A knowledge platform construction management system and method using AI technology

Abstract

The application discloses a knowledge middle platform construction management system and method applying AI technology, relates to the technical field of information management, and stores knowledge information in a database of an enterprise, constructs a knowledge label, determines the generation utilization rate of corresponding knowledge information according to the knowledge label, constructs a knowledge node model to obtain a knowledge node distribution graph, classifies knowledge nodes based on the knowledge node distribution graph, carries out correlation analysis on the knowledge nodes to obtain the correlation strength between the knowledge nodes, constructs a knowledge node combination correlation chain, analyzes the chain affinity between the knowledge node combination correlation chains, divides the knowledge node combination correlation chains into knowledge node communities, analyzes the information aggregation degree of the knowledge nodes in the knowledge point communities, and judges the confidence state of the knowledge node communities.

Description

Technical Field

[0001] This invention relates to the field of information management technology, specifically to a knowledge platform construction management system and method that utilizes AI technology. Background Technology

[0002] The knowledge platform is the core infrastructure of a unified and intelligent enterprise knowledge system. It is used to aggregate, govern, and integrate fragmented knowledge scattered throughout the enterprise, and through intelligent processing, form knowledge service capabilities that can be easily called and reused in various business scenarios.

[0003] With the accelerated advancement of digital transformation, enterprises have an increasingly urgent need for efficient utilization of knowledge assets. Traditional knowledge management typically relies on manually constructing knowledge classification systems and manually completing knowledge processing, tagging, and storage. It depends on domain experts for cataloging and tagging, resulting in low efficiency in knowledge classification and difficulty in scaling up. Furthermore, in terms of knowledge mining, association, and recommendation, it is difficult to meet the complex knowledge management needs within enterprises, resulting in scattered knowledge storage, weak knowledge correlation, and low sharing rate, which is not conducive to users finding and utilizing knowledge. Summary of the Invention

[0004] The purpose of this invention is to provide a technical solution for building a knowledge platform management system and method using AI technology, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for building and managing a knowledge platform using AI technology, comprising the following steps:

[0007] Retrieve knowledge information stored in the enterprise database and construct knowledge tags;

[0008] The generation and utilization rate of corresponding knowledge information is determined based on the knowledge tags, and a knowledge node model is constructed to obtain a knowledge node distribution map;

[0009] Based on the knowledge node distribution map, knowledge nodes are classified, and association analysis is performed on the knowledge nodes to obtain the association strength between knowledge nodes, so as to construct a knowledge node combination association chain.

[0010] By analyzing the chain affinity between the knowledge node combination association chains, the knowledge node combination association chains are divided into knowledge node communities, and the information aggregation degree of knowledge nodes in the knowledge node communities is analyzed to determine the confidence status of the knowledge node communities.

[0011] Furthermore, the knowledge information is business data stored in the enterprise database; the knowledge tag consists of knowledge information and a corresponding time data set; the time data set includes the storage time t of the knowledge information.c Call time t d The number of calls f and the number of productions p; where the number of productions of knowledge information includes the number of times the same knowledge information is repeatedly generated and updated;

[0012] Based on the knowledge tags in the enterprise database, the generation and utilization rate of knowledge information is analyzed using Ur, and the analysis calculation is as follows:

[0013] ;

[0014] Based on the knowledge information generation and utilization rate (Ur) analysis data, this method constructs a knowledge node model and imports the corresponding knowledge information generation and utilization rate (Ur) to perform node mapping on the knowledge information; it constructs a planar coordinate system with the knowledge information storage time (t) as the reference. c Using the x-axis as the horizontal axis and the generation utilization rate Ur as the y-axis, we determine the knowledge information coordinates to determine the node distribution status of the knowledge information, thereby obtaining a knowledge node distribution map.

[0015] Furthermore, based on the knowledge node distribution map, a horizontal ray is taken along the vertical axis and translated in the vertical direction. The density of knowledge nodes covered on different rays is recorded simultaneously. The vertical axis value of the ray with the maximum density is determined and recorded as the classification threshold Uc. By comparing the generation utilization rate Ur of each knowledge node with the classification threshold Uc, each knowledge node is classified. Since the value on the vertical axis corresponds to the generation utilization rate of the knowledge node, rays are constructed on the vertical axis and translated vertically to obtain the number of knowledge nodes covered by each ray corresponding to each value on the vertical axis. The density of covered knowledge nodes is determined by analyzing the proportion of the number of covered knowledge nodes in the knowledge node distribution map. The maximum value is taken as the classification threshold of the knowledge node, which reflects the dispersion of the utilization distribution of the corresponding knowledge information in the database. The concentrated value is taken as the discrimination object to realize the dynamic identification of the utilization of knowledge information.

[0016] If Ur≥Uc, then the corresponding knowledge node will be classified as a main knowledge node.

[0017] If Ur < Uc, then the corresponding knowledge node will be classified as an edge knowledge node;

[0018] Based on the knowledge node classification results, the main knowledge nodes and peripheral knowledge nodes are divided into two categories, and corresponding knowledge node sets are constructed, denoted as the main node set and the peripheral node set.

[0019] Furthermore, we iterate through the set of backbone nodes and select any non-repeating backbone knowledge nodes i and j for co-occurrence association analysis to obtain the co-occurrence association strength cas(i,j) of the corresponding backbone knowledge nodes i and j. The analysis and calculation are as follows:

[0020] ;

[0021] Where i and j are the numbers of the main knowledge nodes; w(i,j) is the weight of the main knowledge nodes i and j in the same text source; M(i,j) is the number of times the main knowledge nodes i and j appear together in the same text source; M(i) and M(j) are the total number of times the main knowledge nodes i and j appear alone in their respective text sources.

[0022] When the co-occurrence association strength cas(i,j)≥cz for any two main knowledge nodes i and j, main knowledge nodes i and j are determined to be associated main knowledge nodes; the associated main knowledge nodes are then connected to construct a main knowledge node association chain; when cas(i,j)<cz, main knowledge nodes i and j cannot form an association and do not constitute an association chain; where cz is the comparison threshold; where a main knowledge node is simultaneously associated with three other main knowledge nodes, then it will have one main knowledge point existing in multiple main knowledge node association chains.

[0023] Furthermore, based on the association chain of the main knowledge nodes, behavioral association analysis is performed on each main knowledge node in the association chain between the edge knowledge nodes and the main knowledge nodes in the edge node set to obtain the behavioral association strength bas(p,k) between each main knowledge node and the edge knowledge node, which is calculated as follows:

[0024] ;

[0025] Where p is the number of the main knowledge node in the main knowledge node association chain; k is the number of the edge knowledge node in the edge node set; w(p,k) is the number of times the main knowledge node p and the edge knowledge node k are called simultaneously; N(p) and N(k) are the number of times the main knowledge node p and the edge knowledge node k are called individually, respectively.

[0026] When the behavioral association strength bas(p,k) ≥ cz between any main knowledge node p in the main knowledge node association chain and any edge knowledge node k in the edge node set, edge knowledge node k is determined to be an associated edge knowledge node of main knowledge node p. Based on the association chain of main knowledge nodes, each main knowledge node is connected with its corresponding associated edge knowledge node to obtain a knowledge node combination association chain. When bas(p,k) < cz, edge knowledge node k cannot form an association with main knowledge node p and does not constitute an association chain. Among these, there exists a main knowledge node that forms an association with multiple edge knowledge nodes.

[0027] Furthermore, by traversing the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, the chain affinity Ct(y,e) of adjacent combination association chains is analyzed based on the number of overlapping knowledge nodes obtained from the traversal; where y and e are the numbers of adjacent knowledge node combination association chains, respectively; the analysis results are as follows:

[0028] Ct(y,e)=[n(y,e)] / [n(y)+n(e)];

[0029] Wherein, n(y,e) is the number of knowledge nodes that overlap between adjacent knowledge node combination association chains y and e; n(y) and n(e) are the number of knowledge nodes on adjacent knowledge node combination association chains y and e, respectively; where adjacent knowledge node combination association chains are knowledge node combination association chains that have overlapping knowledge nodes on the knowledge node distribution map.

[0030] Furthermore, when the chain affinity Ct(y,e) between adjacent knowledge node combination association chains reaches a preset affinity threshold, the current adjacent knowledge node combination association chains are divided into knowledge node communities. Based on the knowledge node community, the information weight of the connection edge between the corresponding knowledge nodes is determined by obtaining the association strength between any knowledge nodes, and the information aggregation degree Igd analysis is performed on the knowledge nodes within the knowledge node community, which is calculated as follows:

[0031] ;

[0032] Where m is the number of interconnected edges between knowledge nodes within the knowledge node community; x and y are the unique node numbers within the knowledge node community; Q x,y The weight of the concatenation edge information between knowledge nodes x and y; Q x,h and Q y,h γ represents the sum of the weights of the concatenated edges connecting to knowledge nodes x and y, respectively; x:y (α, β) are the knowledge node category judgment parameters; where, the knowledge node category judgment parameters are obtained by judging the categories of knowledge nodes x and y. When x and y are of the same category, the parameter value α is taken, and otherwise the parameter value β is taken; where Qx,y is the association strength between knowledge nodes x and y; the association strength includes co-occurrence association strength and behavioral association strength.

[0033] Furthermore, the information confidence status of the current knowledge node community is determined by combining the information aggregation degree Igd with the information aggregation degree confidence threshold IF; only when Igd > IF is the information aggregation of the knowledge nodes in the current knowledge node community considered trustworthy; if Igd ≤ IF, the information aggregation of the knowledge nodes in the current knowledge node community is considered to have a trust risk.

[0034] It outputs a reliable knowledge node community and pushes corresponding knowledge information based on user information retrieval behavior; when a user retrieves knowledge information for the first time, it locates the corresponding knowledge node in the corresponding knowledge community and pushes the associated knowledge nodes level by level.

[0035] A knowledge platform construction and management system that utilizes AI technology:

[0036] The system includes a knowledge retrieval module, a distributed processing module, an association architecture module, and a community evaluation module;

[0037] The knowledge retrieval module retrieves knowledge information stored in the enterprise database and constructs knowledge tags;

[0038] The distribution processing module determines the generation and utilization rate of corresponding knowledge information based on the knowledge tags, and constructs a knowledge node model to obtain a knowledge node distribution map;

[0039] The association architecture module classifies knowledge nodes based on the knowledge node distribution map and performs association analysis on the knowledge nodes to obtain the association strength between knowledge nodes, so as to construct a knowledge node combination association chain.

[0040] The community assessment module analyzes the chain affinity between the knowledge node combination association chains, divides the knowledge node combination association chains into knowledge node communities, analyzes the information aggregation degree of knowledge nodes in the knowledge node communities, and determines the confidence status of the knowledge node communities.

[0041] The knowledge retrieval module includes a database retrieval unit and a data tag composition unit;

[0042] The database retrieval unit is used to retrieve knowledge information stored in the enterprise database;

[0043] The data tag component unit is used to construct knowledge tags based on knowledge information;

[0044] The distribution processing module includes a utilization analysis unit and a distribution map construction unit;

[0045] The utilization analysis unit generates utilization Ur analysis based on each knowledge tag in the enterprise database;

[0046] The distribution map component constructs a knowledge node model and imports the corresponding knowledge information generation utilization rate Ur, performs node mapping on the knowledge information, determines the coordinates of the knowledge information to determine the node distribution status of the knowledge information, and obtains a knowledge node distribution map; and performs classification processing on the knowledge nodes.

[0047] The associated architecture module includes a multi-level association analysis unit and an association chain construction unit;

[0048] The multi-level association analysis unit performs co-occurrence association strength and behavioral association strength analysis on the main knowledge nodes and peripheral knowledge nodes respectively, based on the knowledge node classification results.

[0049] The association chain construction unit is used to construct an association chain of main knowledge nodes based on the co-occurrence association strength analysis data of main knowledge nodes, and to construct a knowledge node combination association chain by combining the behavior association strength analysis data of edge knowledge nodes.

[0050] The community assessment module includes a community division unit and an information aggregation assessment unit;

[0051] The community partitioning unit is used to traverse the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, perform chain affinity Ct(y,e) analysis on adjacent combination association chains, and finally partition knowledge node communities.

[0052] The information aggregation evaluation unit, based on the knowledge node community, determines the information weight of the connection edge between the corresponding knowledge nodes by obtaining the correlation strength between any knowledge nodes, performs information aggregation degree Igd analysis on the knowledge nodes in the knowledge node community, judges the information confidence status of the current knowledge node community, outputs a credible knowledge node community, and pushes out corresponding knowledge information based on the user's information retrieval behavior.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] This invention constructs a knowledge node model by automating the retrieval, tagging, and analysis of enterprise knowledge information, obtaining a node distribution map, and classifying knowledge nodes. This enables multi-level association analysis of knowledge nodes to build node association chains. Based on association chain affinity analysis, it identifies knowledge node communities and performs information aggregation and evaluation on these communities to determine their confidence level. This invention employs automated data tagging and association analysis processes, overcoming the limitations of traditional manual methods for large-scale knowledge information processing. Furthermore, it aggregates and associates scattered knowledge information through association mining, forming knowledge communities. This greatly facilitates user access to and interaction with knowledge information, improving the correlation and sharing of knowledge information. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the structure of a knowledge platform construction and management system that applies AI technology according to the present invention;

[0056] Figure 2 This is a flowchart illustrating a knowledge platform construction and management method using AI technology according to the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Example: Figure 1 As shown, the present invention provides a technical solution:

[0059] A knowledge platform construction and management system that utilizes AI technology:

[0060] The system includes a knowledge retrieval module, a distributed processing module, an association architecture module, and a community evaluation module;

[0061] Among them, the knowledge retrieval module retrieves knowledge information stored in the enterprise database and constructs knowledge tags;

[0062] The distributed processing module determines the generation and utilization rate of corresponding knowledge information based on knowledge tags, and constructs a knowledge node model to obtain a knowledge node distribution map;

[0063] The association architecture module classifies knowledge nodes based on the knowledge node distribution map and performs association analysis on the knowledge nodes to obtain the association strength between knowledge nodes, so as to construct a knowledge node combination association chain;

[0064] The community assessment module analyzes the chain affinity between knowledge node combination association chains, divides the knowledge node combination association chains into knowledge node communities, and analyzes the information aggregation degree of knowledge nodes in the knowledge node communities to determine the confidence status of the knowledge node communities.

[0065] The knowledge retrieval module includes a database retrieval unit and a data tag composition unit.

[0066] The database retrieval unit is used to retrieve knowledge information stored in the enterprise database;

[0067] Data tag components are used to construct knowledge tags based on knowledge information;

[0068] The distributed processing module includes a utilization analysis unit and a distribution map construction unit;

[0069] The utilization analysis unit generates utilization Ur analysis based on the knowledge tags in the enterprise database;

[0070] The distribution map component constructs a knowledge node model and imports the corresponding knowledge information generation and utilization rate Ur. It then maps the knowledge information to nodes, determines the coordinates of the knowledge information to determine the node distribution status of the knowledge information, and obtains a knowledge node distribution map. Finally, it classifies the knowledge nodes.

[0071] The related architecture module includes multi-level related analysis units and related chain construction units;

[0072] Based on the knowledge node classification results, the multi-level association analysis unit performs co-occurrence association strength and behavioral association strength analysis on the main knowledge nodes and peripheral knowledge nodes respectively;

[0073] The association chain construction unit is used to construct the association chain of the main knowledge nodes based on the co-occurrence association strength analysis data of the main knowledge nodes, and to construct the knowledge node combination association chain by combining the behavior association strength analysis data of the peripheral knowledge nodes.

[0074] The community assessment module includes community division units and information aggregation assessment units;

[0075] The community partitioning unit is used to traverse the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, perform chain affinity Ct(y,e) analysis on adjacent combination association chains, and finally partition knowledge node communities;

[0076] The information aggregation evaluation unit, based on the knowledge node community, determines the information weight of the connection edge between the corresponding knowledge nodes by obtaining the correlation strength between any knowledge nodes, and performs information aggregation degree IgD analysis on the knowledge nodes within the knowledge node community; it also judges the information confidence status of the current knowledge node community, outputs a credible knowledge node community, and pushes corresponding knowledge information based on user information retrieval behavior.

[0077] like Figure 2 As shown, the present invention provides another technical solution:

[0078] A method for building and managing a knowledge platform using AI technology, comprising the following steps:

[0079] Retrieve knowledge information stored in the enterprise database and construct knowledge tags;

[0080] The generation and utilization rate of corresponding knowledge information is determined based on knowledge tags, and a knowledge node model is constructed to obtain a knowledge node distribution map;

[0081] Based on the knowledge node distribution map, knowledge nodes are classified and correlation analysis is performed to obtain the correlation strength between knowledge nodes in order to construct a knowledge node combination correlation chain.

[0082] By analyzing the chain affinity between knowledge node combination association chains, knowledge node combination association chains are divided into knowledge node communities, and the information aggregation degree of knowledge nodes in the knowledge node communities is analyzed to determine the confidence status of the knowledge node communities.

[0083] Furthermore, the knowledge information is business data stored in the enterprise database; the knowledge tag consists of the knowledge information and the corresponding time data set; the time data set includes the storage time t of the knowledge information. c Call time t d The number of calls f and the number of productions p; where the number of productions of knowledge information includes the number of times the same knowledge information is repeatedly generated and updated.

[0084] Furthermore, based on the knowledge tags in the enterprise database, a knowledge information generation and utilization rate (Ur) analysis is performed, and the analysis calculation is as follows:

[0085] ;

[0086] Based on the knowledge information generation and utilization rate (Ur) analysis data, this method constructs a knowledge node model and imports the corresponding knowledge information generation and utilization rate (Ur) to perform node mapping on the knowledge information; it constructs a planar coordinate system with the knowledge information storage time (t) as the reference. c Using the x-axis as the horizontal axis and the generation utilization rate Ur as the y-axis, we determine the knowledge information coordinates to determine the node distribution status of the knowledge information, thereby obtaining a knowledge node distribution map.

[0087] Furthermore, based on the knowledge node distribution map, a horizontal ray is taken along the vertical axis and translated in the vertical direction. The density of knowledge nodes covered on different rays is recorded simultaneously. The vertical axis value of the ray with the maximum density is determined and recorded as the classification threshold Uc. By comparing the generation utilization rate Ur of each knowledge node with the classification threshold Uc, each knowledge node is classified. Since the value on the vertical axis corresponds to the generation utilization rate of the knowledge node, rays are constructed on the vertical axis and translated vertically to obtain the number of knowledge nodes covered by each ray corresponding to each value on the vertical axis. The density of covered knowledge nodes is determined by analyzing the proportion of the number of covered knowledge nodes in the knowledge node distribution map. The maximum value is taken as the classification threshold of the knowledge node, which reflects the dispersion of the utilization distribution of the corresponding knowledge information in the database. The concentrated value is taken as the discrimination object to realize the dynamic identification of the utilization of knowledge information.

[0088] If Ur≥Uc, then the corresponding knowledge node will be classified as a main knowledge node.

[0089] If Ur < Uc, then the corresponding knowledge node will be classified as an edge knowledge node;

[0090] Based on the knowledge node classification results, the main knowledge nodes and peripheral knowledge nodes are divided into two categories, and corresponding knowledge node sets are constructed, denoted as the main node set and the peripheral node set.

[0091] Furthermore, we iterate through the set of backbone nodes and select any non-repeating backbone knowledge nodes i and j for co-occurrence association analysis to obtain the co-occurrence association strength cas(i,j) of the corresponding backbone knowledge nodes i and j. The analysis and calculation are as follows:

[0092] ;

[0093] Where i and j are the numbers of the main knowledge nodes; w(i,j) is the weight of the main knowledge nodes i and j in the same text source; M(i,j) is the number of times the main knowledge nodes i and j appear together in the same text source; M(i) and M(j) are the total number of times the main knowledge nodes i and j appear alone in their respective text sources.

[0094] When the co-occurrence association strength cas(i,j)≥cz of any two main knowledge nodes i and j, main knowledge nodes i and j are determined to be associated main knowledge nodes; the associated main knowledge nodes are then connected to construct a main knowledge node association chain; when cas(i,j)<cz, main knowledge nodes i and j cannot form an association and do not constitute an association chain; where cz is the comparison threshold; where a main knowledge node is simultaneously associated with three other main knowledge nodes, then it will have one main knowledge point existing in multiple main knowledge node association chains.

[0095] In this embodiment, when a certain backbone knowledge node cannot form a connection with any other backbone knowledge node, it is a standalone backbone knowledge node and cannot build a connection chain. Therefore, it is still regarded as a connection chain in a special state, and after conducting behavioral association strength analysis with the edge knowledge nodes, it continues to participate in the knowledge community division analysis.

[0096] Furthermore, based on the association chain of the main knowledge nodes, behavioral association analysis is performed on each main knowledge node in the association chain between the edge knowledge nodes and the main knowledge nodes in the edge node set to obtain the behavioral association strength bas(p,k) between each main knowledge node and the edge knowledge node, which is calculated as follows:

[0097] ;

[0098] Where p is the number of the main knowledge node in the main knowledge node association chain; k is the number of the edge knowledge node in the edge node set; w(p,k) is the number of times the main knowledge node p and the edge knowledge node k are called simultaneously; N(p) and N(k) are the number of times the main knowledge node p and the edge knowledge node k are called individually, respectively.

[0099] When the behavioral association strength bas(p,k) ≥ cz between any main knowledge node p in the main knowledge node association chain and any edge knowledge node k in the edge node set, edge knowledge node k is determined to be an associated edge knowledge node of main knowledge node p. Based on the association chain of main knowledge nodes, each main knowledge node is connected with its corresponding associated edge knowledge node to obtain a knowledge node combination association chain. When bas(p,k) < cz, edge knowledge node k cannot form an association with main knowledge node p and does not constitute an association chain. Among these, there exists a main knowledge node that forms an association with multiple edge knowledge nodes.

[0100] Furthermore, by traversing the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, the chain affinity Ct(y,e) of adjacent combination association chains is analyzed based on the number of overlapping knowledge nodes obtained from the traversal; where y and e are the numbers of adjacent knowledge node combination association chains, respectively; the analysis results are as follows:

[0101] Ct(y,e)=[n(y,e)] / [n(y)+n(e)];

[0102] Wherein, n(y,e) is the number of knowledge nodes that overlap between adjacent knowledge node combination association chains y and e; n(y) and n(e) are the number of knowledge nodes on adjacent knowledge node combination association chains y and e, respectively; where adjacent knowledge node combination association chains are knowledge node combination association chains that have overlapping knowledge nodes on the knowledge node distribution map.

[0103] Furthermore, when the chain affinity Ct(y,e) between adjacent knowledge node combination association chains reaches a preset affinity threshold, the current adjacent knowledge node combination association chains are divided into knowledge node communities. Based on the knowledge node community, the information weight of the connection edge between the corresponding knowledge nodes is determined by obtaining the association strength between any knowledge nodes, and the information aggregation degree Igd analysis is performed on the knowledge nodes within the knowledge node community, which is calculated as follows:

[0104] ;

[0105] Where m is the number of interconnected edges between knowledge nodes within the knowledge node community; x and y are the unique node numbers within the knowledge node community; Q x,y The weight of the concatenation edge information between knowledge nodes x and y; Q x,h and Q y,h γ represents the sum of the weights of the concatenated edges connecting to knowledge nodes x and y, respectively; x:y(α, β) are the knowledge node category judgment parameters; where, the knowledge node category judgment parameters are obtained by judging the categories of knowledge nodes x and y. When x and y are of the same category, the parameter value α is taken, and otherwise the parameter value β is taken; where Qx,y is the association strength between knowledge nodes x and y; the association strength includes co-occurrence association strength and behavioral association strength.

[0106] Furthermore, the information confidence status of the current knowledge node community is determined by combining the information aggregation degree Igd with the information aggregation degree confidence threshold IF; only when Igd > IF is the information aggregation of the knowledge nodes in the current knowledge node community considered trustworthy; if Igd ≤ IF, the information aggregation of the knowledge nodes in the current knowledge node community is considered to have a trust risk.

[0107] It outputs a credible knowledge node community and pushes corresponding knowledge information based on user information retrieval behavior; it locates the corresponding knowledge node in the corresponding knowledge community when the user first retrieves knowledge information, and pushes the related knowledge nodes level by level.

[0108] In this embodiment, the step-by-step push for users to retrieve knowledge information is specifically as follows: when a user retrieves the first piece of knowledge information, the knowledge node is located in the knowledge node community, and other knowledge nodes that have a connection edge with it are located at the same time, and the first-level knowledge node is pushed. Subsequently, according to the first-level knowledge node selected by the user, the above operation is repeated to push knowledge nodes step by step until the push of related nodes ends or the user ends the information retrieval.

[0109] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for constructing and managing a knowledge platform using AI technology, characterized in that: Retrieve knowledge information stored in the enterprise database and construct knowledge tags; The generation and utilization rate of corresponding knowledge information is determined based on the knowledge tags, and a knowledge node model is constructed to obtain a knowledge node distribution map; Based on the knowledge node distribution map, knowledge nodes are classified, and association analysis is performed on the knowledge nodes to obtain the association strength between knowledge nodes, so as to construct a knowledge node combination association chain. The knowledge node classification is based on the knowledge node distribution map. A horizontal ray is taken along the vertical axis and translated in the vertical direction. The density of knowledge nodes covered on different rays is recorded simultaneously. The value of the vertical axis of the ray with the maximum density is determined and recorded as the classification threshold Uc. Each knowledge node is classified by comparing its generation utilization rate Ur with the classification threshold Uc. If Ur≥Uc, then the corresponding knowledge node will be classified as a main knowledge node. If Ur < Uc, then the corresponding knowledge node will be classified as an edge knowledge node; Based on the knowledge node classification results, the main knowledge nodes and peripheral knowledge nodes are divided in a unified manner, and corresponding knowledge node sets are constructed, denoted as the main node set and the peripheral node set. The correlation strength between knowledge nodes includes co-occurrence correlation strength cas(i,j) and behavioral correlation strength bas(p,k); The analysis of the co-occurrence association strength cas(i,j) involves traversing the set of backbone nodes and selecting any non-repeating backbone knowledge nodes i and j to perform co-occurrence association analysis, thereby obtaining the co-occurrence association strength cas(i,j) of the corresponding backbone knowledge nodes i and j. When the co-occurrence association strength cas(i,j)≥cz of any two main knowledge nodes i and j, the main knowledge nodes i and j are determined to be related main knowledge nodes. The main knowledge nodes are linked together to construct a main knowledge node association chain; when cas(i,j)<cz, the main knowledge nodes i and j cannot form an association and do not constitute an association chain; where cz is the comparison threshold. The analysis of the behavioral association strength bas(p,k) is based on the association chain of the main knowledge nodes. The analysis is performed on each main knowledge node in the association chain between the edge knowledge nodes and the main knowledge nodes in the edge node set to obtain the behavioral association strength bas(p,k) between each main knowledge node and the edge knowledge node. By analyzing the chain affinity between the knowledge node combination association chains, the knowledge node combination association chains are divided into knowledge node communities, and the information aggregation degree of knowledge nodes in the knowledge node communities is analyzed to determine the confidence status of the knowledge node communities.

2. The knowledge platform construction and management method using AI technology according to claim 1, characterized in that: The analysis and calculation of the co-occurrence correlation strength cas(i,j) are as follows: ； Where i and j are the numbers of the main knowledge nodes; w(i,j) is the weight of the main knowledge nodes i and j in the same text source; M(i,j) is the number of times the main knowledge nodes i and j appear together in the same text source; M(i) and M(j) are the total number of times the main knowledge nodes i and j appear alone in their respective text sources.

3. The knowledge platform construction and management method using AI technology according to claim 1, characterized in that: The analysis and calculation of the behavioral association strength bas(p,k) are as follows: ； Where p is the number of the main knowledge node in the main knowledge node association chain; k is the number of the edge knowledge node in the edge node set; w(p,k) is the number of times the main knowledge node p and the edge knowledge node k are called simultaneously; N(p) and N(k) are the number of times the main knowledge node p and the edge knowledge node k are called individually, respectively. When the behavioral association strength bas(p,k) ≥ cz between any main knowledge node p in the main knowledge node association chain and any edge knowledge node k in the edge node set, edge knowledge node k is determined to be an associated edge knowledge node of main knowledge node p. Based on the association chain of main knowledge nodes, each main knowledge node is connected with its corresponding associated edge knowledge node to obtain a knowledge node combination association chain. When bas(p,k) < cz, edge knowledge node k cannot form an association with main knowledge node p and does not constitute an association chain.

4. The knowledge platform construction and management method using AI technology according to claim 1, characterized in that: By traversing the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, the chain affinity Ct(y,e) of adjacent combination association chains is analyzed based on the number of overlapping knowledge nodes obtained from the traversal; where y and e are the numbers of adjacent knowledge node combination association chains, respectively; the analysis results are as follows: Ct(y,e)=[n(y,e)] / [n(y)+n(e)]; Where n(y,e) is the number of overlapping knowledge nodes between adjacent knowledge node combination association chains y and e; n(y) and n(e) are the number of knowledge nodes on adjacent knowledge node combination association chains y and e, respectively.

5. The knowledge platform construction and management method using AI technology according to claim 1, characterized in that: When the chain affinity Ct(y,e) between adjacent knowledge node combinations reaches a preset affinity threshold, the current adjacent knowledge node combinations are divided into knowledge node communities. Within a knowledge node community, the information weight of the connecting edges between corresponding knowledge nodes is determined by obtaining the association strength between any two knowledge nodes. Information aggregation degree Igd analysis is then performed on the knowledge nodes within the knowledge node community, and its calculation is as follows: ； Where m is the number of interconnected edges between knowledge nodes within the knowledge node community; x and y are the unique node numbers within the knowledge node community; Q x,y The weight of the concatenation edge information between knowledge nodes x and y; Q x,h and Q y,h γ represents the sum of the weights of the concatenated edges connecting to knowledge nodes x and y, respectively; x:y (α, β) are the parameters for determining the category of knowledge nodes; The information confidence status of the current knowledge node community is determined by combining the information aggregation degree Igd with the information aggregation degree confidence threshold IF. Only when Igd > IF is the information aggregation of the knowledge nodes in the current knowledge node community considered trustworthy. If Igd ≤ IF, the information aggregation of the knowledge nodes in the current knowledge node community is considered to have a trust risk. It outputs a trustworthy knowledge node community and pushes corresponding knowledge information based on user information retrieval behavior.

6. The knowledge platform construction and management method using AI technology according to claim 1, characterized in that: The knowledge information is business data stored in the enterprise database; the knowledge tag consists of knowledge information and a corresponding time data set; the time data set includes the storage time t of the knowledge information. c Call time t d The number of calls f and the number of productions p; Based on the knowledge tags in the enterprise database, the generation and utilization rate of knowledge information is analyzed using Ur, and the analysis calculation is as follows: ； Based on the knowledge information generation and utilization rate Ur analysis data, a knowledge node model is constructed and the corresponding knowledge information generation and utilization rate Ur is imported to perform node mapping of knowledge information; It constructs a planar coordinate system to store knowledge information over time t. c Using the x-axis as the horizontal axis and the generation utilization rate Ur as the y-axis, we determine the knowledge information coordinates to determine the node distribution status of the knowledge information, thereby obtaining a knowledge node distribution map.

7. A system for implementing the knowledge platform construction and management method using AI technology as described in any one of claims 1-6, characterized in that: The system includes a knowledge retrieval module, a distributed processing module, an association architecture module, and a community evaluation module; The knowledge retrieval module retrieves knowledge information stored in the enterprise database and constructs knowledge tags; The distribution processing module determines the generation and utilization rate of corresponding knowledge information based on the knowledge tags, and constructs a knowledge node model to obtain a knowledge node distribution map; The association architecture module classifies knowledge nodes based on the knowledge node distribution map and performs association analysis on the knowledge nodes to obtain the association strength between knowledge nodes, so as to construct a knowledge node combination association chain. The community assessment module analyzes the chain affinity between the knowledge node combination association chains, divides the knowledge node combination association chains into knowledge node communities, analyzes the information aggregation degree of knowledge nodes in the knowledge node communities, and determines the confidence status of the knowledge node communities.

8. A knowledge platform construction and management system applying AI technology according to claim 7, characterized in that: The knowledge retrieval module includes a database retrieval unit and a data tag composition unit; The database retrieval unit is used to retrieve knowledge information stored in the enterprise database; The data tag component unit is used to construct knowledge tags based on knowledge information; The distribution processing module includes a utilization analysis unit and a distribution map construction unit; The utilization analysis unit generates utilization Ur analysis based on each knowledge tag in the enterprise database; The distribution map component constructs a knowledge node model and imports the corresponding knowledge information generation utilization rate Ur, performs node mapping on the knowledge information, determines the coordinates of the knowledge information to determine the node distribution status of the knowledge information, and obtains a knowledge node distribution map; and performs classification processing on the knowledge nodes.

9. A knowledge platform construction and management system applying AI technology according to claim 8, characterized in that: The associated architecture module includes a multi-level association analysis unit and an association chain construction unit; The multi-level association analysis unit performs co-occurrence association strength and behavioral association strength analysis on the main knowledge nodes and peripheral knowledge nodes respectively, based on the knowledge node classification results. The association chain construction unit is used to construct the association chain of the main knowledge nodes based on the co-occurrence association strength analysis data of the main knowledge nodes, and to construct the knowledge node combination association chain by combining the behavior association strength analysis data of the peripheral knowledge nodes; the community evaluation module includes a community division unit and an information aggregation evaluation unit. The community partitioning unit is used to traverse the number of overlapping knowledge nodes between adjacent knowledge node combination association chains, perform chain affinity Ct(y,e) analysis on adjacent combination association chains, and finally partition knowledge node communities. The information aggregation evaluation unit, based on the knowledge node community, determines the information weight of the connection edge between corresponding knowledge nodes by obtaining the correlation strength between any knowledge nodes, and performs information aggregation degree Igd analysis on the knowledge nodes within the knowledge node community. It also determines the information confidence status of the current knowledge node community, outputs the reliable knowledge node community, and pushes out corresponding knowledge information based on the user's information retrieval behavior.

Images