A Visual Multidimensional Knowledge Management and Learning System

Abstract

The present invention provides a visual multi-dimensional knowledge management and learning system, including a display, a graphics card, a processor, and a non-transitory storage medium. Among them, the non-transitory storage medium stores program modules that can be run by the processor to implement visual multi-dimensional knowledge management and learning. The program modules include: a digital map module; a knowledge acquisition module; a knowledge selection sub-module, a knowledge parsing module connected to the knowledge acquisition module; a blockification and pooling module, including a blockification sub-module and a pooling sub-module; an input module connected to a computer network; an evaluation module including an online expert system sub-module; and a display module. This system visually manages knowledge from five dimensions: geographical spatial location, knowledge, equipment, heterogeneous behavior, pooling and blockification. At the same time, an online expert system is constructed to provide a diversified all-round perspective management.

Description

Technical Field

[0001] The present invention relates to a multi-dimensional knowledge management and learning system, and particularly to a visual multi-dimensional knowledge management and learning system, belonging to the field of intelligent management. Background Art

[0002] Knowledge management generally realizes module windowing operations on a system platform. However, for the abstract objects of knowledge in multiple fields and multiple algorithms, and the entity objects of multiple devices, how to achieve a concrete system representation. In the prior art, it is usually achieved by using multiple fixed and unmodifiable distribution maps. However, due to the development and change of knowledge, the fields, algorithms, and objects involved may change at any time. Therefore, it is necessary to draw more and updated distribution maps, which is very labor-consuming. Therefore, how to perform operable visual management on these knowledge objects has become an urgent technical problem to be solved. Summary of the Invention

[0003] To solve the above problems of the prior art, the present invention provides a visual multi-dimensional knowledge management and learning system, including a display, a graphics card, a processor, and a non-transitory storage medium. Among them, the non-transitory storage medium stores program modules that can be run by the processor to implement visual multi-dimensional knowledge management and learning. The program modules include:

[0004] A digital map module for drawing and revising a digital geographical map.

[0005] A knowledge acquisition module, including a knowledge selection sub-module. The knowledge acquisition module is used to acquire knowledge data from a knowledge generation source and record the spatial location and generation time of the knowledge generation. The knowledge data includes knowledge related to devices related to the knowledge, specifically including the name of the device, the knowledge generation source and / or user to which it belongs, the uses related to the device, technical principles, as well as scientific and technological resources and social and human science resources, specifically including literature, pictures, videos, and audios. The knowledge also includes the name of the device, the knowledge generation source and / or user to which it belongs, the uses related to the device, technical principles, usage records, historical and real-time indicators, and environmental parameters, and device management information. Among them,

[0006] The knowledge generation source forms original knowledge data by inputting and calibrating its own spatial location, proficient knowledge field, and the knowledge data on the geographical digital map.

[0007] The knowledge selection sub-module is specifically used to request from the pooling sub-module and receive at least one small pool formed by newly input knowledge data by the user for the user to select an interested small pool for subsequent knowledge parsing.

[0008] A knowledge parsing module, connected to the knowledge acquisition module, for:

[0009] First, construct a Heterogeneous Graph Neural Network (HGNN), and use the received knowledge to generate the original knowledge data input by the source and the recorded spatial locations, and construct an Attention Mechanism (AM) based on the heterogeneous graph neural network.

[0010] Second, receive the small pool of interest sent by the knowledge selection sub-module, project it into the HGNN, call out all meta-paths (i.e., all meta-paths that theoretically satisfy the permutations and combinations defined by the user), and save the spatial location and time when the user selects the small pool of interest.

[0011] Third, after constructing the AM, again retrieve the knowledge context based on the knowledge data newly input by the user (in most cases, the new input is a search operation, i.e., inputting relevant knowledge data to search for any relevant knowledge information of interest, or there are other purposes such as use, learning, and organization), and based on the knowledge context, give the analysis result of the attention mechanism, and connect it to the digital map module. Form a first visualization area and a first editing area on the digital geographical map for the HGNN, the knowledge context, the analysis result, and the AM respectively.

[0012] The blockization and pooling module includes a blockization sub-module and a pooling sub-module, which are respectively used to blockize the heterogeneous graph neural network based on the analysis result, set the data management password for each block based on the blockized graph neural network, and update the HGNN and the AM within a predetermined time period, and accordingly continuously update the blocks and the data management password; and,

[0013] Collect the original knowledge data, pool the original knowledge data within a periodic time to form knowledge pools in different knowledge domains, and accept the knowledge data of the new input by the user (i.e., the input after pooling) and pool it to form at least one small pool.

[0014] The pooling sub-module is connected to the digital map module to form a second visualization area on the digital geographical map for all the meta-paths and the at least one small pool.

[0015] An input module, connected to a computer network, is used to receive the newly input knowledge data of the user, input it into the knowledge parsing module through the computer network. The knowledge data is clustered in a preset target field with domain labels. When the user newly inputs knowledge data, it is necessary to select the target field for the newly input knowledge data and select the domain label of the target. At the same time, record the knowledge operation behavior according to the user's new input behavior analysis, including at least one of querying, downloading, uploading, Q&A, discussion and their combinations.

[0016] An evaluation module, including an online expert system sub-module, is used for:

[0017] First, comprehensively analyze the analysis results of the retrieved knowledge context and attention mechanism, and give a final evaluation adapted to the true meaning of the user's search.

[0018] Second, according to all the retrieved meta-paths and the spatial location, give the user and other users' custom meta-paths around the user belonging to the spatial location. Further call the attention meta-paths whose association degree calculated by the attention mechanism exceeds the first threshold, as well as the spatial location and corresponding time corresponding to the attention users corresponding to the attention meta-paths for further knowledge evaluation.

[0019] Third, the online expert system sub-module is specifically used for re-learning based on the knowledge operation behaviors of all users and knowledge generation sources to form an online expert system, and online knowledge teaching based on the online expert system. Among them, the online expert system is integrated into the digital geographical map. By searching and clicking on at least one accurate spatial location on the digital geographical map, the online expert system sub-module pops up, and targeted re-learning and online knowledge teaching are carried out based on at least one accurate spatial location. Among them,

[0020] The evaluation module is connected to the digital map module to form a third visualization area for comprehensive analysis and final evaluation, and a second editing area for the online expert system sub-module on the digital geographical map.

[0021] A display module is used to display and operate the content in the first visualization area, the second visualization area, the first editing area, and the second editing area. Among them,

[0022] The knowledge generation source regularly inputs the knowledge data generated according to the classification of confidentiality levels into the knowledge acquisition module and transmits it to the knowledge parsing module for the update.

[0023] Optionally, the visualization objects in the first visualization area and the second visualization area can be changed, namely the perspectives of HGNN, all meta-paths, at least one small pool, and the comprehensive analysis result, and the sizes and distributions of the first visualization area, the second visualization area, the first editing area, and the second editing area are adjustable. The operations include at least one of dragging and rotating the visualization objects, clicking on them to pop up a dialog box or a search bar for searching and / or viewing, modifying, deleting, and / or restoring knowledge.

[0024] Optionally, a heterogeneous graph neural network is constructed, and the original knowledge data input by the received knowledge generation source and the recorded spatial locations are used. The method for constructing the attention mechanism based on the heterogeneous graph neural network includes:

[0025] S1 Establish a geographical map, mark the spatial locations on the geographical map as nodes of the heterogeneous graph neural network, and construct a heterogeneous graph network based on the real spatial locations;

[0026] S2 Construct a knowledge heterogeneous graph network, a device heterogeneous graph network, and a heterogeneous behavior network based on the nodes, and define the network nodes and the edges between the nodes;

[0027] S3 Construct multiple types of meta-paths in the networks in S2 respectively, and define neighbor nodes;

[0028] S4 First establish a node attention mechanism according to the multiple types of meta-paths, then construct a semantic attention mechanism, and finally construct a subgraph attention mechanism.

[0029] Based on the newly input knowledge data and / or the recorded knowledge operation behaviors of the user, retrieve the knowledge context, and give the analysis result of the attention mechanism based on the knowledge context. The specific method is:

[0030] S5 Input the knowledge into the trained AM to obtain the predicted target classification and / or record the relevant knowledge operation behaviors;

[0031] S6 In the predicted target classification related to the knowledge operation behaviors, draw the knowledge context according to the spatial locations and generation times of the knowledge, and retrieve the drawn knowledge context;

[0032] S7 Mark the nodes and / or subgraphs corresponding to the newly input knowledge data in the drawn knowledge context.

[0033] For the recorded knowledge operation behaviors, if the corresponding knowledge fields are mostly for viewing, that is, only the viewing function is marked. For another example, if the predicted knowledge operation behaviors have high correlations with viewing, downloading, uploading, etc., including high correlations between nodes and high correlations of meta-paths formed by a series of nodes, then mark these operations and the relevant paths.

[0034] The comprehensive analysis includes statistical analysis of at least one of the following types of distributions:

[0035] The knowledge domain corresponding to the target classification related to knowledge operation behavior corresponding to knowledge data, the first distribution on the geographical map, the second distribution in the visualized knowledge heterogeneous graph network, device heterogeneous graph network, and heterogeneous behavior network, and the traversal distribution (i.e., some nodes belong to the first distribution and the remaining part belongs to the second distribution). Moreover, statistical analysis of the knowledge context in the selected types of distributions is performed, and the first distribution and the second distribution are displayed in the second visualization area.

[0036] The method for knowledge evaluation includes: obtaining the spatial location and time of the user and the concerned user, finding the nodes containing at least one small pool according to the concerned meta-path, calculating through the attention mechanism to draw a distribution map of node association degrees, and displaying it in the second visualization area.

[0037] Optionally, the method of blockification includes: performing sectional clustering on the normalization coefficients of the subgraph clustering calculated according to the attention mechanism, and dividing the subgraphs into multiple blocks according to the sections. Beneficial effects

[0038] Visualize the management of knowledge from five dimensions: geographical spatial location, knowledge, device, heterogeneous behavior, pooling, and blockification, and form an online expert system accordingly, thus providing a diversified and all-round perspective management for the recording, imparting, and development of knowledge. Brief description of the drawings

[0039] Figure 1 A schematic diagram of the configuration of a visualized multi-dimensional knowledge management and learning system of the present invention,

[0040] Figure 2 A schematic diagram of the distribution of the first visualization area, second visualization area, first editing area, and second editing area on the display interface of the present invention. Detailed implementation manners

[0041] As Figure 1 shown, a visualized multi-dimensional knowledge management and learning system includes a display, a graphics card, a processor, and a non-transitory storage medium. Among them, the non-transitory storage medium stores program modules that can be run by the processor to implement visualized multi-dimensional knowledge management and learning. The program modules include:

[0042] The digital map module in the brown box; the knowledge acquisition module in the green box; including the knowledge selection sub-module in the green box, the knowledge parsing module in the red box connected to the knowledge acquisition module (represented by the red thin line, the same below); the blockification and pooling module in the blue box, including the blockification sub-module and the pooling sub-module; the input module in the black box connected to the computer network; the evaluation module in the cyan box including the online expert system sub-module; and the display module in the purple box.

[0043] Among them, the digital map module is used to draw and revise the digital geographical map, the knowledge acquisition module is used to obtain knowledge data from the knowledge generation source and record the spatial location and generation time of the knowledge generation. The knowledge data includes knowledge related to the equipment related to the knowledge, specifically including the name of the equipment, the knowledge generation source and user to which it belongs, the uses related to the equipment, the technical principles, as well as scientific and technological resources and social and human science resources, specifically including literature, pictures, videos, and audios. The knowledge also includes the name of the equipment, the knowledge generation source and user to which it belongs, the uses related to the equipment, the technical principles, usage records, historical and real-time indicators, and environmental parameters, and equipment management information. Among them,

[0044] The knowledge generation source forms the original knowledge data by inputting and calibrating its own spatial location, the knowledge field it is good at, and the knowledge data on the digital geographical map.

[0045] The knowledge selection sub-module is specifically used to request from the pooling sub-module and receive the at least one small pool formed by the newly input knowledge data of the user for the user to select the interested small pool for subsequent knowledge parsing.

[0046] The knowledge parsing module is used for:

[0047] First, construct a heterogeneous graph neural network (HGNN), and use the original knowledge data input by the received knowledge generation source and the recorded spatial location to construct an attention mechanism (AM) based on the heterogeneous graph neural network.

[0048] Second, receive the interested small pool sent by the knowledge selection sub-module, project it into the HGNN, call out all meta-paths, and save the spatial location and time when the user selects the interested small pool.

[0049] Third, after the construction of the AM is completed, retrieve the knowledge context again based on the newly input knowledge data of the user and / or the recorded knowledge operation behavior, and give the analysis result of the attention mechanism based on the knowledge context, and connect it to the digital map module to form the first visualization area and the first editing area on the digital geographical map for the HGNN, the knowledge context, the analysis result, and the AM respectively.

[0050] A blockification sub-module and a pooling sub-module are respectively used to blockify the heterogeneous graph neural network based on the analysis result, and at the same time set the data management password for each block based on the blockified graph neural network, and update the HGNN and AM within a predetermined time period, and accordingly continuously update the block and the data management password; and,

[0051] It is used to collect the original knowledge data, pool the original knowledge data within a periodic time to form knowledge pools in different knowledge fields, and accept the knowledge data of the new input of the user (that is, the input after pooling) and pool it to form at least one small pool.

[0052] The pooling sub-module is connected to the digital map module to form a second visualization area on the digital geographical map for all the meta-paths and the at least one small pool.

[0053] The input module is used to receive the new input knowledge data of the user, input it into the knowledge parsing module through a computer network, and the knowledge data is clustered in a preset target field with a domain label. And when the user newly inputs knowledge data, it is necessary to select the target field for the newly input knowledge data, select the target domain label. At the same time, record the knowledge operation behavior according to the new input behavior of the user, including at least one of querying, downloading, uploading, Q&A, discussion and their combinations.

[0054] The evaluation module is used for:

[0055] First, comprehensively analyze the analysis results of the retrieved knowledge context and attention mechanism, and give a final evaluation adapted to the true meaning of the user's search.

[0056] Second, according to all the retrieved meta-paths and the spatial location, give the user-defined meta-paths of the user and other users around the user belonging to the spatial location, further call the attention meta-paths whose correlation degree calculated by the attention mechanism exceeds the first threshold through the attention mechanism, and the spatial location and corresponding time corresponding to the attention users corresponding to the attention meta-paths for further knowledge evaluation.

[0057] Third, the online expert system sub-module is specifically used for re-learning based on the knowledge operation behaviors of all users and knowledge generation sources to form an online expert system, and online knowledge teaching based on the online expert system. Among them, the online expert system is integrated into the digital geographical map. By searching and clicking at at least one accurate spatial location on the digital geographical map, the online expert system sub-module pops up, and targeted re-learning and online knowledge teaching are carried out based on at least one accurate spatial location. Among them,

[0058] The evaluation module is connected to the digital map module, forming a third visualization area for comprehensive analysis and final evaluation on the digital geographical map, as well as a second editing area for the online expert system sub-module.

[0059] The display module is used to display and operate on the content in the first visualization area, the second visualization area, the first editing area, and the second editing area. Among them,

[0060] The knowledge generation source regularly inputs the knowledge data generated according to the classification of secrecy levels into the knowledge acquisition module, and transmits it to the knowledge parsing module for the update.

[0061] As Figure 2 shown, it is the distribution of the first visualization area, the second visualization area, the first editing area, and the second editing area displayed on the display, and the area and distribution of each area can be changed and adjusted. Figure 2 In the first / second visualization area, 11 distribution nodes of the spatial location are given, forming a heterogeneous graph network based on the real spatial location, and a second visualization area is formed near each node of the network for the visualization of the knowledge heterogeneous graph network, the device heterogeneous graph network, and the heterogeneous behavior network. The AM is constructed and trained in the first editing area, and the online expert system sub-module is re-learned and knowledge is imparted in the second editing area. By clicking on the spatial location node in the first visualization area, the online expert system sub-module in the second editing area can be popped up.

[0062] In the first editing area, to construct a heterogeneous graph neural network and use the original knowledge data input by the received knowledge generation source and the recorded spatial location to construct a method for the attention mechanism based on the heterogeneous graph neural network, the steps include:

[0063] S1 Establish a geographical map, mark the spatial location on the geographical map as the nodes of the heterogeneous graph neural network, and construct a heterogeneous graph network based on the real spatial location;

[0064] S2 Construct a knowledge heterogeneous graph network, a device heterogeneous graph network, and a heterogeneous behavior network based on the nodes, and define the network nodes and the edges between the nodes;

[0065] S3 Construct multiple types of meta-paths in the networks in S2 respectively, and define the neighbor nodes;

[0066] S4 First establish a node attention mechanism according to the multiple types of meta-paths, then construct a semantic attention mechanism, and finally construct a sub-graph attention mechanism.

[0067] Based on the newly input knowledge data and / or the recorded knowledge operation behavior of the user, retrieve the knowledge context, and give the analysis result of the attention mechanism based on the knowledge context. The specific method is:

[0068] S5 inputs the knowledge into the trained AM to obtain the predicted target classification and / or record the relevant knowledge operation behaviors;

[0069] S6 In the predicted target classification related to the knowledge operation behaviors, draw the knowledge context according to the corresponding spatial location and generation time of the knowledge, and retrieve the drawn knowledge context;

[0070] S7 Mark the nodes and / or subgraphs corresponding to the newly input knowledge data in the drawn knowledge context.

[0071] The predicted target classification and / or the recorded relevant knowledge operation behaviors, and the drawn knowledge context are all displayed in the third visualization area.

[0072] The comprehensive analysis includes statistical analysis of at least one of the following types of distributions:

[0073] The knowledge field corresponding to the target classification related to the knowledge operation behaviors corresponding to the knowledge data, the first distribution on the geographical map, the second distribution in the visualized knowledge heterogeneous graph network, device heterogeneous graph network, and heterogeneous behavior network, and the crossing distribution. Moreover, statistical analysis is performed on the knowledge context in the selected types of distributions. The first and second distributions are both displayed in the second visualization area.

[0074] The method for knowledge evaluation includes: obtaining the spatial location and time of the user and the users being concerned, finding the nodes containing at least one small pool according to the concerned meta-path, calculating and drawing the node correlation degree distribution map through the attention mechanism, and displaying it in the second visualization area.

[0075] The method of blockification includes: performing sectional clustering on the normalized coefficients of the subgraph clustering calculated according to the attention mechanism, and dividing the subgraphs into multiple blocks according to the sections.

Claims

1. A visual multi-dimensional knowledge management and learning system, characterized in that, It includes a display, a graphics card, a processor, and a non-transitory storage medium. Among them, the non-transitory storage medium stores program modules that can be run by the processor to implement visual multi-dimensional knowledge management and learning. The program modules include: A digital map module for drawing and revising digital geographical maps. A knowledge acquisition module, including a knowledge selection sub-module. The knowledge acquisition module is used to obtain knowledge data from knowledge generation sources and record the spatial location and generation time of knowledge generation. The knowledge data includes knowledge related to devices related to knowledge, specifically including the name of the device, the knowledge generation source and / or user to which it belongs, the uses related to the device, technical principles, as well as scientific and technological resources and social and human science resources, specifically including literature, pictures, videos, and audios. The knowledge also includes the name of the device, the knowledge generation source and / or user to which it belongs, the uses related to the device, technical principles, usage records, historical and real-time indicators, and environmental parameters, and device management information. Among them, The knowledge generation source forms original knowledge data by inputting and calibrating its own spatial location, proficient knowledge field, and the knowledge data on the geographical digital map. The knowledge selection sub-module is specifically used to request at least one small pool formed by the user's newly input knowledge data received by the pooling sub-module for the user to select the interested small pool for subsequent knowledge parsing. A knowledge parsing module, connected to the knowledge acquisition module, is used for: First, construct a Heterogeneous Graph Neural Networks (HGNN) and use the original knowledge data input by the received knowledge generation source and the recorded spatial location to construct an Attention Mechanism (AM) based on the heterogeneous graph neural network. Second, receive the interested small pool sent by the knowledge selection sub-module, project it into the HGNN, call out all meta-paths, and save the spatial location and time when the user selects the interested small pool. Third, after the construction of the AM is completed, retrieve the knowledge context again based on the user's newly input knowledge data and / or recorded knowledge operation behaviors, and give the analysis result of the attention mechanism based on the knowledge context. Connect to the digital map module to form a first visualization area and a first editing area on the digital geographical map for the HGNN, the knowledge context, the analysis result, and the AM respectively. The blockification and pooling module includes a blockification sub-module and a pooling sub-module, which are respectively used to blockify the heterogeneous graph neural network based on the analysis result, set the data management password for each block based on the blockified graph neural network, and update the HGNN and AM within a predetermined time period, and accordingly continuously update the blocks and the data management password; and, It is used to collect original knowledge data, pool the original knowledge data within a cycle time to form knowledge pools in different knowledge fields, and accept the user's newly input knowledge data and pool it to form at least one small pool. The pooling sub-module is connected to the digital map module, and forms a second visualization area on the digital geographical map for all the meta-paths and the at least one small pool. An input module, connected to a computer network, is configured to receive the newly input knowledge data of the user, input it into the knowledge parsing module through the computer network. The knowledge data is clustered in a preset target domain with a domain label. When the user newly inputs knowledge data, the target domain of the newly input knowledge data needs to be selected, and the domain label of the target is selected. At the same time, according to the new input behavior of the user, the knowledge operation behavior is analyzed, including at least one of querying, downloading, uploading, Q&A, discussion and their combinations. An evaluation module, including an online expert system sub-module, is configured to: First, comprehensively analyze the analyzed results of the retrieved knowledge context and the attention mechanism, and give a final evaluation adapted to the true meaning of the user's search. Second, according to all the meta-paths and the spatial location called out, give the user and other users' customized meta-paths around the user belonging to the spatial location. Through the attention mechanism, further call the attention meta-paths whose association degree calculated by the attention mechanism exceeds the first threshold, and the spatial locations and corresponding times corresponding to the attention users corresponding to the attention meta-paths for further knowledge evaluation. Third, the online expert system sub-module is specifically used for re-learning based on the knowledge operation behaviors of all users and knowledge generation sources to form an online expert system, and online knowledge teaching based on the online expert system. Among them, the online expert system is integrated into the digital geographical map. By searching and clicking on at least one accurate spatial location on the digital geographical map, the online expert system sub-module pops up, and targeted re-learning and online knowledge teaching are carried out based on at least one accurate spatial location. Among them, The evaluation module is connected to the digital map module, and forms a third visualization area for comprehensive analysis and final evaluation, and a second editing area of the online expert system sub-module on the digital geographical map. A display module is configured to display and operate the content in the first visualization area, the second visualization area, the first editing area, and the second editing area. Among them, The knowledge generation source regularly inputs the knowledge data generated according to the classification level into the knowledge acquisition module and transmits it to the knowledge parsing module for the update.

2. The system according to claim 1, wherein The visualization objects in the first visualization area and the second visualization area can be changed, that is, the perspectives of HGNN, all meta-paths, at least one small pool, and the comprehensive analysis results can be changed. The sizes and distributions of the first visualization area, the second visualization area, the first editing area, and the second editing area are adjustable. The operations include dragging and rotating the visualization objects, and clicking on them to pop up a dialog box or a search bar for at least one of searching and / or viewing, modifying, deleting, and / or restoring knowledge.

3. The system according to claim 1 or 2, characterized in that, Construct a heterogeneous graph neural network, and generate the original knowledge data input by the source and the recorded spatial locations by using the received knowledge. The method for constructing an attention mechanism based on the heterogeneous graph neural network includes: S1 Establish a geographical map, mark the spatial location as a node of the heterogeneous graph neural network on the geographical map, and construct a heterogeneous graph network based on the real spatial location. S2 Construct a knowledge heterogeneous graph network, a device heterogeneous graph network, and a heterogeneous behavior network based on the nodes, and define network nodes and the edges between the nodes. S3 Construct multiple types of meta-paths in the networks in S2 respectively, and define neighbor nodes. S4 First establish a node attention mechanism according to the multiple types of meta-paths, then construct a semantic attention mechanism, and finally construct a sub-graph attention mechanism.

4. The system according to claim 3, characterized in that, Retrieve the knowledge context based on the newly input knowledge data and / or the recorded knowledge operation behaviors of the user, and give the analysis result of the attention mechanism based on the knowledge context. The specific method is: S5 Input the knowledge into the trained AM to obtain the predicted target classification and / or record the relevant knowledge operation behaviors. S6 In the predicted target classification related to the knowledge operation behaviors, draw the knowledge context according to the corresponding spatial location and generation time of the knowledge, and retrieve the drawn knowledge context. S7 Mark the nodes and / or sub-graphs corresponding to the newly input knowledge data in the drawn knowledge context.

5. The system according to claim 4, characterized in that, The comprehensive analysis includes statistically analyzing at least one of the following types of distributions: The first distribution of the knowledge field corresponding to the target classification related to the knowledge operation behaviors corresponding to the knowledge data on the geographical map, the second distribution in the visualized knowledge heterogeneous graph network, device heterogeneous graph network, and heterogeneous behavior network, and the crossing distribution. And statistically analyze the knowledge context in the selected types of distributions, and the first distribution and the second distribution are displayed in the second visualization area.

6. The system according to claim 4 or 5, characterized in that, The method for knowledge evaluation includes: obtaining the spatial locations and times of the user and the users being followed, finding the nodes containing at least one small pool according to the meta-path being followed, calculating through the attention mechanism to draw a node correlation degree distribution map, and displaying it in the second visualization area.

7. The system according to claim 6, wherein The method for block partitioning includes: performing segmented clustering on the normalization coefficients of the sub-graph clustering calculated according to the attention mechanism, and dividing the sub-graphs into multiple blocks according to the segments.

Images