read manner. As shown in Table 2, it is a statistical table of the change in the proportion of structured data corresponding to different data similarities during the statistical analysis process:

[0058] Table 2 Statistical table of the change in the proportion of structured data

[0059]

[0060]

[0061] It should be understood that by analyzing the distribution of entity information types and relationship information types, the main content and domain coverage of the knowledge graph can be understood. Secondly, by analyzing the average number of knowledge relationships, the tightness of the connection between entity information can be evaluated. In addition, by comparing the statistical analysis results of different time periods, the change trend of the knowledge graph can be observed. The statistical analysis of the knowledge graph can be systematically carried out using data similarity, realizing more accurate acquisition of standard information data, further improving the accuracy of standard information data acquisition, and effectively solving the problem of low accuracy of standard information data acquisition in the prior art.

[0062] Furthermore, the specific steps for iterative optimization of the constructed knowledge graph include: monitoring the performance of the knowledge graph in real time according to the access query process of the access user to the knowledge graph within a preset time period to obtain performance data, where the performance data includes access query time data and throughput data, and the throughput data represents the number of access queries that the knowledge graph can process within a preset time period; positioning the performance bottleneck in the access query process of the knowledge graph based on the obtained performance data and combining with the real-time feedback results of the user and making a capacity expansion plan, and the capacity expansion plan includes load balancing and distributed deployment; optimizing the construction algorithm and query algorithm of the knowledge graph in combination with the results of the capacity expansion plan until the preset standard information index is met.

[0063] In this embodiment, according to user feedback and the identified problems, the standard information data in the knowledge graph is updated in real time. If necessary, the structure of the knowledge graph can be adjusted to better meet the needs of users and improve performance. This usually includes modifying the definitions and representation methods of entities, relationships, and attributes, or adjusting the concept hierarchy. After each iterative optimization, the performance of the knowledge graph and user satisfaction are re-evaluated, and the construction algorithm and query algorithm in the knowledge graph are tested using standard information data. Usually, according to the response time of the access query operation of the real-time monitoring knowledge graph, and by gradually increasing the scale of the data set to observe the changes in the performance and response time of the query algorithm. If the preset standard information index is reached, it can be considered that the knowledge graph has been optimized. Otherwise, iterative optimization needs to continue. It should be noted that the entire iterative optimization process should be a cyclic process, by continuously collecting user feedback, evaluating performance, identifying problems, updating data and structure, and re-evaluating until the preset standard information index is met, which helps to ensure that the knowledge graph always remains consistent with the needs of users, continuously improves its performance and user satisfaction, and at the same time, can also continuously discover and solve the problems and deficiencies existing in the knowledge graph, realizing the improvement of the accuracy and reliability of the knowledge graph optimization.

[0064] Such as Figure 5As shown in the figure, it is a schematic structural diagram of a standard information management system based on a knowledge graph provided by an embodiment of the present application. The standard information management system based on a knowledge graph provided by an embodiment of the present application includes: a schema layer acquisition module for the knowledge graph to be constructed, a knowledge graph generation module, a standard information data acquisition module, and a knowledge graph iterative optimization module; among them, the schema layer acquisition module for the knowledge graph to be constructed is used to collect information data to be managed within a preset time period and perform preprocessing, and at the same time perform knowledge extraction and knowledge processing on the preprocessed information data to obtain the schema layer of the knowledge graph to be constructed. The information data includes structured data and unstructured data. The preprocessing is used to remove duplicate data and noise data in the information data. The knowledge extraction includes entity recognition and relationship extraction. The knowledge processing is used to perform knowledge classification and knowledge reasoning on the preprocessed information data. The schema layer of the knowledge graph to be constructed is used to instantiate the preprocessed information data according to a preset knowledge graph schema to form a semantic network of a graph structure; the knowledge graph generation module is used to perform concept abstraction on the schema layer of the knowledge graph to be constructed and generate a corresponding knowledge graph in combination with the corresponding ontology structure and knowledge correlation index. The probability abstraction is used to integrate the information data after knowledge processing with the constructed schema layer into a schema framework. The knowledge correlation index is used to measure the degree of association between the schema layer of the knowledge graph to be constructed and the knowledge graph within a preset time period. The knowledge graph is used to visualize the complexity change relationship between information data within a preset time period; the standard information data acquisition module is used to perform statistical analysis on the knowledge graph and obtain data similarity according to the results of the statistical analysis, and at the same time update the knowledge relationship information in the knowledge graph according to the data similarity to obtain standard information data. The data similarity is used to measure the similarity degree between information data in the knowledge graph. The statistical analysis represents the occurrence frequency of structured data in the information data in the knowledge graph. The standard information data is used to update the knowledge relationship information in the knowledge graph within a preset time period. The knowledge relationship information represents the similar knowledge information between the information data and the standard information data within a preset time period; the knowledge graph iterative optimization module is used to perform interactive management on the standard information data and perform iterative optimization on the constructed knowledge graph in combination with the feedback results of the user and the preset standard information indicators. The interactive management means to display the standard information data in the knowledge graph in real time according to the requests of the accessing users. The preset standard information indicators represent the reference standards set according to the performance of the knowledge graph within a preset time period. The iterative optimization means to perform real-time update on the standard information data in the knowledge graph according to the feedback results of the accessing users within a preset time period until the preset standard information indicators are met.

[0065] In this embodiment, according to the calculated data similarity, the standard information data acquisition module will identify possible duplicate, redundant, and inconsistent knowledge relationship information, and update or merge it to ensure the accuracy and consistency of the knowledge graph. During the process of updating the knowledge relationship information, this module will obtain new standard information data as needed to supplement or correct the content in the knowledge graph; the iterative optimization module of the knowledge graph supports interactive management of the knowledge graph by the access user (including browsing, querying, editing, and modifying). The access user can add, delete, or modify entities, relationships, and attributes in the knowledge graph through the interface. Then, this module will evaluate and compare the knowledge graph according to preset standard information metrics (including accuracy, integrity, and consistency). These metrics can help the module identify problems and deficiencies in the knowledge graph, achieving an improvement in the timeliness of accessing the knowledge graph.

[0066] In summary, the embodiment of the present application collects the information data to be managed within a preset time period and performs preprocessing to obtain the schema layer of the knowledge graph to be constructed, then generates the corresponding knowledge graph in combination with the corresponding ontology structure and knowledge correlation index, and finally updates the knowledge relationship information in the knowledge graph to obtain the standard information data and iteratively optimize the constructed knowledge graph, thereby achieving an improvement in the accuracy of the schema layer of the knowledge graph to be constructed, and further achieving an improvement in the accuracy of obtaining the standard information data, effectively solving the problem of low accuracy of obtaining the standard information data in the prior art.

[0067] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can be implemented in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can be implemented in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0069] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements in the process Figure 1 a process or processes and / or blocks Figure 1 the functions specified in a block or blocks.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in a process Figure 1 a process or processes and / or blocks Figure 1 a block or blocks.

[0071] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0072] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A standard information management method based on knowledge graph, characterized in that: The following steps are involved: S1, collect information data to be managed within a preset time period and preprocess it, and at the same time perform knowledge extraction and knowledge processing on the preprocessed information data to obtain the model layer of the knowledge graph to be constructed; S2, performing conceptual abstraction on the pattern layer of the knowledge graph to be constructed and generating a corresponding knowledge graph in combination with the corresponding ontology structure and the knowledge relevance index, wherein the knowledge relevance index is used to measure the degree of association between the pattern layer of the knowledge graph to be constructed and the knowledge graph within a preset time period; S3, performing statistical analysis on the knowledge graph and obtaining data similarity based on the results of the statistical analysis, and updating the knowledge relationship information in the knowledge graph based on the data similarity to obtain standard information data, wherein the data similarity is used to measure the similarity between information data in the knowledge graph; S4, interactively manage the standard information data and iteratively optimize the constructed knowledge graph based on user feedback and preset standard information indicators; The specific steps for obtaining the node text matching degree are: S21, performing word segmentation processing on the information data within a preset time period and the node text in the pattern layer structure, and obtaining a string matching degree according to the result of the word segmentation processing, wherein the string matching degree is used to measure the matching degree between the information data and the phrase string in the pattern layer structure; S22, identifying entity semantic information in the information data and node text of the pattern layer structure, and obtaining semantic similarity based on cosine similarity in the vector space; S23, extracting the attribute value corresponding to the target node text attribute from the information data and the pattern layer structure and performing attribute verification, and judging whether the result of the attribute verification meets the preset target node text requirement, if so, stopping the attribute verification and inputting the attribute value into the pattern layer, otherwise executing S24; S24, re-checking the attribute values ​​that do not meet the preset target node text requirements to obtain attribute check values ​​and perform data conversion according to the requirements of the model layer structure, and at the same time obtaining a preset attribute check correction factor from the database according to the preset attribute check standard, and combining the string matching degree and the semantic similarity to obtain the node text matching degree; The knowledge relevance index is obtained by the following method: The concept coverage is obtained based on the number of knowledge concepts in the knowledge graph and the total number of knowledge probabilities in the pattern layer of the knowledge graph to be constructed, and the entity matching degree is obtained based on the alignment of the pattern layer of the knowledge graph to be constructed and the entity information in the knowledge graph; The relationship information is scored for consistency based on the degree of agreement between the model layer of the knowledge graph to be constructed and the knowledge graph in terms of the preset relationship information. At the same time, the corresponding weight is set based on the result of the consistency score to obtain the relationship consistency, and the knowledge relevance index is obtained by combining the concept coverage and entity matching degree. The specific steps of obtaining the data similarity include: The entity information and relationship information in the knowledge graph are transformed into vectors and the entity vector value and relationship vector value are obtained by combining the corresponding spatial angles. At the same time, the spatial position changes of the entity information and relationship information in the knowledge graph during the vector transformation process are monitored in real time and the preset cosine similarity correction factor is obtained from the database. Monitor the changes of information data in real time during the construction of the knowledge graph, obtain the structural coefficient based on the results of position mapping, and combine it with the preset cosine similarity correction factor to obtain the data similarity.

2. A standard information management method based on knowledge graph as claimed in claim 1, characterized in that: The specific acquisition process of the pattern layer of the knowledge graph to be constructed is: Align the information data after knowledge processing according to the ontology library of the information data to be managed within a preset time period to obtain a graph structure; Based on the acquired graph structure and in combination with the preset standard information management target, a pattern layer structure is defined, and the defined pattern layer structure is stored in the ontology library; The node text matching degree is obtained according to the correlation between the pattern layer structures within a preset time period, and the information data after alignment is filled with the nodes in the defined pattern layer structure to obtain the pattern layer of the knowledge graph to be constructed. The node text matching degree is used to measure the matching degree between the node text in the information data after alignment and the node text in the pattern layer structure.

3. A standard information management method based on knowledge graph as claimed in claim 2, characterized in that: The node text matching degree is calculated by the following formula: Where g is the number of the preset time period, g = 1, 2, ..., G, G is the total number of preset time periods, YI g represents the node text matching degree of information data in the g-th preset time period, e is a natural constant, A g represents the string matching degree of the information data in the g-th preset time period, A0 represents the string reference matching degree, B g represents the semantic similarity of information data in the g-th preset time period, B0 represents the semantic reference similarity, C g represents the attribute verification value of the information data in the g-th preset time period, C0 represents the attribute reference verification value, and α represents the attribute verification correction factor.

4. A standard information management method based on knowledge graph as claimed in claim 1, characterized in that: The specific steps of generating the knowledge graph include: Map the preprocessed information data to the ontology structure of the model layer of the knowledge graph to be constructed, and generate the structural framework of the knowledge graph to be constructed based on the result of the position mapping; The knowledge relevance index is obtained according to the co-occurrence frequency between entity information in the ontology structure and the generated structural framework is verified. At the same time, according to the results of information verification, the attribute relationship between information data is improved to generate a knowledge graph.

5. A standard information management method based on knowledge graph as claimed in claim 1, characterized in that: The specific steps of obtaining the data similarity also include: The data similarity is calculated using the following formula: Where g is the number of the preset time period, g = 1, 2, ..., G, G is the total number of preset time periods, TN g represents the data similarity of the knowledge graph in the g-th preset time period, δ represents the cosine similarity correction factor, δ≥1, e is a natural constant, H g represents the entity vector of the entity information in the g-th preset time period, K g Represents the relationship vector of the relationship information in the g-th preset time period, W g It represents the structural coefficient of the structured data in the g-th preset time period in the knowledge graph, and W0 represents the structural reference coefficient.

6. A standard information management method based on knowledge graph as claimed in claim 1, characterized in that: The specific process of obtaining the standard information data is as follows: Extract structured data from the knowledge graph based on data similarity and in combination with relational database query language and perform frequency statistics; According to the preset standard information indicators, the structured data after frequency analysis is coded and classified, and the frequency distribution of the structured data is visualized in the form of charts; Based on the results of coding and classification within a preset time period, standard information data is generated and verified. At the same time, the entity information and relationship information in the knowledge graph are updated in real time according to the verification results.

7. A standard information management method based on knowledge graph as claimed in claim 1, characterized in that: The specific steps of iteratively optimizing the constructed knowledge graph include: Monitor the performance of the knowledge graph in real time to obtain performance data based on the access query process of the users to the knowledge graph within a preset time period; Based on the acquired performance data and combined with the real-time feedback from users, we can locate the performance bottlenecks of the knowledge graph during the access query process and make expansion plans. Combined with the results of the expansion plan, the knowledge graph construction algorithm and query algorithm are optimized until the preset standard information indicators are met.

8. A standard information management system based on knowledge graph, used to execute the method according to any one of claims 1 to 7, characterized in that: include: The model layer acquisition module, knowledge graph generation module, standard information data acquisition module and knowledge graph iterative optimization module of the knowledge graph to be constructed; The pattern layer acquisition module of the knowledge graph to be constructed is used to collect and preprocess the information data to be managed within a preset time period, and at the same time perform knowledge extraction and knowledge processing on the preprocessed information data to obtain the pattern layer of the knowledge graph to be constructed; The knowledge graph generation module is used to perform conceptual abstraction on the pattern layer of the knowledge graph to be constructed and generate a corresponding knowledge graph in combination with the corresponding ontology structure and the knowledge relevance index, and the knowledge relevance index is used to measure the degree of association between the pattern layer of the knowledge graph to be constructed and the knowledge graph within a preset time period; The standard information data acquisition module is used to perform statistical analysis on the knowledge graph and obtain data similarity based on the results of the statistical analysis, and at the same time update the knowledge relationship information in the knowledge graph based on the data similarity to obtain standard information data. The data similarity is used to measure the similarity between information data in the knowledge graph; The knowledge graph iterative optimization module is used to interactively manage standard information data and iteratively optimize the constructed knowledge graph in combination with user feedback results and preset standard information indicators.

Images