AI-based enterprise knowledge base intelligent management system and method

Abstract

The invention provides an AI-based enterprise knowledge base intelligent management system and method, and the system comprises a data access and preprocessing module which is used for carrying out the real-time data connection of a plurality of data sources in an enterprise, obtaining different knowledge data, and carrying out the Chinese entity recognition of the knowledge data; the data processing module is used for performing knowledge association mining analysis on the knowledge data based on the Chinese entity recognition result, constructing a dynamic knowledge graph, and updating the dynamic knowledge graph based on the change characteristics of the knowledge data in different data sources and the aging verification result of the knowledge data; and the interaction management module is used for analyzing the query data submitted by the user, matching related knowledge data from the enterprise knowledge base according to an analysis result and the dynamic knowledge graph, and feeding back the related knowledge data to the user terminal. The efficiency and accuracy of user query are improved, and the management effect of the enterprise knowledge base is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of enterprise knowledge management technology, and in particular to an AI-based intelligent management system and method for enterprise knowledge bases. Background Technology

[0002] Currently, with the development of enterprises and the expansion of their businesses, the knowledge and information accumulated within enterprises are growing exponentially. Therefore, effective management of enterprise knowledge is particularly important.

[0003] However, traditional knowledge bases rely on manual classification, resulting in low retrieval efficiency and delayed updates. Furthermore, the classification of knowledge is often inaccurate, leading to low efficiency for employees when searching for the knowledge they need, and the functions rely on preset rules. In addition, traditional knowledge bases are not updated in a timely manner, leaving some outdated information in the knowledge base, which can mislead employees' decisions. At the same time, they lack the ability to deeply analyze knowledge content and cannot uncover the relationships between knowledge, which in turn makes it difficult for employees to quickly and comprehensively retrieve the data they need, greatly reducing the management effectiveness of the enterprise knowledge base.

[0004] Therefore, in order to overcome the above-mentioned shortcomings, the present invention provides an AI-based intelligent management system and method for enterprise knowledge base. Summary of the Invention

[0005] This invention provides an AI-based intelligent management system and method for enterprise knowledge bases. It connects knowledge data from different data sources within the enterprise and performs Chinese entity recognition on the connected knowledge data. This facilitates the determination of relationships between different knowledge data based on the Chinese entity recognition results, enabling knowledge association mining and analysis, and constructing a dynamic knowledge graph. This ensures the rigor and reliability of the dynamic knowledge graph. Furthermore, it updates the dynamic knowledge graph based on the changing characteristics of knowledge data from different data sources and the timeliness verification results of the knowledge data, ensuring accurate and reliable query criteria for users and improving the timeliness of knowledge updates in the knowledge base. Finally, it responds to user-submitted queries and, combined with the dynamic knowledge graph, matches relevant knowledge data from the enterprise knowledge base, improving the efficiency and accuracy of user queries and significantly enhancing the management effectiveness of the enterprise knowledge base.

[0006] This invention provides an AI-based intelligent management system for enterprise knowledge bases, comprising:

[0007] The data access and preprocessing module is used to connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and to perform Chinese entity recognition on the knowledge data.

[0008] The data processing module is used to perform knowledge association mining and analysis on knowledge data based on Chinese entity recognition results, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the changing characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data.

[0009] The interaction management module is used to parse the query data submitted by users and, based on the parsing results and dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base and feed it back to the user terminal.

[0010] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a data access and preprocessing module, comprising:

[0011] The interface allocation unit is used to identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways.

[0012] The data connection unit connects with each data source based on the allocated data flow interface, and configures the data flow interface for data throughput based on the business data generation volume of each data source based on the connection results.

[0013] The data receiving and preprocessing unit is used to receive business data generated by different data sources in real time based on the data throughput configuration results, obtain different knowledge data, and perform preprocessing operations on the different knowledge data.

[0014] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a data receiving and preprocessing unit, comprising:

[0015] The data area is a molecular unit used to distinguish different knowledge data based on the data source, and to classify and cache different knowledge data based on the distinction results;

[0016] The data preprocessing subunit is used for:

[0017] Based on the classification results, the corresponding benchmark business indicators of the knowledge data are obtained from the business side of the corresponding category, and the knowledge data of the corresponding category is processed for business standardization based on the benchmark business indicators.

[0018] Based on the processing results of business specifications, data cleaning and deduplication are performed on each category of business data to complete the preprocessing operations for different knowledge data.

[0019] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a data access and preprocessing module, comprising:

[0020] The data identification unit is used to identify the data types of different knowledge data and obtain the data modal corresponding to different knowledge data.

[0021] Chinese entity recognition unit, used for:

[0022] Based on the data modality, knowledge data of the same category are sorted into queues, and the Chinese entity recognition strategy corresponding to each category is retrieved based on the queue sorting results.

[0023] Based on the Chinese entity recognition strategy, Chinese entity recognition is performed on the knowledge data after each category queue is sorted in parallel, and the Chinese entities obtained after recognition are originally associated with the corresponding knowledge data.

[0024] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a data processing module comprising:

[0025] The association mining and analysis unit is used for:

[0026] Obtain Chinese entity recognition results for different knowledge data, and classify the different knowledge data for business purposes based on the semantics of Chinese entities;

[0027] Based on the business classification results, the semantics of Chinese entities are transformed into semantic vectors. At the same time, the pre-labeled entities and relations are used as seed data to learn the semantic vectors, and the first association relationship between different Chinese entities is obtained based on the learning results.

[0028] The first association is mapped across different knowledge data to obtain the second association between different knowledge data.

[0029] The graph construction unit is used for:

[0030] Based on the second association relationship, the nodes and edges of the knowledge graph are determined for knowledge data in different modalities. Based on the determined nodes and edges, the graph structure corresponding to the knowledge data in different modalities is constructed. In the graph modal data, the nodes are objects in the image and the edges are the association relationships between objects. In the text modal data, the nodes are words and the edges are the grammatical relationships between words. In the speech modal data, the nodes are speech elements and the edges are the combination relationships between speech elements.

[0031] Based on the graph neural network, the graph structure corresponding to knowledge data under different modalities is jointly learned according to the second association relationship, and the joint learning is monitored in real time based on the limited loss function under different modalities;

[0032] When the preset requirements are met based on real-time monitoring results, the nodes in different graph structures are associated according to the joint learning results;

[0033] The key concerns during the construction of the dynamic knowledge graph are obtained from the management terminal, and the preference parameters of the attention mechanism are adapted based on the key concerns.

[0034] The distance in the node association results in different graph structures is adaptively corrected based on the preference parameter adaptation results, and a dynamic knowledge graph is obtained based on the adaptive correction results.

[0035] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a graph construction unit comprising:

[0036] The map sampling subunit is used for:

[0037] Randomly lock nodes in the dynamic knowledge graph, and determine the data parameters that are associated with the locked nodes based on the locking results and the dynamic knowledge graph.

[0038] The corresponding sampling data set is retrieved from the enterprise knowledge base based on the data parameters;

[0039] The quality assessment subunit is used for:

[0040] Perform quality checks on the correlation of the sampled data set, and determine the actual correlation between each sampled data in the sampled data set based on the quality check results;

[0041] The actual relationships are compared with the representational relationships of the dynamic knowledge graph, and the quality of the dynamic knowledge graph is evaluated based on the comparison results.

[0042] The quality inspection of the dynamic knowledge graph was completed based on the quality assessment results.

[0043] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes a data processing module comprising:

[0044] The data monitoring unit is used for:

[0045] Real-time monitoring of the upload dynamics of knowledge data from different data sources, and comparison of the differences in upload dynamics at adjacent times based on the upload time series of knowledge data from different data sources;

[0046] Based on the difference comparison, the change characteristics of knowledge data in different data sources are determined, and when there is updated knowledge data, the updated knowledge data is used as the first update parameter;

[0047] The validity period verification unit is used to periodically traverse the validity period of different knowledge data in the enterprise knowledge base, determine the expired knowledge data based on the periodic traversal results, and use the expired knowledge data as the second update parameter.

[0048] The map update unit is used for:

[0049] Determine the associated nodes of the first update parameter on the dynamic knowledge graph, and expand the branch structure of the dynamic knowledge graph based on the relationship between the first update parameter and the associated nodes;

[0050] Simultaneously, the target node set and edge set of the second update parameter are determined on the dynamic knowledge graph, and the target node set and edge set are pruned on the dynamic knowledge graph to complete the update of the dynamic knowledge graph.

[0051] Preferably, an AI-based intelligent management system for enterprise knowledge bases includes an interactive management module, comprising:

[0052] The query parsing unit is used to obtain the query data submitted by the user, and to perform natural language parsing on the query data to determine the target semantics corresponding to the query data;

[0053] Data matching unit, used for:

[0054] A global scan of the dynamic knowledge graph is performed based on the target semantics, and the target node corresponding to the query data is determined based on the global scan results.

[0055] Based on the topological structure of the target node in the dynamic knowledge graph, the relevant data corresponding to the query data is determined, and the relevant data is filtered based on the association level index to obtain the relevant knowledge data.

[0056] The data feedback unit is used to package and compress relevant knowledge data and then send the packaged and compressed results back to the user terminal.

[0057] This invention provides an AI-based intelligent management method for enterprise knowledge bases, comprising:

[0058] Step 1: Connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and perform Chinese entity recognition on the knowledge data;

[0059] Step 2: Based on the Chinese entity recognition results, perform knowledge association mining and analysis on the knowledge data, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the change characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data;

[0060] Step 3: Parse the query data submitted by the user, and match relevant knowledge data from the enterprise knowledge base based on the parsing results and dynamic knowledge graph, and feed it back to the user terminal.

[0061] Preferably, in an AI-based intelligent management method for enterprise knowledge bases, step 1 involves real-time data integration with multiple data sources within the enterprise to obtain different knowledge data, including:

[0062] Identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways;

[0063] The allocated data flow interface is connected to each data source, and the data flow interface is configured to handle data throughput based on the business data generation volume of each data source according to the connection results.

[0064] Based on the data throughput configuration results, business data generated from different data sources are received in real time to obtain different knowledge data, and preprocessing operations are performed on the different knowledge data.

[0065] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0066] By connecting knowledge data from different data sources within the enterprise and performing Chinese entity recognition on the connected knowledge data, the system facilitates the determination of relationships between different knowledge data based on the Chinese entity recognition results. This enables knowledge association mining and analysis, and the construction of a dynamic knowledge graph. This ensures the rigor and reliability of the dynamic knowledge graph. Furthermore, the system can update the dynamic knowledge graph based on the changing characteristics of knowledge data from different data sources and the timeliness verification results of the knowledge data. This ensures that users have accurate and reliable query criteria when conducting knowledge searches and improves the timeliness of knowledge updates in the knowledge base. Finally, the system can respond to user-submitted queries and, in conjunction with the dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base, improving the efficiency and accuracy of user queries and significantly enhancing the management effectiveness of the enterprise knowledge base.

[0067] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0068] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0069] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0070] Figure 1 This is a structural diagram of an AI-based intelligent management system for enterprise knowledge base, as described in an embodiment of the present invention.

[0071] Figure 2 This is a structural diagram of the data access and preprocessing module in an AI-based intelligent management system for enterprise knowledge bases, as described in an embodiment of the present invention.

[0072] Figure 3 This is a flowchart of an AI-based intelligent management method for enterprise knowledge base in an embodiment of the present invention. Detailed Implementation

[0073] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0074] Example 1:

[0075] This embodiment provides an AI-based intelligent management system for enterprise knowledge bases, such as... Figure 1 As shown, it includes:

[0076] The data access and preprocessing module is used to connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and to perform Chinese entity recognition on the knowledge data.

[0077] The data processing module is used to perform knowledge association mining and analysis on knowledge data based on Chinese entity recognition results, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the changing characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data.

[0078] The interaction management module is used to parse the query data submitted by users and, based on the parsing results and dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base and feed it back to the user terminal.

[0079] In this embodiment, real-time data connection to multiple data sources within the enterprise refers to accessing data streams from business systems such as ERP / CRM through an API gateway.

[0080] In this embodiment, Chinese entity recognition includes text parsing, image parsing, and speech-to-text conversion.

[0081] In this embodiment, the dynamic knowledge graph is used to represent the relationships between different knowledge data.

[0082] In this embodiment, the change characteristics of knowledge data in different data sources refer to the need to update the dynamic knowledge graph when new knowledge data is generated in a data source.

[0083] In this embodiment, the timeliness verification result of knowledge data refers to updating the dynamic knowledge graph according to the validity period of the knowledge data. That is, when data in the enterprise knowledge base becomes invalid, the dynamic knowledge graph needs to be updated while the invalid data is removed.

[0084] In this embodiment, parsing the query data submitted by the user can be done by parsing the query data using natural language processing technology, thereby determining the semantics of the query data submitted by the user, and thus determining the user's query purpose.

[0085] The beneficial effects of the above technical solution are as follows: By connecting knowledge data generated from different data sources within the enterprise and performing Chinese entity recognition on the connected knowledge data, it is easy to determine the correlation between different knowledge data based on the Chinese entity recognition results. This enables knowledge association mining and analysis of the knowledge data, constructing a dynamic knowledge graph, ensuring the rigor and reliability of the dynamic knowledge graph. Simultaneously, the dynamic knowledge graph can be updated based on the changing characteristics of knowledge data from different data sources and the timeliness verification results of the knowledge data, ensuring that users can have accurate and reliable query basis when conducting knowledge queries, and improving the timeliness of knowledge updates in the knowledge base. Finally, it can respond to user-submitted query data and, combined with the dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base, improving the efficiency and accuracy of user queries and greatly enhancing the management effect of the enterprise knowledge base.

[0086] Example 2:

[0087] Based on Example 1, this example provides an AI-based intelligent management system for enterprise knowledge bases, such as... Figure 2 As shown, the data access and preprocessing module includes:

[0088] The interface allocation unit is used to identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways.

[0089] The data connection unit connects with each data source based on the allocated data flow interface, and configures the data flow interface for data throughput based on the business data generation volume of each data source based on the connection results.

[0090] The data receiving and preprocessing unit is used to receive business data generated by different data sources in real time based on the data throughput configuration results, obtain different knowledge data, and perform preprocessing operations on the different knowledge data.

[0091] In this embodiment, business attributes refer to the types of business corresponding to different data sources, such as email data.

[0092] In this embodiment, data throughput configuration refers to configuring the data throughput of the data flow interface, that is, the amount of data passing through per unit time.

[0093] In this embodiment, the preprocessing operation refers to data cleaning and deduplication operations on different knowledge data.

[0094] The beneficial effects of the above technical solution are: by connecting multiple data sources within the enterprise, the data flow interface can be configured according to the connection results, thereby enabling the receipt of business data generated by different data sources, thus ensuring the comprehensive and effective acquisition of different knowledge data, and providing reliable data support for enterprise knowledge base management.

[0095] Example 3:

[0096] Based on Example 2, this example provides an AI-based intelligent management system for enterprise knowledge bases, including a data receiving and preprocessing unit:

[0097] The data area is a molecular unit used to distinguish different knowledge data based on the data source, and to classify and cache different knowledge data based on the distinction results;

[0098] The data preprocessing subunit is used for:

[0099] Based on the classification results, the corresponding benchmark business indicators of the knowledge data are obtained from the business side of the corresponding category, and the knowledge data of the corresponding category is processed for business standardization based on the benchmark business indicators.

[0100] Based on the processing results of business specifications, data cleaning and deduplication are performed on each category of business data to complete the preprocessing operations for different knowledge data.

[0101] In this embodiment, the benchmark business indicator refers to the business standard corresponding to different knowledge data, such as the data value range of the industry in which different knowledge data are located.

[0102] The beneficial effects of the above technical solution are: by distinguishing and classifying the received different knowledge data and caching them, it is convenient to perform corresponding business specification processing, data cleaning and deduplication operations on different types of knowledge data, thus ensuring the accuracy and reliability of the final knowledge data.

[0103] Example 4:

[0104] Based on Example 1, this example provides an AI-based intelligent management system for enterprise knowledge bases, including a data access and preprocessing module:

[0105] The data identification unit is used to identify the data types of different knowledge data and obtain the data modal corresponding to different knowledge data.

[0106] Chinese entity recognition unit, used for:

[0107] Based on the data modality, knowledge data of the same category are sorted into queues, and the Chinese entity recognition strategy corresponding to each category is retrieved based on the queue sorting results.

[0108] Based on the Chinese entity recognition strategy, Chinese entity recognition is performed on the knowledge data after each category queue is sorted in parallel, and the Chinese entities obtained after recognition are originally associated with the corresponding knowledge data.

[0109] In this embodiment, the data category status includes text category, image category, and voice category.

[0110] In this embodiment, data modality refers to the type of knowledge data, such as image modality, text modality, and voice modality.

[0111] In this embodiment, the Chinese entity recognition strategy is pre-set. Different data modalities correspond to different Chinese entity recognition strategies, which can be corresponding algorithms, etc.

[0112] The beneficial effects of the above technical solution are: by determining the data modal corresponding to different knowledge data, the corresponding Chinese entity recognition strategy can be called according to the data modal to perform Chinese entity recognition, which ensures the accuracy and reliability of knowledge data recognition for different data modalities and provides a reliable guarantee for the construction of dynamic knowledge graphs.

[0113] Example 5:

[0114] Based on Example 1, this example provides an AI-based intelligent management system for enterprise knowledge bases, including a data processing module:

[0115] The association mining and analysis unit is used for:

[0116] Obtain Chinese entity recognition results for different knowledge data, and classify the different knowledge data for business purposes based on the semantics of Chinese entities;

[0117] Based on the business classification results, the semantics of Chinese entities are transformed into semantic vectors. At the same time, the pre-labeled entities and relations are used as seed data to learn the semantic vectors, and the first association relationship between different Chinese entities is obtained based on the learning results.

[0118] The first association is mapped across different knowledge data to obtain the second association between different knowledge data.

[0119] The graph construction unit is used for:

[0120] Based on the second association relationship, the nodes and edges of the knowledge graph are determined for knowledge data in different modalities. Based on the determined nodes and edges, the graph structure corresponding to the knowledge data in different modalities is constructed. In the graph modal data, the nodes are objects in the image and the edges are the association relationships between objects. In the text modal data, the nodes are words and the edges are the grammatical relationships between words. In the speech modal data, the nodes are speech elements and the edges are the combination relationships between speech elements.

[0121] Based on the graph neural network, the graph structure corresponding to knowledge data under different modalities is jointly learned according to the second association relationship, and the joint learning is monitored in real time based on the limited loss function under different modalities;

[0122] When the preset requirements are met based on real-time monitoring results, the nodes in different graph structures are associated according to the joint learning results;

[0123] The key concerns during the construction of the dynamic knowledge graph are obtained from the management terminal, and the preference parameters of the attention mechanism are adapted based on the key concerns.

[0124] The distance in the node association results in different graph structures is adaptively corrected based on the preference parameter adaptation results, and a dynamic knowledge graph is obtained based on the adaptive correction results.

[0125] In this embodiment, semantic vector refers to the result of converting the semantics of Chinese entities into vector form.

[0126] In this embodiment, the entities and relationships pre-labeled are known in advance and are used to characterize the association relationships between known entities.

[0127] In this embodiment, seed data refers to Chinese entity data that has already been labeled with association relationships.

[0128] In this embodiment, the first association relationship refers to the association relationship between different Chinese entities. It is the association relationship between entities with pre-defined annotations and serves as a reference for constructing a knowledge graph.

[0129] In this embodiment, the second association relationship refers to the association relationship between different knowledge data obtained after learning and analyzing different knowledge data through the first association relationship.

[0130] In this embodiment, the graph structure is constructed based on the nodes and edges corresponding to different knowledge data.

[0131] In this embodiment, the graph neural network is known in advance.

[0132] In this embodiment, the defined loss function is pre-set and is a tool used to monitor the joint learning process, to evaluate the learning loss during the joint learning process, and thus to determine whether the joint learning standard has been met.

[0133] In this embodiment, the preset requirements are set in advance.

[0134] In this embodiment, the key focus refers to the knowledge data or structure that needs to be emphasized when constructing a dynamic knowledge graph.

[0135] In this embodiment, preference parameter adaptation refers to configuring the parameters of the attention mechanism according to key concerns, so as to ensure that the data or parameters of concern can be effectively analyzed when constructing a dynamic knowledge graph.

[0136] In this embodiment, distance adaptive correction refers to adjusting the distance based on the strength of the association between different nodes. The stronger the association, the shorter the distance between the two nodes, and vice versa.

[0137] The beneficial effects of the above technical solution are as follows: First, by classifying different knowledge data according to the semantics of Chinese entities, and determining the association relationship of Chinese entities under each category based on the semantics, then determining the nodes and edges of the knowledge graph for knowledge data under different modalities based on the determined association relationship, thereby realizing the accurate and effective construction of the graph structure corresponding to the knowledge data under different modalities based on the nodes and edges. Finally, by jointly learning different graph structures, a comprehensive and reliable construction of dynamic knowledge graph is achieved, thereby ensuring the accuracy and reliability of intelligent management of enterprise knowledge base.

[0138] Example 6:

[0139] Based on Example 5, this example provides an AI-based intelligent management system for enterprise knowledge bases, including a graph construction unit comprising:

[0140] The map sampling subunit is used for:

[0141] Randomly lock nodes in the dynamic knowledge graph, and determine the data parameters that are associated with the locked nodes based on the locking results and the dynamic knowledge graph.

[0142] The corresponding sampling data set is retrieved from the enterprise knowledge base based on the data parameters;

[0143] The quality assessment subunit is used for:

[0144] Perform quality checks on the correlation of the sampled data set, and determine the actual correlation between each sampled data in the sampled data set based on the quality check results;

[0145] The actual relationships are compared with the representational relationships of the dynamic knowledge graph, and the quality of the dynamic knowledge graph is evaluated based on the comparison results.

[0146] The quality inspection of the dynamic knowledge graph was completed based on the quality assessment results.

[0147] In this embodiment, the data parameter refers to the set of data that is associated with the locked node, as determined by the dynamic knowledge graph.

[0148] In this embodiment, the sampling data set refers to the specific data retrieved from the enterprise knowledge base according to the determined data parameters.

[0149] The beneficial effects of the above technical solution are: by sampling the dynamic knowledge graph and monitoring its quality based on the sampling results, the reliability of the final dynamic knowledge graph is ensured, which also provides convenience and security for enterprise knowledge management.

[0150] Example 7:

[0151] Based on Example 1, this example provides an AI-based intelligent management system for enterprise knowledge bases, including a data processing module:

[0152] The data monitoring unit is used for:

[0153] Real-time monitoring of the upload dynamics of knowledge data from different data sources, and comparison of the differences in upload dynamics at adjacent times based on the upload time series of knowledge data from different data sources;

[0154] Based on the difference comparison, the change characteristics of knowledge data in different data sources are determined, and when there is updated knowledge data, the updated knowledge data is used as the first update parameter;

[0155] The validity period verification unit is used to periodically traverse the validity period of different knowledge data in the enterprise knowledge base, determine the expired knowledge data based on the periodic traversal results, and use the expired knowledge data as the second update parameter.

[0156] The map update unit is used for:

[0157] Determine the associated nodes of the first update parameter on the dynamic knowledge graph, and expand the branch structure of the dynamic knowledge graph based on the relationship between the first update parameter and the associated nodes;

[0158] Simultaneously, the target node set and edge set of the second update parameter are determined on the dynamic knowledge graph, and the target node set and edge set are pruned on the dynamic knowledge graph to complete the update of the dynamic knowledge graph.

[0159] In this embodiment, the upload time series refers to the upload time information of knowledge data from different data sources.

[0160] In this embodiment, the change feature refers to whether there are differences in knowledge data from different data sources at adjacent times. When there are differences, it is determined to store newly uploaded knowledge data.

[0161] In this embodiment, the first update parameter refers to the newly uploaded knowledge data.

[0162] In this embodiment, expired knowledge data refers to knowledge data in the enterprise knowledge base that has reached its expiration date.

[0163] In this embodiment, the second update parameter refers to the failure knowledge data.

[0164] In this embodiment, the associated node refers to the specific node on the dynamic knowledge graph corresponding to the first update parameter.

[0165] In this embodiment, branch structure expansion refers to expanding the structure of associated nodes according to the first update parameter, so as to represent the relationship between updated data and existing knowledge data in more detail.

[0166] In this embodiment, pruning refers to adjusting the dynamic knowledge graph, that is, removing the nodes corresponding to invalid knowledge data from the dynamic knowledge graph.

[0167] The beneficial effects of the above technical solution are as follows: by determining the change characteristics of knowledge data in different data sources, the updated data can be identified when there is updated knowledge data. At the same time, the validity period of different knowledge data in the enterprise knowledge base can be determined, and the invalid data in the enterprise knowledge base can be identified. Finally, the dynamic knowledge graph is updated by updating the updated data and invalid data, which ensures the correlation between the dynamic knowledge graph and the knowledge data in the enterprise knowledge base and guarantees the intelligent management effect of the enterprise knowledge base.

[0168] Example 8:

[0169] Based on Example 1, this example provides an AI-based intelligent management system for enterprise knowledge bases, including an interactive management module:

[0170] The query parsing unit is used to obtain the query data submitted by the user, and to perform natural language parsing on the query data to determine the target semantics corresponding to the query data;

[0171] Data matching unit, used for:

[0172] A global scan of the dynamic knowledge graph is performed based on the target semantics, and the target node corresponding to the query data is determined based on the global scan results.

[0173] Based on the topological structure of the target node in the dynamic knowledge graph, the relevant data corresponding to the query data is determined, and the relevant data is filtered based on the association level index to obtain the relevant knowledge data.

[0174] The data feedback unit is used to package and compress relevant knowledge data and then send the packaged and compressed results back to the user terminal.

[0175] In this embodiment, natural language parsing refers to determining the semantics of the query data, that is, obtaining the corresponding target semantics (i.e., the specific meaning of the text).

[0176] In this embodiment, the target node refers to the specific node on the dynamic knowledge graph corresponding to the query data.

[0177] In this embodiment, the association level index is pre-set and is used to characterize the association level that needs to be achieved.

[0178] The beneficial effects of the above technical solution are: by parsing the query data submitted by the user, the system can obtain the corresponding knowledge data from the enterprise knowledge base based on the query results and dynamic knowledge graph, and then feed the obtained knowledge data back to the user terminal, thereby improving the efficiency and effectiveness of the user's enterprise knowledge query.

[0179] Example 9:

[0180] This embodiment provides an AI-based intelligent management method for enterprise knowledge bases, such as... Figure 3 As shown, it includes:

[0181] Step 1: Connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and perform Chinese entity recognition on the knowledge data;

[0182] Step 2: Based on the Chinese entity recognition results, perform knowledge association mining and analysis on the knowledge data, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the change characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data;

[0183] Step 3: Parse the query data submitted by the user, and match relevant knowledge data from the enterprise knowledge base based on the parsing results and dynamic knowledge graph, and feed it back to the user terminal.

[0184] The beneficial effects of the above technical solution are as follows: By connecting knowledge data generated from different data sources within the enterprise and performing Chinese entity recognition on the connected knowledge data, it is easy to determine the correlation between different knowledge data based on the Chinese entity recognition results. This enables knowledge association mining and analysis of the knowledge data, constructing a dynamic knowledge graph, ensuring the rigor and reliability of the dynamic knowledge graph. Simultaneously, the dynamic knowledge graph can be updated based on the changing characteristics of knowledge data from different data sources and the timeliness verification results of the knowledge data, ensuring that users can have accurate and reliable query basis when conducting knowledge queries, and improving the timeliness of knowledge updates in the knowledge base. Finally, it can respond to user-submitted query data and, combined with the dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base, improving the efficiency and accuracy of user queries and greatly enhancing the management effect of the enterprise knowledge base.

[0185] Example 10:

[0186] Building upon Example 9, this example provides an AI-based intelligent management method for enterprise knowledge bases. In step 1, real-time data integration with multiple data sources within the enterprise is performed to obtain different knowledge data, including:

[0187] Identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways;

[0188] The allocated data flow interface is connected to each data source, and the data flow interface is configured to handle data throughput based on the business data generation volume of each data source according to the connection results.

[0189] Based on the data throughput configuration results, business data generated from different data sources are received in real time to obtain different knowledge data, and preprocessing operations are performed on the different knowledge data.

[0190] The beneficial effects of the above technical solution are: by connecting multiple data sources within the enterprise, the data flow interface can be configured according to the connection results, thereby enabling the receipt of business data generated by different data sources, thus ensuring the comprehensive and effective acquisition of different knowledge data, and providing reliable data support for enterprise knowledge base management.

[0191] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An AI-based intelligent management system for enterprise knowledge bases, characterized in that, include: The data access and preprocessing module is used to connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and to perform Chinese entity recognition on the knowledge data. The data processing module is used to perform knowledge association mining and analysis on knowledge data based on Chinese entity recognition results, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the changing characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data. The interaction management module is used to parse the query data submitted by users and, based on the parsing results and dynamic knowledge graph, match relevant knowledge data from the enterprise knowledge base and feed it back to the user terminal.

2. The AI-based intelligent management system for enterprise knowledge base according to claim 1, characterized in that, The data access and preprocessing module includes: The interface allocation unit is used to identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways. The data connection unit connects with each data source based on the allocated data flow interface, and configures the data flow interface for data throughput based on the business data generation volume of each data source based on the connection results. The data receiving and preprocessing unit is used to receive business data generated by different data sources in real time based on the data throughput configuration results, obtain different knowledge data, and perform preprocessing operations on the different knowledge data.

3. The AI-based intelligent management system for enterprise knowledge base according to claim 2, characterized in that, The data receiving and preprocessing unit includes: The data area is a molecular unit used to distinguish different knowledge data based on the data source, and to classify and cache different knowledge data based on the distinction results; The data preprocessing subunit is used for: Based on the classification results, the corresponding benchmark business indicators of the knowledge data are obtained from the business side of the corresponding category, and the knowledge data of the corresponding category is processed for business standardization based on the benchmark business indicators. Based on the processing results of business specifications, data cleaning and deduplication are performed on each category of business data to complete the preprocessing operations for different types of knowledge data.

4. The AI-based intelligent management system for enterprise knowledge base according to claim 1, characterized in that, The data access and preprocessing module includes: The data identification unit is used to identify the data types of different knowledge data and obtain the data modal corresponding to different knowledge data. Chinese entity recognition unit, used for: Based on the data modality, knowledge data of the same category are sorted into queues, and the Chinese entity recognition strategy corresponding to each category is retrieved based on the queue sorting results. Based on the Chinese entity recognition strategy, Chinese entity recognition is performed on the knowledge data after each category queue is sorted in parallel, and the Chinese entities obtained after recognition are originally associated with the corresponding knowledge data.

5. The AI-based intelligent management system for enterprise knowledge base according to claim 1, characterized in that, The data processing module includes: The association mining and analysis unit is used for: Obtain Chinese entity recognition results for different knowledge data, and classify the different knowledge data for business purposes based on the semantics of Chinese entities; Based on the business classification results, the semantics of Chinese entities are transformed into semantic vectors. At the same time, the pre-labeled entities and relations are used as seed data to learn the semantic vectors, and the first association relationship between different Chinese entities is obtained based on the learning results. The first association is mapped across different knowledge data to obtain the second association between different knowledge data. The graph construction unit is used for: Based on the second association relationship, the nodes and edges of the knowledge graph are determined for knowledge data in different modalities. Based on the determined nodes and edges, the graph structure corresponding to the knowledge data in different modalities is constructed. In the graph modal data, the nodes are objects in the image and the edges are the association relationships between objects. In the text modal data, the nodes are words and the edges are the grammatical relationships between words. In the speech modal data, the nodes are speech elements and the edges are the combination relationships between speech elements. Based on the graph neural network, the graph structure corresponding to knowledge data under different modalities is jointly learned according to the second association relationship, and the joint learning is monitored in real time based on the limited loss function under different modalities; When the preset requirements are met based on real-time monitoring results, the nodes in different graph structures are associated according to the joint learning results; The key concerns during the construction of the dynamic knowledge graph are obtained from the management terminal, and the preference parameters of the attention mechanism are adapted based on the key concerns. The distance in the node association results in different graph structures is adaptively corrected based on the preference parameter adaptation results, and a dynamic knowledge graph is obtained based on the adaptive correction results.

6. The AI-based intelligent management system for enterprise knowledge base according to claim 5, characterized in that, The map construction unit includes: The map sampling subunit is used for: Randomly lock nodes in the dynamic knowledge graph, and determine the data parameters that are associated with the locked nodes based on the locking results and the dynamic knowledge graph. The corresponding sampling data set is retrieved from the enterprise knowledge base based on the data parameters; The quality assessment subunit is used for: Perform quality checks on the correlation of the sampled data set, and determine the actual correlation between each sampled data in the sampled data set based on the quality check results; The actual relationships are compared with the representational relationships of the dynamic knowledge graph, and the quality of the dynamic knowledge graph is evaluated based on the comparison results. The quality inspection of the dynamic knowledge graph was completed based on the quality assessment results.

7. The AI-based intelligent management system for enterprise knowledge base according to claim 1, characterized in that, The data processing module includes: The data monitoring unit is used for: Real-time monitoring of the upload dynamics of knowledge data from different data sources, and comparison of the differences in upload dynamics at adjacent times based on the upload time series of knowledge data from different data sources; Based on the difference comparison, the change characteristics of knowledge data in different data sources are determined, and when there is updated knowledge data, the updated knowledge data is used as the first update parameter; The validity period verification unit is used to periodically traverse the validity period of different knowledge data in the enterprise knowledge base, determine the expired knowledge data based on the periodic traversal results, and use the expired knowledge data as the second update parameter. The map update unit is used for: Determine the associated nodes of the first update parameter on the dynamic knowledge graph, and expand the branch structure of the dynamic knowledge graph based on the relationship between the first update parameter and the associated nodes; Simultaneously, the target node set and edge set of the second update parameter are determined on the dynamic knowledge graph, and the target node set and edge set are pruned on the dynamic knowledge graph to complete the update of the dynamic knowledge graph.

8. The AI-based intelligent management system for enterprise knowledge base according to claim 1, characterized in that, The interactive management module includes: The query parsing unit is used to obtain the query data submitted by the user, and to perform natural language parsing on the query data to determine the target semantics corresponding to the query data; Data matching unit, used for: A global scan of the dynamic knowledge graph is performed based on the target semantics, and the target node corresponding to the query data is determined based on the global scan results. Based on the topological structure of the target node in the dynamic knowledge graph, the relevant data corresponding to the query data is determined, and the relevant data is filtered based on the association level index to obtain the relevant knowledge data. The data feedback unit is used to package and compress relevant knowledge data and then send the packaged and compressed results back to the user terminal.

9. An AI-based intelligent management method for enterprise knowledge bases, characterized in that, include: Step 1: Connect with multiple data sources within the enterprise in real time to obtain different knowledge data, and perform Chinese entity recognition on the knowledge data; Step 2: Based on the Chinese entity recognition results, perform knowledge association mining and analysis on the knowledge data, construct a dynamic knowledge graph, and update the dynamic knowledge graph based on the change characteristics of knowledge data in different data sources and the timeliness verification results of knowledge data; Step 3: Parse the query data submitted by the user, and match relevant knowledge data from the enterprise knowledge base based on the parsing results and dynamic knowledge graph, and feed it back to the user terminal.

10. The AI-based intelligent management method for enterprise knowledge base according to claim 9, characterized in that, In step 1, real-time data integration is performed with multiple data sources within the enterprise to obtain different knowledge data, including: Identify multiple data sources within the enterprise, build gateways in the main control center based on the business attributes of different data sources, and allocate data flow interfaces for each data source based on the gateways; The allocated data flow interface is connected to each data source, and the data flow interface is configured to handle data throughput based on the business data generation volume of each data source according to the connection results. Based on the data throughput configuration results, business data generated from different data sources are received in real time to obtain different knowledge data, and preprocessing operations are performed on the different knowledge data.

Images