A knowledge base management system and enhancement method

Abstract

The present invention discloses a knowledge base management system and an enhancement method, belonging to the field of data processing technology, which specifically includes: obtaining user query information containing multiple semantic elements, and determining its IT service scenario based on the semantic elements and a pre-built IT service knowledge graph; dividing the query information into high-priority and low-priority semantic information areas according to the knowledge base dimension; quickly locating core knowledge based on the high-priority semantic information area, and improving the search results in combination with the low-priority semantic information area; triggering a multi-semantic query if a semantic element has multiple semantics; the present invention effectively improves the accuracy and efficiency of IT service knowledge retrieval and semantic understanding, and enhances the application value of the knowledge base.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a knowledge base management system and an enhancement method. Background Art

[0002] As the core system for storing and managing IT service-related knowledge, the performance of the IT service knowledge base directly affects the response speed and problem-solving capabilities of IT services.

[0003] However, traditional IT service knowledge bases face numerous challenges. For one thing, knowledge is typically stored in structured or semi-structured formats, lacking semantic connections. This makes it difficult to accurately understand the intent of complex user queries, leading to inaccurate or incomplete search results. For example, when a user enters a query containing multiple semantic elements, the knowledge base cannot identify the connections between the elements and cannot provide matching solutions. Furthermore, knowledge in the IT service field updates rapidly, while traditional knowledge bases lag behind, making them unable to respond to emerging issues. Furthermore, knowledge reuse and sharing are insufficient, resulting in wasted resources.

[0004] While existing information retrieval and semantic understanding technologies have been introduced for knowledge base optimization, they still have limitations. Traditional information retrieval relies on keyword matching and fails to understand query semantics and context. It can easily return irrelevant results for ambiguous keywords and struggles to process complex queries. Furthermore, natural language processing technology, limited by the flexibility of natural language and domain knowledge, struggles to accurately understand user intent. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention proposes a knowledge base management system and an enhancement method, which obtains user query information containing multiple semantic elements, determines its IT service scenario based on the semantic elements and the pre-built IT service knowledge graph; divides the query information into high- and low-priority semantic information areas according to the knowledge base dimension; quickly locates core knowledge based on the high-priority semantic information area, and improves the search results in combination with the low-priority semantic information area; triggers multi-semantic query if the semantic element has multiple semantics; the present invention effectively improves the accuracy and efficiency of IT service knowledge retrieval and semantic understanding, and enhances the application value of the knowledge base.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A knowledge base enhancement method, comprising:

[0008] Obtaining user query information; the user query information includes multiple semantic elements expressing the user's intention;

[0009] Determine IT service scenarios for all semantic elements based on the semantic elements and the pre-built IT service knowledge graph;

[0010] Dividing user query information into a high-priority semantic information area and a low-priority semantic information area based on the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario; the knowledge base dimension includes a semantic matching degree dimension, a case reuse rate dimension, and a timeliness dimension;

[0011] According to the high-priority semantic information region and the low-priority semantic information region, IT service knowledge retrieval and semantic understanding are performed on the user query information.

[0012] Specifically, determining the IT service scenarios of all semantic elements based on the semantic elements and the pre-built IT service knowledge graph includes:

[0013] Preprocess the pre-built IT service knowledge graph and establish an index of the mapping relationship between semantic elements and IT service scenarios;

[0014] Match the extracted semantic elements with the mapping relationship index to find all IT service scenarios that cover the semantic elements.

[0015] If there are multiple IT service scenarios covering the semantic elements, the searched IT service scenarios are sorted according to the strength of association between the semantic elements and the commonality of the IT service scenarios, and the IT service scenario that best meets the user's intention expression is selected as the final IT service scenario.

[0016] Specifically, dividing the user query information into a high-priority semantic information area and a low-priority semantic information area according to the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario includes:

[0017] For a specific IT service scenario, analyze the importance of different semantic elements in terms of semantic matching, case reuse rate, and timeliness;

[0018] Calculate the semantic similarity between each semantic element and the knowledge in the IT service knowledge graph, and classify the semantic elements with semantic similarity higher than the preset semantic similarity threshold into high-priority semantic information areas;

[0019] Count the frequency and reuse of each semantic element in historical IT service scenarios, and classify semantic elements with frequencies and reuse times exceeding preset frequency and reuse thresholds as high-priority semantic information areas;

[0020] Obtaining the update time of the information involved in each semantic element, and classifying the semantic elements whose update time and current time difference is less than a preset time difference threshold as high-priority semantic information areas;

[0021] Semantic elements that are not classified as high-priority semantic information areas are classified as low-priority semantic information areas.

[0022] Specifically, performing IT service knowledge retrieval and semantic understanding on user query information based on the high-priority semantic information region and the low-priority semantic information region includes:

[0023] First, locate the core knowledge in the IT service knowledge graph based on the high-priority semantic information areas, and then improve and supplement the search results by combining the low-priority semantic information areas;

[0024] If the identified semantic element has two semantics in the IT service knowledge graph, a multi-semantic query of the knowledge graph is triggered;

[0025] After a multi-semantic query on the knowledge graph, the system determines whether the semantic element is a polysemous term. If it is a polysemous term, the system displays polysemous options and related cases. If it is not a polysemous term, the system sorts the search results by the knowledge base dimension.

[0026] Return retrieval results with semantic labels, collect user feedback data, and feed the feedback data back to the IT service knowledge graph.

[0027] Specifically, the method first locates core knowledge in the IT service knowledge graph based on high-priority semantic information areas, and then improves and supplements the search results by combining low-priority semantic information areas, including:

[0028] Combining and associating semantic elements in high-priority semantic information areas to construct semantic query patterns;

[0029] Using the constructed semantic query pattern, search the IT service knowledge graph and find the core knowledge nodes that match the semantic query pattern through the graph traversal algorithm;

[0030] Filter and sort the found core knowledge nodes, calculate the comprehensive score of each core knowledge node based on the three dimensions of semantic matching, case reuse rate, and timeliness, and select the core knowledge node with the highest comprehensive score as the located core knowledge;

[0031] Conduct correlation analysis between the semantic elements in the low-priority semantic information area and the located core knowledge to find out the information that is not fully covered by the core knowledge;

[0032] According to the results of association analysis, low-priority semantic information area knowledge is extracted from the IT service knowledge graph as supplementary knowledge;

[0033] The supplementary knowledge is integrated with the located core knowledge to form a complete search result.

[0034] Specifically, if the identified semantic element has two semantics in the IT service knowledge graph, a knowledge graph multi-semantic query is triggered, including:

[0035] In the process of identifying the semantics of semantic elements, a semantic disambiguation algorithm is used to determine whether the semantic elements have two or more semantics;

[0036] If there are two semantics, different semantic query patterns are constructed according to the corresponding two semantics;

[0037] By using the two constructed semantic query modes, parallel queries are performed in the IT service knowledge graph to obtain two types of semantic knowledge.

[0038] Specifically, the use of a semantic disambiguation algorithm to determine whether a semantic element has two or more semantics includes: first extracting features from the context in which the semantic element is located, and then determining whether the semantic element has two or more semantics based on the extracted features.

[0039] Specifically, obtaining user query information includes:

[0040] Design user query access methods, including text input box, voice input interface, and graphical query interface;

[0041] When a user enters a query in the text input box, the word segmentation, part-of-speech tagging, and named entity recognition algorithms in natural language processing technology are used to extract multiple semantic elements that represent the user's intention from the input text;

[0042] When a user enters a query through the voice input interface, the voice signal is first converted into text. Then, using natural language processing technologies such as word segmentation, part-of-speech tagging, and named entity recognition algorithms, multiple semantic elements that represent the user's intended expression are extracted from the converted text.

[0043] When a user enters a query through the graphical query interface, multiple semantic elements expressing the user's intention are parsed through the interaction and association relationships of the interface elements.

[0044] Specifically, the construction process of the pre-built IT service knowledge graph includes:

[0045] Define entities, relationships, and attributes specific to the IT field and establish a knowledge model for the IT field;

[0046] Collect existing data in the IT service knowledge base, including structured data, semi-structured data, and unstructured data;

[0047] Preprocess the collected data, including data cleaning, data conversion and data integration;

[0048] Using knowledge extraction technology, entities, relationships, and attributes are extracted from pre-processed data and organized according to the knowledge model of the IT field;

[0049] Store the organized knowledge in the knowledge graph database to complete the construction of the IT service knowledge graph.

[0050] A knowledge base management system, comprising: a user query module, a service scenario determination module, a region division module, and a semantic understanding module;

[0051] The user query module is used to obtain user query information, pre-process the user query information, and extract semantic elements;

[0052] The service scenario determination module is used to determine the IT service scenarios of all semantic elements based on the extracted semantic elements and the pre-built IT service knowledge graph, and to sort and filter the IT service scenarios;

[0053] The region division module is configured to divide the user query information into a high-priority semantic information region and a low-priority semantic information region according to the determined IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario;

[0054] The semantic understanding module is used to perform IT service knowledge retrieval and semantic understanding on user query information based on high-priority semantic information areas and low-priority semantic information areas, return retrieval results with semantic tags, and collect user usage feedback data.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The present invention proposes a knowledge base management system and optimizes and improves the architecture, operation steps and processes. The system has the advantages of simple processes, low investment and operation costs, and low production work costs.

[0057] 2. The present invention proposes a knowledge base enhancement method. By obtaining user query information, the pre-built IT service knowledge graph is used to determine the IT service scenarios of semantic elements, and the user query information is divided into high- and low-priority semantic information areas according to the knowledge base dimensions, and then knowledge retrieval and semantic understanding are performed. This method can accurately locate the user's query intention, assign retrieval priority according to the importance of different semantic elements, improve retrieval efficiency and accuracy, quickly focus on core knowledge, avoid interference from irrelevant information, and provide users with retrieval results that are more in line with their needs.

[0058] 3. The present invention proposes a knowledge base enhancement method. When processing multi-semantic semantic elements, it triggers multi-semantic queries and displays multi-semantic options and related cases, which helps users clearly understand semantic differences and make accurate choices; at the same time, it collects user feedback and flows it back to the IT service knowledge graph, which can continuously optimize the knowledge graph and improve the accuracy and practicality of the knowledge base; in addition, multiple user query access methods and perfect semantic element extraction methods enhance the system's usability and adaptability to meet the needs of different users; and the standardized process of pre-constructing the IT service knowledge graph ensures the quality and reliability of the knowledge graph, providing a solid knowledge foundation for the entire method. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic diagram of a knowledge base enhancement method according to the present invention;

[0060] Figure 2 This is a principle flow chart of a knowledge base enhancement method of the present invention;

[0061] Figure 3 This is an architecture diagram of a knowledge base management system of the present invention. DETAILED DESCRIPTION

[0062] Example 1:

[0063] See also Figure 1 and Figure 2 The present invention provides an embodiment of a knowledge base enhancement method, the method comprising steps S1 to S4, including the following steps:

[0064] S1: Obtain user query information; the user query information includes multiple semantic elements expressing the user's intention;

[0065] S2: Determine IT service scenarios for all semantic elements based on the semantic elements and the pre-built IT service knowledge graph;

[0066] S3: Divide the user query information into a high-priority semantic information area and a low-priority semantic information area based on the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario; the knowledge base dimension includes a semantic matching degree dimension, a case reuse rate dimension, and a timeliness dimension;

[0067] S4: performing IT service knowledge retrieval and semantic understanding on the user query information according to the high-priority semantic information region and the low-priority semantic information region.

[0068] The obtaining of user query information specifically includes:

[0069] S1.1: Design user query access methods, including but not limited to text input boxes, voice input interfaces, and graphical query interfaces;

[0070] S1.2: When a user enters a query into the text input box, the user uses natural language processing techniques such as word segmentation, part-of-speech tagging, and named entity recognition to extract multiple semantic elements that represent the user's intended meaning from the input text. Natural language processing techniques are state-of-the-art in this field and do not constitute the inventive concept of this application, so their detailed description is omitted here.

[0071] S1.3: When a user enters a query through the voice input interface, the voice signal is first converted to text. Then, natural language processing algorithms such as word segmentation, part-of-speech tagging, and named entity recognition are used to extract multiple semantic elements that represent the user's intended expression from the converted text.

[0072] S1.4: When a user enters a query through the graphical query interface, the multiple semantic elements expressing the user's intention are parsed through the interaction and association relationships of the interface elements.

[0073] Determining IT service scenarios for all semantic elements based on the semantic elements and the pre-built IT service knowledge graph includes:

[0074] S2.1: Preprocess the pre-built IT service knowledge graph and establish an index of the mapping relationship between semantic elements and IT service scenarios;

[0075] The construction process of the pre-built IT service knowledge graph includes:

[0076] S2.1.1: Define entities, relationships, and attributes specific to the IT domain and establish a knowledge model for the IT domain;

[0077] S2.1.2: Collect existing data in the IT service knowledge base, including structured data, semi-structured data, and unstructured data;

[0078] S2.1.3: Preprocess the collected data, including data cleaning, data conversion, and data integration;

[0079] S2.1.4: Use knowledge extraction techniques to extract entities, relationships, and attributes from preprocessed data and organize them according to the knowledge model of the IT domain;

[0080] Furthermore, knowledge extraction technology is used to extract entities, relationships, and attributes from the preprocessed data, including:

[0081] (1) For structured data, directly extract entities, relationships, and attributes based on the structure and field information of the database table;

[0082] (2) For semi-structured data, such as XML and JSON files, extract entities, relationships, and attributes by parsing the structure and content of the files;

[0083] (3) For unstructured data, such as text reports and fault descriptions, named entity recognition, relationship extraction, and attribute extraction algorithms in natural language processing technology are used to extract entities, relationships, and attributes.

[0084] S2.1.5: Store the organized knowledge in the knowledge graph database to complete the construction of the IT service knowledge graph.

[0085] Furthermore, an index of mapping relationships between semantic elements and IT service scenarios is established, including:

[0086] (1) Collect IT service scenario information from internal enterprise IT service records, case libraries, technical documents, user feedback and other channels. For example, obtain common scenarios such as hardware fault repair, software system upgrades, and network configuration adjustments from IT service records;

[0087] (2) Extract semantic elements from users’ historical query records, professional terms in IT service documents, and FAQs. For example, extract semantic elements such as “printer” and “cannot print” from the user’s query “how to solve the problem of printer not printing”;

[0088] (3) Based on the collected IT service scenarios and semantic elements, the annotation personnel manually associate each semantic element with the relevant IT service scenario. At the same time, they establish annotation standards and review mechanisms to cross-verify and review the annotation results to ensure the accuracy and consistency of the annotations. For example, the semantic element "database connection failure" is associated with the "database troubleshooting and repair" scenario.

[0089] (4) Select an inverted index structure with semantic elements as keys and the corresponding IT service scenario list as values;

[0090] (5) Using the graph index method, semantic elements are regarded as nodes in the graph, and edge connections are established with related IT service scenario nodes. The weight of the edge can represent the association strength between the semantic element and the scenario, and a mapping relationship index structure is obtained.

[0091] S2.2: Match the extracted semantic elements with the mapping relationship index to find all IT service scenarios that cover the semantic elements.

[0092] In the present invention, a similarity method is used for matching, and the similarity method is a prior art in this field and is not an inventive solution of the present application, so it will not be described in detail here.

[0093] S2.3: If there are multiple IT service scenarios covering the semantic element, the searched IT service scenarios are sorted according to the strength of association between the semantic elements and the commonality of the IT service scenarios, and the IT service scenario that best matches the user's intent is selected as the final IT service scenario.

[0094] Among them, the commonness of IT service scenarios is based on the frequency of occurrence. It refers to the number of times the IT service scenario has occurred in the actual operation and maintenance and service processes in the past. For example, in an enterprise IT environment, the scenario of application lag caused by insufficient server memory may occur several times a week; while the system paralysis caused by a new ransomware attack may only occur once every few months. In contrast, the commonness of the scenario of insufficient server memory lag is higher because its frequency of occurrence is significantly higher than the latter. Moreover, the frequency of occurrence of different IT service scenarios can be calculated through statistical analysis of historical work orders, fault records, service requests and other data. The higher the frequency, the greater the commonness. For example, the number of work orders corresponding to various IT scenarios in the IT service management system in the past year is counted. Assuming that there are 200 work orders corresponding to network switch port failures and 50 work orders corresponding to data center precision air conditioning failures, the commonness of network switch port failure scenarios is relatively higher.

[0095] Furthermore, the specific steps of S2.3 include:

[0096] (1) Obtain semantic elements and count the number of times each pair of semantic elements co-occurs in the same document. For example, when analyzing fault reports, if database connection failure and server memory shortage appear in the same report, their co-occurrence count will increase.

[0097] (2) Create a two-dimensional matrix, where rows and columns represent different semantic elements, and the element values ​​in the matrix are the co-occurrence times of the corresponding semantic element pairs, to obtain the co-occurrence matrix;

[0098] (3) Convert the co-occurrence matrix into a vector form and calculate the cosine similarity between each pair of semantic elements. The cosine similarity calculation method is a prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0099] (4) Calculating the association strength by comprehensively considering the cosine similarity and the number of co-occurrences between the semantic elements; the association strength is the weighted sum of the first sum and the second sum; the first sum is the product of the cosine similarity between the semantic elements and the corresponding weight; the second sum is the ratio of the number of co-occurrences to the maximum value in the co-occurrence matrix multiplied by the corresponding weight;

[0100] (5) Count the number of times each IT service scenario occurs from data sources such as IT service records, user feedback, and historical cases. For example, count the number of times the "database troubleshooting and repair" scenario occurred in the past year.

[0101] (6) Calculate the frequency of each scene, that is, the ratio of the number of times the scene appears to the total number of times the scene appears;

[0102] (7) For each IT service scenario, count the semantic elements contained therein and calculate the sum of the correlation strengths between these semantic elements and the semantic elements in the user query to obtain the correlation between the semantic elements and the scenario;

[0103] (8) Perform a weighted summation of the relevance between semantic elements and scenes and the frequency of occurrence of each scene to obtain a comprehensive score for each scene;

[0104] (9) Sort all candidate IT service scenarios in descending order according to their comprehensive scores;

[0105] (10) The IT service scenario with the highest score is selected as the final scenario.

[0106] The dividing the user query information into a high-priority semantic information area and a low-priority semantic information area according to the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario includes:

[0107] S3.1: For a specific IT service scenario, analyze the importance of different semantic elements in the IT service scenario in terms of semantic matching, case reuse rate, and timeliness.

[0108] S3.2: Calculate the semantic similarity between each semantic element and the knowledge in the IT service knowledge graph, and classify semantic elements with semantic similarity higher than a preset semantic similarity threshold as high-priority semantic information areas;

[0109] S3.3: Count the frequency of occurrence and number of reuses of each semantic element in historical IT service scenarios, and classify semantic elements whose frequency of occurrence and number of reuses exceed the preset frequency and reuse thresholds as high-priority semantic information areas;

[0110] S3.4: Obtain the update time of the information involved in each semantic element, and classify the semantic elements whose update time and current time difference is less than a preset time difference threshold as high-priority semantic information areas;

[0111] S3.5: Classify semantic elements that are not classified as high-priority semantic information areas into low-priority semantic information areas.

[0112] The performing IT service knowledge retrieval and semantic understanding on the user query information based on the high-priority semantic information area and the low-priority semantic information area includes:

[0113] S4.1: First locate core knowledge in the IT service knowledge graph based on high-priority semantic information areas, and then improve and supplement the search results by combining low-priority semantic information areas;

[0114] S4.2: If the identified semantic element has two semantics in the IT service knowledge graph, a multi-semantic query in the knowledge graph is triggered;

[0115] S4.3: After a multi-semantic query on the knowledge graph, the system determines whether the semantic element is a polysemous term. If so, it displays polysemous options and related cases. If not, it directly sorts the search results by the knowledge base dimensions.

[0116] Furthermore, the specific steps of S4.3 include:

[0117] Analyze the results of multi-semantic queries on the knowledge graph to determine whether the semantic element actually has two or more semantic meanings, that is, whether it is a polysemous term;

[0118] If it is a polysemous term, relevant cases related to each semantic option are extracted from the IT service knowledge base, and the polysemous options and related cases are presented to the user in a visual interface;

[0119] If it is not a polysemous term, the search results will be comprehensively sorted according to the three dimensions of semantic matching, case reuse rate and timeliness to generate a sorted search result list.

[0120] Furthermore, the search results are comprehensively ranked based on the three dimensions of semantic matching, case reuse rate, and timeliness, including:

[0121] (1) Assign weights to the three dimensions of semantic matching, case reuse rate, and timeliness. The weight values ​​are set according to the characteristics of the IT service scenario and user needs.

[0122] (2) Calculate the scores of each search result in terms of semantic matching, case reuse rate, and timeliness;

[0123] (3) Calculate the comprehensive score of each search result based on the assigned weights and the scores of each search result in the three dimensions;

[0124] (4) Sort the search results in descending order of comprehensive scores.

[0125] S4.4: Return search results with semantic labels, collect user feedback data, and return the feedback data to the IT service knowledge graph.

[0126] Furthermore, returning the search results with semantic tags includes:

[0127] (1) Perform semantic analysis on each knowledge node in the sorted search results to determine its semantic category;

[0128] (2) Add corresponding semantic labels to each knowledge node. The semantic labels can include information such as semantic category and semantic source;

[0129] (3) Return the search results with semantic tags to the user in the form of a visual interface or data interface.

[0130] Furthermore, collecting user feedback data includes:

[0131] (1) Design a user feedback collection mechanism, including but not limited to user satisfaction surveys, analysis of user click behavior on search results, and analysis of user selection behavior on polysemous options;

[0132] (2) When users use the search results, user feedback information is collected in real time, including the user's satisfaction score with the search results, the number of clicks on the search results, and the user's selection of polysemous options.

[0133] Furthermore, returning the feedback data to the IT service knowledge graph includes:

[0134] (1) Design knowledge graph quality evaluation indicators, including accuracy, completeness, consistency, and timeliness;

[0135] (2) Using the designed evaluation indicators, evaluate the updated and optimized IT service knowledge graph and calculate the scores of each indicator;

[0136] (3) Based on the evaluation results, determine whether the updated and optimized IT service knowledge graph meets the quality requirements; if not, further analyze the reasons and update and optimize the IT service knowledge graph again.

[0137] The method first locates core knowledge in the IT service knowledge graph based on high-priority semantic information areas, and then improves and supplements the search results by combining low-priority semantic information areas, including:

[0138] S4.1.1: Perform combination and association analysis on semantic elements in high-priority semantic information regions to construct semantic query patterns;

[0139] S4.1.2: Utilize the constructed semantic query pattern to quickly search the IT service knowledge graph and use a graph traversal algorithm to find the core knowledge nodes that match the semantic query pattern. The graph traversal algorithm is prior art in this field and does not constitute an inventive solution of this application, so it will not be described in detail here.

[0140] S4.1.3: Filter and sort the found core knowledge nodes, calculate the comprehensive score of each core knowledge node based on the three dimensions of semantic matching, case reuse rate, and timeliness, and select the core knowledge node with the highest comprehensive score as the quickly located core knowledge;

[0141] S4.1.4: Analyze the association between the semantic elements in the low-priority semantic information area and the quickly located core knowledge to identify information that is related to the core knowledge but not fully covered by it.

[0142] S4.1.5: Based on the results of association analysis, extract low-priority semantic information area knowledge from the IT service knowledge graph as supplementary knowledge;

[0143] S4.1.6: Integrate supplementary knowledge with quickly located core knowledge to form complete search results.

[0144] If the identified semantic element has two semantics in the IT service knowledge graph, a knowledge graph multi-semantic query is triggered, including:

[0145] S4.2.1: In the process of identifying the semantic meaning of a semantic element, use a semantic disambiguation algorithm to determine whether the semantic element has two or more semantic meanings;

[0146] S4.2.2: If there are two semantics, construct different semantic query patterns according to the two corresponding semantics;

[0147] S4.2.3: Use the two constructed semantic query modes to perform parallel queries in the IT service knowledge graph to obtain two types of semantic knowledge.

[0148] The method of using a semantic disambiguation algorithm to determine whether a semantic element has two or more possible semantics includes: first extracting features from the context in which the semantic element is located, and then determining whether the semantic element has two or more possible semantics based on the extracted features.

[0149] Example 2:

[0150] See also Figure 3 Another embodiment of the present invention provides a knowledge base management system, comprising:

[0151] User query module, service scenario determination module, area division module, and semantic understanding module;

[0152] The user query module is used to obtain user query information, pre-process the user query information, and extract semantic elements to provide a basis for subsequent knowledge retrieval and semantic understanding;

[0153] The service scenario determination module is used to determine the IT service scenarios of all semantic elements based on the extracted semantic elements and the pre-built IT service knowledge graph, and to sort and filter the IT service scenarios;

[0154] a region division module, configured to divide user query information into a high-priority semantic information region and a low-priority semantic information region according to the determined IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario;

[0155] The semantic understanding module is used to perform IT service knowledge retrieval and semantic understanding of user query information based on high-priority semantic information areas and low-priority semantic information areas, return retrieval results with semantic labels, and collect user usage feedback data.

[0156] The user query module includes: a query access unit, a text processing unit, a voice processing unit, and a graphic processing unit;

[0157] The query access unit is used to provide multiple user query access methods, including text input box, voice input interface, and graphical query interface, so that users can submit queries in different ways;

[0158] The text processing unit, when a user enters a query through the text input box, uses word segmentation, part-of-speech tagging, and named entity recognition algorithms in natural language processing technology to extract multiple semantic elements that the user intends to express from the input text;

[0159] The voice processing unit converts the voice signal into text when the user enters a query through the voice input interface, and then uses the same processing method of the text processing unit to extract semantic elements;

[0160] The graphical processing unit, when a user enters a query through the graphical query interface, parses the multiple semantic elements that the user intends to express through the interaction and association of interface elements.

[0161] The service scenario determination module includes: a graph processing unit, a scenario matching unit, and a scenario sorting unit;

[0162] The graph processing unit is used to pre-process the pre-built IT service knowledge graph and establish a mapping relationship index between semantic elements and IT service scenarios for fast matching;

[0163] The scenario matching unit is used to match the extracted multiple semantic elements with the mapping relationship index to find all IT service scenarios that cover the semantic elements;

[0164] The scenario ranking unit is used to sort the found IT service scenarios according to the association strength between semantic elements and the commonality of IT service scenarios, and determine the IT service scenarios.

[0165] The regional division module includes: a dimension analysis unit and a regional division unit;

[0166] The dimension analysis unit is used to analyze the importance of different semantic elements in the corresponding IT service scenarios in terms of semantic matching, case reuse rate, and timeliness.

[0167] The region division unit is used to divide the semantic element into a high-priority semantic information region and a low-priority semantic information region.

[0168] The semantic understanding module includes: knowledge positioning unit, result improvement unit, multi-semantic query unit, and feedback collection unit;

[0169] The knowledge location unit is used to combine and analyze the semantic elements in high-priority semantic information areas, build semantic query patterns, and quickly locate core knowledge;

[0170] The result improvement unit is used to analyze the association between the semantic elements in the low-priority semantic information area and the core knowledge quickly located, and extract the knowledge related to the low-priority semantic information area from the IT service knowledge graph as supplementary knowledge to form a complete search result;

[0171] A multi-semantic query unit is used to determine whether a semantic element has two or more possible semantics by using a semantic disambiguation algorithm in the process of identifying the semantics of the semantic element;

[0172] The feedback collection unit is used to return search results with semantic tags, collect user feedback data, and return the feedback data to the IT service knowledge graph.

[0173] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific embodiments. The above-mentioned specific embodiments are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention. These are all protected by the present invention.

Claims

1. A knowledge base enhancement method, characterized in that: include: Obtaining user query information; the user query information includes multiple semantic elements expressing the user's intention; Determine IT service scenarios for all semantic elements based on the semantic elements and the pre-built IT service knowledge graph; Dividing user query information into a high-priority semantic information area and a low-priority semantic information area based on the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario; the knowledge base dimension includes a semantic matching degree dimension, a case reuse rate dimension, and a timeliness dimension; performing IT service knowledge retrieval and semantic understanding on the user query information according to the high-priority semantic information region and the low-priority semantic information region; The performing IT service knowledge retrieval and semantic understanding on the user query information based on the high-priority semantic information area and the low-priority semantic information area includes: First, locate the core knowledge in the IT service knowledge graph based on the high-priority semantic information areas, and then improve and supplement the search results by combining the low-priority semantic information areas; If the identified semantic element has two semantics in the IT service knowledge graph, a multi-semantic query of the knowledge graph is triggered; After a multi-semantic query on the knowledge graph, the system determines whether the semantic element is a polysemous term. If it is a polysemous term, the system displays polysemous options and related cases. If it is not a polysemous term, the system sorts the search results by the knowledge base dimension. Return search results with semantic tags, collect user feedback data, and return the feedback data to the IT service knowledge graph; The method first locates core knowledge in the IT service knowledge graph based on high-priority semantic information areas, and then improves and supplements the search results by combining low-priority semantic information areas, including: Combining and associating semantic elements in high-priority semantic information areas to construct semantic query patterns; Using the constructed semantic query pattern, search the IT service knowledge graph and find the core knowledge nodes that match the semantic query pattern through the graph traversal algorithm; Filter and sort the found core knowledge nodes, calculate the comprehensive score of each core knowledge node based on the three dimensions of semantic matching, case reuse rate, and timeliness, and select the core knowledge node with the highest comprehensive score as the located core knowledge; Conduct correlation analysis between the semantic elements in the low-priority semantic information area and the located core knowledge to find out the information that is not fully covered by the core knowledge; According to the results of association analysis, low-priority semantic information area knowledge is extracted from the IT service knowledge graph as supplementary knowledge; The supplementary knowledge is integrated with the located core knowledge to form a complete search result.

2. A knowledge base enhancement method according to claim 1, characterized in that: Determining IT service scenarios for all semantic elements based on the semantic elements and the pre-built IT service knowledge graph includes: Preprocess the pre-built IT service knowledge graph and establish an index of the mapping relationship between semantic elements and IT service scenarios; Match the extracted semantic elements with the mapping relationship index to find all IT service scenarios that cover the semantic elements. If there are multiple IT service scenarios covering the semantic elements, the searched IT service scenarios are sorted according to the strength of association between the semantic elements and the commonality of the IT service scenarios, and the IT service scenario that best meets the user's intention expression is selected as the final IT service scenario.

3. A knowledge base enhancement method according to claim 2, characterized in that: The dividing the user query information into a high-priority semantic information area and a low-priority semantic information area according to the IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario includes: For a specific IT service scenario, analyze the importance of different semantic elements in terms of semantic matching, case reuse rate, and timeliness; Calculate the semantic similarity between each semantic element and the knowledge in the IT service knowledge graph, and classify the semantic elements with semantic similarity higher than the preset semantic similarity threshold into high-priority semantic information areas; Count the frequency and reuse of each semantic element in historical IT service scenarios, and classify semantic elements with frequencies and reuse times exceeding preset frequency and reuse thresholds as high-priority semantic information areas; Obtaining the update time of the information involved in each semantic element, and classifying the semantic elements whose update time and current time difference is less than a preset time difference threshold as high-priority semantic information areas; Semantic elements that are not classified as high-priority semantic information areas are classified as low-priority semantic information areas.

4. A knowledge base enhancement method according to claim 3, characterized in that: If the identified semantic element has two semantics in the IT service knowledge graph, a knowledge graph multi-semantic query is triggered, including: In the process of identifying the semantics of semantic elements, a semantic disambiguation algorithm is used to determine whether the semantic elements have two or more semantics; If there are two semantics, different semantic query patterns are constructed according to the corresponding two semantics; By using the two constructed semantic query modes, parallel queries are performed in the IT service knowledge graph to obtain two types of semantic knowledge.

5. A knowledge base enhancement method according to claim 4, characterized in that: The method of using a semantic disambiguation algorithm to determine whether a semantic element has two or more semantics includes: first extracting features from the context in which the semantic element is located, and then determining whether the semantic element has two or more semantics based on the extracted features.

6. A knowledge base enhancement method according to claim 1, characterized in that: The obtaining of user query information specifically includes: Design user query access methods, including text input box, voice input interface, and graphical query interface; When a user enters a query in the text input box, the word segmentation, part-of-speech tagging, and named entity recognition algorithms in natural language processing technology are used to extract multiple semantic elements that represent the user's intention from the input text; When a user enters a query through the voice input interface, the voice signal is first converted into text. Then, using natural language processing technologies such as word segmentation, part-of-speech tagging, and named entity recognition algorithms, multiple semantic elements that represent the user's intent are extracted from the converted text. When a user enters a query through the graphical query interface, multiple semantic elements expressing the user's intention are parsed through the interaction and association relationships of the interface elements.

7. A knowledge base enhancement method according to claim 1, characterized in that: The construction process of the pre-built IT service knowledge graph includes: Define entities, relationships, and attributes specific to the IT field and establish a knowledge model for the IT field; Collect existing data in the IT service knowledge base, including structured data, semi-structured data, and unstructured data; Preprocess the collected data, including data cleaning, data conversion and data integration; Using knowledge extraction technology, entities, relationships, and attributes are extracted from pre-processed data and organized according to the knowledge model of the IT field; Store the organized knowledge in the knowledge graph database to complete the construction of the IT service knowledge graph.

8. A knowledge base management system, used to implement a knowledge base enhancement method according to any one of claims 1 to 7, characterized in that: include: User query module, service scenario determination module, area division module, and semantic understanding module; The user query module is used to obtain user query information, pre-process the user query information, and extract semantic elements; The service scenario determination module is used to determine the IT service scenarios of all semantic elements based on the extracted semantic elements and the pre-built IT service knowledge graph, and to sort and filter the IT service scenarios; The region division module is configured to divide the user query information into a high-priority semantic information region and a low-priority semantic information region according to the determined IT service scenario and at least one knowledge base dimension corresponding to the IT service scenario; The semantic understanding module is used to perform IT service knowledge retrieval and semantic understanding on user query information based on high-priority semantic information areas and low-priority semantic information areas, return retrieval results with semantic tags, and collect user usage feedback data.

Images