Intelligent association organization and review analysis and summary method based on multi-dimensional historical data

Abstract

The application discloses a kind of based on multidimensional historical data intelligent association organization and review analysis summary method, comprising the following steps: constructing comprehensive review dataset;Comprehensive review dataset is analyzed from multiple angles;Analysis process and its result are encapsulated as reusable review analysis component, and the uniform registration, discovery and calling mechanism of component is provided, different analysis components are selected and combined to form customized review analysis special subject product according to the needs of review task;Review analysis summary report is generated;The application realizes the free customization of data board display content.

Description

Technical Field

[0001] This invention relates to the field of retrospective analysis, and in particular to a method for intelligent association organization and retrospective analysis summarization based on multi-dimensional historical data. Background Technology

[0002] With the deepening of information technology construction, the accumulation of various historical data is becoming increasingly rich, covering multiple key areas such as key objectives, event handling, and task status. However, current historical data management still faces many challenges: data sources are scattered and storage methods are independent, forming "data silos" that make rapid integration and correlation analysis difficult; the inherent connections between data have not been effectively explored, resulting in a lack of systematicness and depth in retrospective analysis; comprehensive analysis methods for multi-dimensional information are relatively scarce, making it difficult to meet the analysis needs in complex scenarios; at the same time, data visualization methods are not uniform enough, and the ability to intuitively present information is insufficient, affecting the efficiency and effectiveness of retrospective analysis. To address these issues, it is urgent to construct a comprehensive, multi-dimensional retrospective analysis and summary method, focusing on breakthroughs in the following key areas: First, establishing unified historical data standards to achieve standardized data management and deep correlation; second, developing multi-dimensional information analysis tools to improve data mining and correlation analysis capabilities; third, designing unified visualization styles and intuitive presentation methods to enhance the expressive effect of analysis results; and fourth, constructing a retrospective analysis framework based on unified standards to provide customized scenario analysis support for various business areas. Summary of the Invention

[0003] Purpose of the invention: The purpose of this invention is to provide an intelligent association organization and retrospective analysis summary method based on multi-dimensional historical data. By integrating multi-dimensional historical data, conducting in-depth data mining analysis, customizing retrospective scenarios, performing various retrospective scenarios, and generating auxiliary analysis reports, this invention solves the problems existing in the background technology.

[0004] Technical Solution: The present invention provides an intelligent association organization and retrospective analysis method based on multi-dimensional historical data, comprising the following steps: Step 1: Through predefined data entity description specifications, multi-source historical data are uniformly published as standardized retrospective data entities. Based on knowledge graph technology, according to the retrospective objects and analysis objectives that users are interested in, the standardized retrospective data entities are associated across data sources to form a comprehensive retrospective dataset. Step 2: Conduct multi-faceted analysis of the comprehensive review dataset, including using deep learning models to uncover the objective patterns contained in the data, depicting the evolution of the review scenario over time, and making predictive judgments on future development trends; Step 3: Encapsulate the analysis process and its results into reusable debriefing analysis components, and provide a unified registration, discovery and invocation mechanism for the components. This allows for the selection and combination of different analysis components according to the needs of the debriefing task, forming customized debriefing analysis thematic products. Step 4: Under the unified review and broadcast control framework, each selected review and analysis component can synchronously respond to the control commands of the broadcast timeline, dynamically present the evolution of historical data over time and the corresponding analysis results, and automatically generate a review and analysis summary report based on the preset template.

[0005] Furthermore, in step 1, the predefined data entity description specification specifically defines the metadata model for each type of retrospective data entity, including basic data information, corresponding display component names, data lifecycle description on the timeline, and key data fields used for identification, filtering, and association; standardize the retrospective data entity types, including at least simple entities describing the state of a single object, complex entities describing multi-object interaction events, and analysis result entities describing the conclusions of manual or automated analysis.

[0006] Furthermore, in step 1, the data association based on knowledge graph technology is specifically as follows: relying on a unified data model and data exchange standard, and using an object-oriented data abstraction method, core elements and their interrelationships are extracted from multi-source heterogeneous business data, and a public data model that can fully describe key entities and their attributes is constructed to achieve a unified logical representation and on-demand organization of historical data from different sources and in different forms.

[0007] Furthermore, in step 2, the objective patterns contained in the data are mined using deep learning models as follows: analyzing the logistical and collaborative relationships between the targets of interest; analyzing the typical movement trajectories, hotspot activity areas and time periods of the targets; identifying the abnormal activities of the targets and their key changes; and comparing the resource investment intensity and capability differences of different stakeholders in specific time periods.

[0008] Furthermore, in step 2, the evolution of the review scenario over time is depicted, specifically: extracting time and geographic location information from text information related to the target, event, or task; arranging all text in chronological order based on time information and segmenting it according to the set time slices; calculating and extracting the activity themes reflected in the text subsets within each time slice, and identifying key time nodes; and generating an evolutionary path diagram describing the development of the event based on the changes in these themes and key nodes.

[0009] Furthermore, in step 2, the predictive assessment of future development trends specifically involves: using automatic classification technology to filter out symptoms related to the task or event of interest from multi-source information; merging and deduplicating symptom information from different sources using information fusion technology; semi-automatically extracting typical symptom event patterns using pattern mining technology; determining the alarm level of symptom events by combining expert knowledge; and establishing a probability assessment model from symptom events to final alarm events to assist in assessing possible future events and their early warning timing.

[0010] Furthermore, in step 3, the retrospective analysis components include: a map display component responsible for drawing spatial information that changes over time on the electronic map as needed; a chart display component responsible for displaying the analysis results of business data in the form of statistical charts; and all retrospective analysis components must follow unified development specifications and have the ability to receive external time synchronization messages.

[0011] Furthermore, the unified review and broadcast control framework broadcasts playback progress control messages to all running review and analysis components through a standardized time synchronization message communication mechanism. After receiving the message, each component autonomously prepares data for the corresponding historical moment, updates the displayed content, and recalculates the analysis results based on the time information contained in the message, thereby achieving precise synchronization between map display and various charts on the timeline.

[0012] Furthermore, in step 3, the customized debriefing analysis products are generated as follows: a visual arrangement tool is provided, which pre-integrates a variety of common display components, including heatmaps, mind maps, Gantt charts, radar charts, and scatter plots; users can freely customize data dashboards, including maps, charts, and analysis conclusions, by dragging and dropping, configuring these components, and associating them with registered debriefing data entities and analysis components, thereby generating thematic products that meet the needs of different debriefing scenarios.

[0013] Furthermore, step 4 is as follows: Predefine customized report templates for different types of debriefing scenarios. The templates specify the report structure, chapters, and analysis items that need to be filled in. During or after the debriefing broadcast, the system automatically extracts the completed pattern mining results, event progress analysis conclusions, and trend prediction results according to the template, and fills them into the corresponding positions in the report template, ultimately dynamically generating a debriefing analysis summary report.

[0014] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: (1) It can publish multi-source heterogeneous data based on a unified review data entity interface specification, realize unified data aggregation, and support unified historical data query, filtering and compilation; (2) It can flexibly organize various review data entities as needed based on knowledge graphs, organize and generate comprehensive review data of the entities of interest, and support review analysis and broadcast based on entities; (3) It provides the ability to analyze the inherent laws of data based on historical data, analyze the evolution process of review scenarios, and predict future development trends, and provides review analysis capabilities at all times and in all dimensions on the basis of intuitively displaying the development process of events and situations; (4) It provides review display components as a unified carrier for review analysis software. Based on a unified review analysis component integration mechanism, the review display content can be expanded at any time; (5) It can customize and arrange the review scenario display content to meet the multi-dimensional and multi-style display needs of users; (6) It can realize the synchronization of the display on the graph and the display content of the components based on a unified review broadcast framework, and support the generation of review analysis reports, which is convenient for the review analysis summary after the event. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the present invention; Figure 2 This is a flowchart of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0017] like Figures 1-2 As shown in the figure, this embodiment of the invention provides a method for intelligent association organization and retrospective analysis and summarization based on multi-dimensional historical data, including the following steps: Step 1: Flexible organization of review data entities. Provide interface specifications for review data entities and flexibly organize various review data entities as needed to generate comprehensive review data for entities of interest to the user. This includes the following steps: (11) Review data aggregation, based on the review data entity interface specification, publish multi-source heterogeneous data.

[0018] The data specification model for review data entities defines the basic descriptions of various business review data, including basic data information, processing and display component names, data lifecycle descriptions, and key data fields. By publishing metadata interfaces, the situation review framework can identify and integrate review data.

[0019] Metadata types are mainly divided into simple entities, complex entities, and analysis results. Simple entity data includes historical entity movement information, basic entity information, and entity operational status, while complex entity data includes event handling information and related facilities, as well as basic event information. Analysis result data includes the event process, resource damage, and so on.

[0020] Data field descriptions are the data subject field attributes that each business needs to provide, including filter identifiers, tag information, etc. (a) Filter identifier. Describes the data fields that support filtering and their enumerated filter items. For example, a certain entity data can be filtered by level, and the specific enumerated filter items are "Level 1, Level 2, Level 3, Level 4".

[0021] (b) Tag Information. This describes the data fields labeled with tags, which generally include the unique identifier of the data entity, the entity location, the entity name, and the data time. It supports the extraction of various types of information by time for review and analysis within the situation review framework.

[0022] (12) Each business publishes business data resources in accordance with the retrospective data entity interface specification. The retrospective framework uniformly connects to the data entity according to the metadata description, performs historical data query, filtering and compilation, and distributes it to each business display component during the broadcast.

[0023] (13) Based on knowledge graphs, relying on unified data models and exchange standards, and adopting the design concept of object-oriented data models, we abstract and organize multi-source heterogeneous business data to form a standardized public data model that aims to fully describe key elements and important concepts, thereby realizing the digitization of the entities of concern for review.

[0024] Step 2: Review and analyze the data processing, which involves analyzing the inherent patterns in the data, analyzing the evolution of the review scenario, and predicting future trends. This includes the following steps: (21) Analyze the inherent patterns in the data. For the movement of the entities of interest, it can analyze the target correlation, including logistics, coordination, etc.; it can analyze the typical movement trajectory of the target; it can analyze the hot activity areas of the target; it can analyze the hot activity periods of the target; it can analyze the abnormal activities and key changes of the target.

[0025] (22) Analyze and review the evolution process of the scenario, extract the time and geographical information related to the target, event and task, segment the text of the open source information, and construct the word frequency matrix corresponding to the text; arrange the time description text set according to the time information in the document-spatiotemporal feature set in chronological order, divide the text set according to time slices, and generate a time slice division scheme; calculate and extract the activity themes contained in the text subsets in each time slice divided by time slices, analyze the key time points, and form the evolution path of the event accordingly; visualize the evolution path of time and generate a development context map.

[0026] (23) Predict future development trends. Based on open source information and text information, use SVM-based information automatic classification technology to automatically filter and focus on target situations and events related to tasks and events. Through parallel fusion technology, merge events from different sources. Use target synergy law mining technology to semi-automatically extract related symptom events. Combine big data analysis technology and expert knowledge to determine the alarm level of symptom events. Based on the sorted symptom events and their alarm parameters, determine the early warning coefficient and time effect between symptom events, etc., and establish a probability model from symptom events to early warning events to assist in judging and predicting trends.

[0027] Step 3: Customize the retrospective analysis components. Provide a unified registration, publishing, and invocation mechanism for all retrospective analysis components, supporting unified management of data access, service access, and result display for various analysis components. It also provides multiple general-purpose display components such as heatmaps, mind maps, Gantt charts, radar charts, and scatter plots. A thematic retrospective product organization and display engine is provided, utilizing chart component management and visualization tools to achieve free customization of the retrospective data dashboard display content, including chart content and retrospective analysis display components. This includes the following steps: (31) Provide development specifications for the retrospective analysis components. The retrospective display components are divided into retrospective map display components and retrospective chart display components. The retrospective map display components are responsible for drawing and displaying information on the map as needed. The retrospective chart display components serve as the display carrier for retrospective data processing and analysis, and are responsible for displaying business statistical analysis information as needed. Each business implements the display components according to the specifications and integrates them into the retrospective product customization and retrospective product broadcasting functions.

[0028] (32) Provide a time synchronization specification for the review analysis component. Based on the message specification for time synchronization of the display component, each service receives messages and controls the corresponding display logic through a unified communication mechanism in accordance with the standard. During the review product broadcast stage, each display component dynamically changes the displayed content according to the playback status of the review broadcast timeline.

[0029] (33) Provides a variety of general display components such as heatmaps, mind maps, Gantt charts, radar charts, and scatter plots to support the display of general review data. Heatmaps can provide statistical display capabilities of target activity heat areas; mind maps can provide event association display capabilities; Gantt charts can provide the ability to schedule, progress, and dependencies of events and tasks; radar charts can provide the ability to compare performance in various dimensions; and scatter plots can provide the ability to analyze data relationships.

[0030] (34) Review product customization: Each business unit must release component resources in accordance with the review display component specifications to support the customized generation of review status diagram products. Based on the released component resources, review scenario products are customized and orchestrated as needed.

[0031] Step 4: Debriefing Scenario Analysis and Broadcast. This provides a framework for debriefing broadcasts, supporting the preparation, dynamic presentation, and analysis of historical business data, and complementing the overall debriefing product broadcast. A debriefing analysis report template is provided to assist in generating the report. This includes the following steps: (41) Review the product broadcast in the scenario. Based on the review broadcast framework, each review map display component and review chart display component receives the playback progress control message according to the interface requirements, and responds in real time to the control message of the broadcast time progress bar to process historical business data preparation, changes, presentation and other processing.

[0032] (42) Generate a debriefing analysis report. Customize the debriefing analysis report template according to the needs of the debriefing scenario. Dynamically generate a debriefing analysis summary report based on actual data pattern mining, event progress analysis, trend prediction and other needs.

[0033] Experimental Case The following is a detailed description of the invention, using a retrospective analysis of a merchant ship's illegal approach to a certain waterway as an example: Step 1: Organize the review data entities flexibly. Based on the review data entity interface specification, publish review data for ships, aircraft, observation stations, regions, and meteorology. Based on key areas of focus, generate review entity information for ships, aircraft, observation stations, regions, and key events. Filter and query entity data that meets the spatiotemporal requirements of the review based on the region and time period of interest, including the platforms of each party executing tasks, the regions of interest, and key events. Based on open-source information, supplement and compile annotation information at key time points and locations, marking key events, emergencies, etc., for example, annotating at XX time: "XX merchant ship discovered at XX location by XX observation station," "XX merchant ship illegally approached XX region at XX time," "XX merchant ship disappeared at XX location XX kilometers away from XX place at XX time."

[0034] Step 2: Review and analyze the data. Based on the review entity data selected in Step 1, conduct review and analysis. Analyze the inherent patterns in the data, focusing on the frequency and duration of a merchant ship's recent activities in a key waterway, and the weather changes in the key area; analyze the evolution of the review scenario, including the entire process from the initial observation of the merchant ship approaching and leaving the waterway, including the ship's activities and tracking; predict future trends, analyze the inherent correlation between previous illegal approaching incidents and this incident, determine the alarm level for subsequent incidents, and determine the alarm coefficient and time effect between incidents, etc., to help generate suggestions for improvement in subsequent handling and response.

[0035] Step 3: Customize the debriefing analysis components. Extract the various debriefing analysis components developed in Step 2 according to the debriefing analysis component development specifications. Based on the needs of the debriefing scenario, mainly customize analysis components such as merchant ship activity analysis, regional meteorological statistics, event development context, key indicators, trend prediction, and map display to form a debriefing analysis special product.

[0036] Step 4: Debriefing and Analysis Broadcast. The broadcast features a customized debriefing and analysis product, with various analysis components dynamically changing their displayed content based on the broadcast timeline. Dynamic displays include map displays, event timelines, and key indicators; static analysis displays include merchant ship activity analysis, regional weather statistics, and trend forecasts.

[0037] As can be seen from the above, this invention can unify various review entities based on the review data entity interface specification, and can associate and generate entity review data to support review analysis and broadcasting based on activity entities. For various review entity data, it can conduct multi-dimensional review analysis, covering historical analysis, in-process analysis, and post-event prediction, enriching the display of review analysis topics. It can flexibly customize and generate review analysis topic products from various review analysis components to meet users' diverse and multi-dimensional analysis and display needs. It can unify the synchronous display of maps and component pages based on the broadcast timeline, compatible with dynamic and static display products, and can generate review analysis summary reports based on template customization, providing prior knowledge for subsequent event and task handling. This invention can meet the needs of review data organization, analysis, and broadcasting summaries for different scenarios and tasks.

Claims

1. A method for intelligent association organization and retrospective analysis based on multi-dimensional historical data, characterized in that, Includes the following steps: Step 1: Through predefined data entity description specifications, multi-source historical data are uniformly published as standardized retrospective data entities. Based on knowledge graph technology, according to the retrospective objects and analysis objectives that users are interested in, the standardized retrospective data entities are associated across data sources to form a comprehensive retrospective dataset. Step 2: Conduct multi-faceted analysis of the comprehensive review dataset, including using deep learning models to uncover the objective patterns contained in the data, depicting the evolution of the review scenario over time, and making predictive judgments on future development trends; Step 3: Encapsulate the analysis process and its results into reusable debriefing analysis components, and provide a unified registration, discovery and invocation mechanism for the components. This allows for the selection and combination of different analysis components according to the needs of the debriefing task, forming customized debriefing analysis thematic products. Step 4: Under the unified review and broadcast control framework, each selected review and analysis component can synchronously respond to the control commands of the broadcast timeline, dynamically present the evolution of historical data over time and the corresponding analysis results, and automatically generate a review and analysis summary report based on the preset template.

2. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 1, the predefined data entity description specification specifies the metadata model for each type of retrospective data entity, including basic data information, corresponding display component names, data lifecycle description on the timeline, and key data fields for identification, filtering, and association. Standardized retrospective data entity types should include at least simple entities describing the state of a single object, complex entities describing multi-object interaction events, and analytical outcome entities describing the conclusions of manual or automated analysis.

3. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 1, the data association based on knowledge graph technology is carried out as follows: relying on a unified data model and data exchange standard, and using an object-oriented data abstraction method, the core elements and their interrelationships are extracted from multi-source heterogeneous business data, and a public data model that can fully describe key entities and their attributes is constructed to achieve a unified logical representation and on-demand organization of historical data from different sources and in different forms.

4. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 2, the objective patterns hidden in the data are mined using deep learning models as follows: analyzing the logistical and collaborative relationships between the targets of interest; analyzing the typical movement trajectories, hotspot activity areas and time periods of the targets; identifying abnormal activities of the targets and their key changes; and comparing the resource investment intensity and capability differences of different stakeholders in specific time periods.

5. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 2, the evolution of the review scenario over time is depicted, specifically: extracting time and geographic location information from text information related to the target, event, or task; arranging all text in chronological order based on time information and segmenting it according to the set time slices; calculating and extracting the activity themes reflected in the text subsets within each time slice and identifying key time nodes; and generating an evolutionary path diagram describing the development of the event based on the changes in these themes and key nodes.

6. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 2, the predictive assessment of future development trends specifically involves: using automatic classification technology to filter out symptoms related to the task or event of interest from multi-source information; and using information fusion technology to merge and deduplicate symptoms from different sources. Typical symptom event patterns are extracted semi-automatically using pattern mining techniques; the alarm level of the symptom events is determined by combining expert knowledge. A probability assessment model for the transition from symptom events to final alarm events is established to assist in the assessment of potential future events and the timing of early warnings.

7. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 3, the retrospective analysis components include: a map display component responsible for drawing spatial information that changes over time on the electronic map as needed; a chart display component responsible for displaying the analysis results of business data in the form of statistical charts; and all retrospective analysis components must follow unified development specifications and have the ability to receive external time synchronization messages.

8. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 7, characterized in that, The unified review and broadcast control framework broadcasts playback progress control messages to all running review and analysis components through a standardized time synchronization message communication mechanism. After receiving the message, each component autonomously prepares data for the corresponding historical moment, updates the displayed content, and recalculates the analysis results based on the time information contained in the message, so as to achieve precise synchronization between map display and various charts on the timeline.

9. The intelligent association organization and retrospective analysis method based on multi-dimensional historical data according to claim 1, characterized in that, In step 3, the customized debriefing analysis products are generated as follows: a visual arrangement tool is provided, which pre-integrates a variety of common display components, including heatmaps, mind maps, Gantt charts, radar charts, and scatter plots. Users can freely customize data dashboards, including maps, charts, and analysis conclusions, by dragging and dropping, configuring these components, and associating them with registered debriefing data entities and analysis components, thereby generating thematic products that meet the needs of different debriefing scenarios.

10. The method for intelligent association organization and retrospective analysis based on multi-dimensional historical data according to claim 1, characterized in that, Step 4 is as follows: Predefine customized report templates for different types of debriefing scenarios. The templates specify the report structure, chapters, and analysis items that need to be filled in. During or after the debriefing broadcast, the system automatically extracts the completed pattern mining results, event progress analysis conclusions, and trend prediction results according to the template, and fills them into the corresponding positions in the report template, ultimately dynamically generating a debriefing analysis summary report.

Images