High-value information mining method based on expert thinking chain large model agent

Abstract

The invention discloses a high-value information mining method based on an expert thinking chain large model agent, which comprises the following steps of: inputting a name and a summary of a task, extracting entity and event information keywords in the task through a large language model, and performing text vectorization processing on the keywords and an expert thinking chain template; the cosine distance similarity between the keyword and the expert thinking chain template vector is calculated, and the expert thinking chain template with the highest similarity is obtained through matching; calling a search engine to search entity and event information keywords extracted from the task name and the summary, associating the keywords and searching related contents by utilizing a large model and the search engine aiming at each information keyword, and obtaining and recording a title, a publishing mechanism, publishing time, contents and a website of the searched webpage; extracting recorded titles, publishing structures and contents to obtain an entity, relationship and event triple, and organizing into a public security domain knowledge system for storage; performing intelligent analysis on the constructed public security domain knowledge system according to the expert thinking chain template to obtain a visual statistical chart and a structured analysis conclusion; and according to the visual statistical chart and the structured analysis conclusion, an analysis report is written and generated, and a final information mining analysis report is obtained. According to the method, a large number of complex data sets can be quickly identified, classified and analyzed, so that the process of information extraction and knowledge discovery is accelerated.

Description

Technical Field

[0001] The invention relates to natural language processing, and more particularly to a high-value information mining method based on an expert thinking chain large model intelligent body. Background Art

[0002] With the rapid development of information technology, the information field is facing the challenges of massive data processing and high-value information mining. High-value information mining based on large language models is a hot research direction in the field of natural language processing. The large language model agent is an artificial intelligence model trained based on a large model dataset. It understands and generates natural language by learning the statistical laws of language, and can capture the complexity and diversity of information in the public security field. Although the large language model agent has achieved unprecedented results in many tasks, its working mechanism lacks transparency and explainability, which can lead to the problem of hallucination.

[0003] Expert thinking chain refers to a series of logical reasoning and information mining steps used by experts when solving problems. It can simulate the expert thinking process through structured knowledge representation and reasoning rules, thereby improving the accuracy and reliability of intelligent agents in solving problems. The expert thinking chain method can significantly improve the performance of large language models in common sense question answering, logical reasoning and other tasks.

[0004] Therefore, using information-oriented methods to guide intelligent agents to conduct logical reasoning and information mining in the public security field can improve the autonomous analysis and decision-making capabilities of intelligent agents and provide new solutions for solving complex analysis problems. In information mining tasks, there are many problems such as the lack of effectiveness of intelligent information mining methods based on expert experience, the inability to solidify and trace expert experience in information mining tasks, and the difficulty in effectively reusing high-value information mining behaviors. This has become a pain point in the research of information mining. This is mainly because at the beginning of the design of the existing information system, expert experience knowledge, analysis patterns, cognitive thinking, etc. were not included in the basic design. By combining the large language model with the expert thinking chain for high-value information mining, the problem of low accuracy and effectiveness of the large language model in information mining can be effectively solved. At the same time, based on the guidance of the expert thinking chain in the public security field, the quality and efficiency of information intelligent processing can be improved by utilizing the high-value information mining capabilities and excellent natural language processing capabilities of "AI Agent + large model". Summary of the invention

[0005] The purpose of the present invention is to propose a high-value information mining method based on an expert thinking chain large model intelligent agent for the public security field, so as to solve the problems such as the insufficient effectiveness of intelligent information mining and processing methods based on expert experience.

[0006] The technical solution to achieve the purpose of the present invention is: a high-value information mining method based on an expert thinking chain large model intelligent agent, comprising the following steps:

[0007] Step 1: Input the name and overview of the task, extract entities including people, time, place, and event information keywords through the large language model, and perform text vectorization on the keywords and the expert thinking chain template. By calculating the cosine distance similarity between the keywords and the expert thinking chain template vectors, the expert thinking chain template with the highest similarity is matched;

[0008] Step 2: Call the search engine to search for the entity and event information keywords extracted from the task name and overview. For each information keyword, use the big model and search engine to associate keywords and search for related content. Obtain and record the title, publishing agency, publishing time, content, and URL of the searched web page. Extract the recorded title, publishing structure, and content to obtain entity, relationship, and event triples, which are organized into a public safety domain knowledge system for storage.

[0009] Step 3: Based on the expert thinking chain template, intelligently analyze the constructed public safety knowledge system to obtain visual statistical charts and structured analysis conclusions;

[0010] Step 4: Write and generate an analysis report based on the visualized statistical charts and structured analysis conclusions to obtain the final information mining analysis report.

[0011] Furthermore, in step 1, the name and overview of the task are input, and entities are extracted through the large language model, including people, time, place, and event information keywords, and the keywords and the expert thinking chain template are processed by text vectorization. By calculating the cosine distance similarity between the keywords and the expert thinking chain template vectors, the expert thinking chain template with the highest similarity is matched, where:

[0012] The expert thinking chain template includes two categories: man-made events and natural disasters, among which:

[0013] The template for man-made events is as follows: (1) Please analyze the cause of the event; (2) Please analyze whether the event is a domestic event or an international event; (3) Please analyze the impact of public cognition in the event; (4) Please analyze the strategy and guiding ideology of the public security event; (5) Please analyze the hard support of cognitive confrontation in the event; (6) Please analyze the historical background and characteristics of the times in the event;

[0014] The natural disaster template is as follows: (1) Please analyze the cause of the incident; (2) Please analyze the international relations and diplomatic factors of the incident; (3) Please analyze the impact of public cognition in the incident; (4) Please analyze the strategy and guiding ideology of the public safety incident, the disaster emergency plan, the strategies and response measures of the government and rescue organizations; (5) Please analyze the hard support of cognitive confrontation in the incident; (6) Please analyze the historical background and characteristics of the times in the incident.

[0015] Further, in step 3, based on the expert thinking chain template, the constructed public safety knowledge system is intelligently analyzed to obtain visual statistical charts and structured analysis conclusions, among which:

[0016] Intelligent analysis includes two parts: visual data and structured analysis. In the visual data part, a bar chart is made for data containing numerical sizes of different categories or time points, a line chart is made for data containing changes over time, a radar chart is made for data containing the relative sizes of different variables, a trajectory chart is made for data containing state or position changes at different time points, a timeline is made for data containing time sequence and the sequence of events, and a pie chart is made for data containing the proportion of each part to the whole. Six types of visual statistical charts are made, including bar charts, line charts, radar charts, trajectory charts, timelines, and pie charts. In the structured analysis part, the starburst method, startup list method, or quadrant processing method is called for analysis to obtain the conclusions of the structured analysis method.

[0017] Furthermore, it also includes recording operation logs. The logs include two parts, behavioral actions and behavioral content. Behavioral actions include searching, analyzing, and writing. Behavioral content includes search content, search results, statistical chart data and structured analysis conclusions obtained from analysis, and written analysis reports.

[0018] Furthermore, it also includes adding behavioral actions that obtain high-value information to the expert thinking chain template.

[0019] A method for high-value information mining based on a large model intelligent agent of an expert thinking chain is implemented to realize high-value information mining based on a large model intelligent agent of an expert thinking chain.

[0020] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for mining high-value information based on a large model intelligent agent of an expert thinking chain is implemented to achieve mining high-value information based on a large model intelligent agent of an expert thinking chain.

[0021] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the high-value information mining method based on the expert thinking chain large model intelligent agent, thereby realizing high-value information mining based on the expert thinking chain large model intelligent agent.

[0022] Compared with the prior art, the present invention has the following significant advantages: 1) It utilizes the expert thinking chain to guide the large-model intelligent body to conduct high-value information mining in the public security field, providing an efficient technical means for the information processing field and significantly improving the work efficiency of information processing personnel. Through the algorithm-driven and data processing capabilities of multiple intelligent agents, a large number of complex data sets can be quickly identified, classified and parsed, thereby accelerating the process of information extraction and knowledge discovery; 2) By implementing an interactive operation mechanism and combining the auxiliary report generation function of the intelligent body, the experience of experts can be effectively solidified and traced back. This method not only promotes the reuse of high-value information mining behaviors, but also enhances the knowledge base of information mining through intelligent recording and feedback loops, providing rich reference resources and decision-making support for future information mining. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a structural diagram of a high-value information mining method based on the expert thinking chain large model intelligent agent;

[0024] Figure 2 This is an example diagram of the expert thinking chain template;

[0025] Figure 3 It is a schematic diagram of the solidification and iteration of the expert thinking chain. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0027] The present invention is directed to the field of public safety, where public safety refers to accidents or emergencies that may have a significant impact on the life, health, property or environment of the public. High-value information mining based on the expert thinking chain large model intelligent agent involves a central control system and a large model intelligent agent (hereinafter referred to as the intelligent agent). The central control system is mainly used for interaction with users and for assigning tasks to corresponding intelligent agents according to the decomposed tasks. The intelligent agent part includes a search intelligent agent, an extraction intelligent agent, an analysis intelligent agent, and a writing intelligent agent. Each type of intelligent agent includes four core components: tools, memory, planning, and actions. Tools refer to algorithms and models used to process and analyze information data, memory refers to the ability of the intelligent agent to store information, planning is the action plan formulated by the intelligent agent according to the target and the current environmental status, and action is the response of the intelligent agent to the information mining task. The search agent tool mainly includes search tools such as search engines, whose main function is to search and associate keywords for the task name and keywords in the overview entered by the user; the extraction agent tool mainly includes extraction tools such as extraction big models, whose main function is to extract based on the task name and overview entered by the user and the text information obtained by the search agent, and generate structured information such as entities, relationships, and events by using the extraction big model; the analysis agent tool includes a code compiler, whose main function is to analyze and process the text information obtained by the search agent and the structured information extracted by the extraction agent, mainly including generating statistics Charts and analysis texts contain numerical information such as comparison and change. Code compilers are used to generate statistical charts. Corresponding visual analysis interfaces are generated based on common structured analysis methods in the field of information mining (such as starburst method, startup list method, quadrant processing method, etc.), and corresponding analysis results are generated based on structured analysis methods. The writing agent tool mainly includes writing a large model. Its main function is to use the language generation capability of the writing large model to intelligently generate and write analysis reports based on search agents, extraction agents, and analysis agents, search information generated by search agents, extracted structured knowledge, statistical charts generated by analysis agents, and conclusions of structured analysis methods. Among them, determining that the high-value information is the result of a certain search, analysis action or behavior in the process of information mining is helpful for the final generated analysis report, or that part of the generated analysis report is obtained from the results of the search, analysis, etc.

[0028] like Figure 1 As shown, a high-value information mining method based on an expert thinking chain large model intelligent agent includes the following steps:

[0029] Step 1: Match the task with the expert thinking chain:

[0030] First, the user inputs the "task name and overview", which consists of text. Based on the text content of the task name and overview, the large language model is used to extract the entity and event information keywords such as people, time, and place, and the keywords and expert thinking chain template are processed by text vectorization. The cosine distance similarity between the keyword vector A and the expert thinking chain template vector B is calculated to match them. The cosine distance similarity θ is calculated by the dot product and the vector length. A i and B i Represent the components of vectors A and B respectively. The calculation method of cosine similarity is as follows:

[0031]

[0032] Sort the cosine similarity from high to low, and get the expert thinking chain template with the highest cosine similarity as the matching result of the expert thinking chain template. Figure 2 As shown in the figure, it is constructed manually based on experience. The artificially constructed expert thinking chain template for the public security field includes two categories: man-made events (such as criminal acts, etc.) and natural disasters (such as earthquakes, floods, hurricanes, etc.). The man-made event template is as follows: (1) Please analyze the cause of the event, such as safety hazards, illegal operations, social emotions, etc.; (2) Please analyze whether the event is a domestic event or an international event. If it is a domestic event, please consider the impact of domestic politics, economy, society, etc. If it is an international event, please consider international relations and diplomatic factors, such as tensions between countries, international intervention and support; (3) Please analyze the impact of public cognition in the event, such as media reports, public opinion, social media mobilization and propaganda wars; (4) Please analyze the strategy and guiding ideology of public security events, such as the goals, strategies and ideological basis of the organizers of the event; (5) Please analyze the hard support of cognitive confrontation in the event, such as media tools, network platforms, and propaganda channels; (6) Please analyze the historical background and characteristics of the times in the event, such as similar events in history, long-accumulated social contradictions and the background of the times. The natural disaster template is as follows: (1) Please analyze the cause of the incident, such as natural factors, such as geological activities, meteorological changes, environmental degradation, etc.; (2) Please analyze the international relations and diplomatic factors of the incident, such as international rescue cooperation, cross-border resource mobilization, and assistance and support from international organizations; (3) Please analyze the impact of public cognition in the incident, such as media reports, information dissemination on social media, and public reaction; (4) Please analyze the strategy and guiding ideology of public safety incidents, disaster emergency plans, strategies and response measures of governments and rescue organizations; (5) Please analyze the hard support of cognitive confrontation in the incident, emergency information systems, media publicity, and information release channels; (6) Please analyze the historical background and characteristics of the times in the incident, such as the lessons learned from similar disasters in the past, the current trend of climate change and its impact.

[0033] After obtaining the matched expert thinking chain template, the central control system will decompose the task and assign the search, extraction, analysis, and writing tasks to the corresponding search, extraction, analysis, and writing agents to complete the search, extraction, analysis, and writing tasks respectively.

[0034] Step 2: Construction of search agent and knowledge system:

[0035] According to the expert thinking chain content matched in step 1, the search agent calls the search engine to search for the information keywords extracted from the task name and overview. Using the large model and the search engine's associative keyword capabilities, the large model and the search engine are used to associate 10 keywords for each information keyword and search for related content, and obtain and record the title, publishing agency, publishing time, content, URL and other attribute information of the searched news, information, announcements and other web pages. The recorded title, publishing structure, and content attribute text information are passed to the extraction agent for extraction to obtain information such as entities, relationships, and events. The extraction process relies on the extraction function of the existing large language model to convert unstructured text information into structured information such as triples. In the information extraction stage, the extraction agent uses natural language processing technology to extract structured information from text-formatted data. This information may include geographic information, time series information, statistical data such as population changes, etc., which can be used as a data source for the analysis agent. The extracted key information of entities, relationships, and events is synchronously updated to the knowledge system in the public safety field. The structured information can assist in subsequent information mining and reasoning, facilitate the construction of relationships between different entities such as people and things, and obtain metaphorical clues from the structured knowledge system.

[0036] Step 3: Analyze the construction of intelligent agents and analyze structured information:

[0037] When the analytical agent receives the central control instruction, it calls the structured analysis method to mine information based on the expert thinking chain information and the structured information extracted from the agent. Among them, structured analysis mainly acts on the reasoning process, strengthens the structural norms of the analysis process, ensures an objective and neutral attitude to ensure the consistency of the evaluation criteria, avoids premature exclusion or affirmation of a hypothesis or a clue, and thus reduces information errors. Applying structured analysis to information mining can make related work more in-depth and refined, and the results are more accurate and reliable. After the data is structured, the analytical agent can perform in-depth information mining such as structured analysis based on the structured information. This may include statistical analysis, trend prediction, association rule mining, cluster analysis, etc. Through these information mining methods, the analytical agent can reveal the patterns and associations behind the data and provide support for decision-making.

[0038] The analytical agent organizes key information such as entities, relationships, and events into structured formats, such as tables, charts, and databases. This step usually involves data integration and association, unifying data from different sources and formats to form a complete data set, such as analyzing and comparing the trend of statistical data over time, generating year-on-year or month-on-month data, etc. Based on the statistical data results of geographic location changes, time series, and population changes extracted from the information, organize the geographic information, time series information, population changes, and other statistical data contained therein. Make a bar chart for data containing numerical values ​​of different categories or time points, a line chart for data containing changes over time, a radar chart for data containing the relative sizes of different variables, a trajectory chart for data containing changes in status or position at different time points, a timeline for data containing time sequence and the order of events, and a pie chart for data containing the proportion of each part to the whole. Make six types of visual chart data: bar chart, line chart, radar chart, trajectory chart, timeline, and pie chart.

[0039] In addition, the analysis agent uses structured analysis methods in the field of information mining, such as the starburst method, the startup list method, and the quadrant processing method, based on the content obtained by the search agent to obtain conclusions from the structured analysis method.

[0040] Step 4: Write the agent construction and analysis result generation:

[0041] The writing agent generates an analysis report based on the previous structured analysis and visual analysis results to assist decision making. The analysis report generated by the writing agent not only relies on the analysis charts generated by the analysis agent, but also relies on the conclusions inferred by the analysis agent based on structured analysis methods such as trend prediction, association rule mining, and cluster analysis. This information and conclusions can help assist decision making. The suggestions and strategies in the report are based on in-depth data mining and logical reasoning, and are designed to help decision makers more clearly understand the nature of the problem, evaluate different solutions, and choose the best path of action when facing complex problems.

[0042] In addition, the entire information mining process can also be manually intervened to perform supplementary interactive search, analysis and other information mining operations. The present invention simultaneously records the operation logs of user interactive mining and large-model intelligent agent information mining. The logs include two parts, behavioral actions and behavioral content. Behavioral actions include search, analysis, writing and other actions. Behavioral content includes the search content, search results (including web page title, publishing agency, publishing time, content, URL and other attribute information), statistical chart data and structured analysis conclusions obtained from the analysis, and written analysis reports.

[0043] Step 5: Solidify the iterative expert thinking chain template:

[0044] During the information mining process, the information mining process from step 1 to step 4 can be traced back to extract high-value information, and the behavioral actions for obtaining the high-value information can be added to the expert thinking chain template.

[0045] If the results of a search, analysis action or behavior in the information mining process are helpful for the final generated analysis report, or part of the generated analysis report is obtained from the search, analysis and other results, then the information is considered to be high-value. Based on the behavior content in the behavior log of the intelligent agent and manual interactive operation, high-value information is extracted, and the behavior action that obtains the high-value information is added to the expert thinking chain template, such as Figure 3 As shown in the figure, the solidification iteration is carried out, and each time information mining is carried out, the expert thinking chain template can be iterated once.

[0046] Example

[0047] In order to verify the effectiveness of the scheme of the present invention, the following experiment was carried out.

[0048] In the first step, enter the task name as "In-depth analysis of the '12.10' fire incident in City A" and the task overview as "At around 15:28 on December 10, 2023, a fire broke out at Company B in City A, killing 6 people, burning an area of ​​more than 5,000 square meters, and causing direct economic losses of more than 20 million yuan. This task aims to collect clues and information to uncover the real reasons behind the incident."

[0049] Match the expert thinking chain template. Through the task name and overview, you can match the thinking chain template of "man-made incidents" in the public safety field. Generate a synthetic expert thinking chain: "(1) Please analyze the cause of the '12.10' fire incident in City A, and consider the root causes of the incident, such as safety hazards, illegal operations, social emotions, etc.; (2) Please analyze whether the '12.10' fire incident in City A is a domestic event or an international event. If it is a domestic event, please consider the impact of domestic politics, economy, society, etc. If it is an international event, please consider international relations and diplomatic factors, such as tensions between countries, international intervention and support; (3) Please analyze the impact of public cognition on the '12.10' fire incident in City A, such as media reports, public opinion, social media mobilization and propaganda war; (4) Please analyze the strategy and guiding ideology of the '12.10' fire incident in City A, such as the goals, strategies and ideological basis of the organizers of the event; (5) Please analyze the hard support of cognitive confrontation in the '12.10' fire incident in City A, such as media tools, network platforms, and propaganda channels; (6) Please analyze the historical background and characteristics of the times in the '12.10' fire incident in City A, such as similar events in history, long-accumulated social contradictions and the background of the times.

[0050] In the second step, the search agent can extract key information such as "A City '12·10' Fire Incident", "15:28 on December 10, 2023", "A City", "Company B", "6 people died", "burned area of ​​more than 5,000 square meters", "direct economic losses exceeded 20 million yuan" from the task name and task overview, and can call the search engine to check relevant information based on key information and expert thinking chain. Search for content related to the incident based on the keywords extracted from the task name "A City '12·10' Fire Incident" and the task overview "15:28 on December 10, 2023", "Company B", "6 people died", etc. In addition, the extraction agent can also extract triple information based on the searched text content. For example, from the text "At 15:28:30 on the same day, a flash appeared in the area near the north window on the west side of the No. 5 flower twisting machine in a workshop, and then a fire was visible in the area north of the No. 5 flower twisting machine. The fire quickly spread to the warehouse through the two-story illegal building connected to the north. The high-temperature toxic and harmful smoke generated by the fire filled the building and the evacuation stairwell. Company B failed to organize the evacuation of the employees of the fourth-floor workshop in the warehouse in time, and the outdoor evacuation staircase safety exit was locked, resulting in the death of 6 employees." Extract triple information such as "Flash appeared in the area near the north window on the west side of the No. 5 flower twisting machine in a workshop."

[0051] Based on the extracted triple information, a relationship network is established, and the new triple knowledge is integrated into the knowledge system to assist in understanding and analyzing the location and time of the '12·10' fire incident in City A.

[0052] In the third step, the analysis agent uses common structured analysis methods such as the "starburst method" to analyze the event elements such as when (15:28 on December 10, 2023), where (Company B in City A), what (fire occurred), and how (6 people died, the burned area was more than 5,000 square meters, and the direct economic loss exceeded 20 million yuan). Based on the thought chain prompt information and the extracted event time, event location, burned area, number of casualties and other data information, the timeline of the '12.10' fire event in City A is generated. In addition, based on all the fire events in City A in the past three years that have been searched, the cause of the fire, the process of fire spread, the number of casualties, the burned area and other data can be analyzed, and statistical charts such as "bar chart of casualties in fire events in City A in the past three years", "timeline of the spread process of fire events in City A in the past three years", and "radar chart of fire causes in fire events in City A in the past three years" can be produced.

[0053] The fourth step is to write an analysis report on the "In-depth Analysis of the '12·10' Fire Incident in City A" based on the previous structured analysis and visualization analysis results.

[0054] The fifth step is to integrate the thought chain generated by the analysis agent during the analysis process with the thought chain generated interactively by users during the analysis process to solidify and iterate. Generate a new "public safety" type expert thinking chain template: "(1) Please analyze the cause of the incident, such as safety hazards, illegal operations, social emotions, etc.; (2) Please analyze whether the incident is a domestic incident or an international incident. If it is a domestic incident, please consider the impact of domestic politics, economy, society and other aspects. If it is an international incident, please consider international relations and diplomatic factors, such as tensions between countries, international intervention and support; (3) Please analyze the impact of public cognition in the incident, such as media reports, public opinion, social media mobilization and propaganda war; (4) Please analyze the strategy and guiding ideology of the public safety incident, such as the goals, strategies and ideological basis of the organizer of the incident; (5) Please analyze the hard support of cognitive confrontation in the incident, such as media tools, network platforms, and propaganda channels; (6) Please analyze the historical background and characteristics of the times in the incident, such as similar events in history, long-accumulated social contradictions and the background of the times; (7) Please use the starburst method and other event analysis related structured analysis methods to analyze the event elements such as when, who, where, what, and how in the process of the incident." This will help to mine information and assist decision-making for subsequent similar events.

[0055] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0056] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A high-value information mining method based on an expert thinking chain large model agent, characterized in that: The steps include: Step 1: Input the name and overview of the task, extract entities including people, time, place, and event information keywords through the large language model, and perform text vectorization on the keywords and the expert thinking chain template. By calculating the cosine distance similarity between the keywords and the expert thinking chain template vectors, the expert thinking chain template with the highest similarity is matched; Step 2: Call the search engine to search for the entity and event information keywords extracted from the task name and overview. For each information keyword, use the big model and search engine to associate keywords and search for related content. Obtain and record the title, publishing agency, publishing time, content, and URL of the searched web page. Extract the recorded title, publishing structure, and content to obtain entity, relationship, and event triples, which are organized into a public safety domain knowledge system for storage. Step 3: Based on the expert thinking chain template, intelligently analyze the constructed public safety knowledge system to obtain visual statistical charts and structured analysis conclusions; Step 4: Write and generate an analysis report based on the visualized statistical charts and structured analysis conclusions to obtain the final information mining analysis report.

2. The high-value information mining method based on the expert thinking chain large model intelligent agent according to claim 1 is characterized in that: Step 1: Input the name and overview of the task, extract entities including people, time, place, and event information keywords through the large language model, and perform text vectorization on the keywords and the expert thinking chain template. By calculating the cosine distance similarity between the keywords and the expert thinking chain template vectors, the expert thinking chain template with the highest similarity is matched, where: The expert thinking chain template includes two categories: man-made events and natural disasters, among which: The template for man-made events is as follows: (1) Please analyze the cause of the event; (2) Please analyze whether the event is a domestic event or an international event; (3) Please analyze the impact of public cognition in the event; (4) Please analyze the strategy and guiding ideology of the public security event; (5) Please analyze the hard support of cognitive confrontation in the event; (6) Please analyze the historical background and characteristics of the times in the event; The natural disaster template is as follows: (1) Please analyze the cause of the incident; (2) Please analyze the international relations and diplomatic factors of the incident; (3) Please analyze the impact of public cognition in the incident; (4) Please analyze the strategy and guiding ideology of the public safety incident, the disaster emergency plan, the strategies and response measures of the government and rescue organizations; (5) Please analyze the hard support of cognitive confrontation in the incident; (6) Please analyze the historical background and characteristics of the times in the incident.

3. The high-value information mining method based on the expert thinking chain large model agent according to claim 1 is characterized in that: Step 3: Based on the expert thinking chain template, intelligently analyze the constructed public safety knowledge system to obtain visual statistical charts and structured analysis conclusions, including: Intelligent analysis includes two parts: visual data and structured analysis. In the visual data part, a bar chart is made for data containing numerical sizes of different categories or time points, a line chart is made for data containing changes over time, a radar chart is made for data containing the relative sizes of different variables, a trajectory chart is made for data containing state or position changes at different time points, a timeline is made for data containing time sequence and the sequence of events, and a pie chart is made for data containing the proportion of each part to the whole. Six types of visual statistical charts are made, including bar charts, line charts, radar charts, trajectory charts, timelines, and pie charts. In the structured analysis part, the starburst method, startup list method, or quadrant processing method is called for analysis to obtain the conclusions of the structured analysis method.

4. The high-value information mining method based on the expert thinking chain large model agent according to claim 1 is characterized in that: It also includes recording operation logs. The logs include two parts: behavioral actions and behavioral content. Behavioral actions include searching, analyzing, and writing. Behavioral content includes the search content, search results, statistical chart data and structured analysis conclusions obtained from the analysis, and written analysis reports.

5. The high-value information mining method based on the expert thinking chain large model agent according to claim 1 is characterized in that: It also includes adding behavioral actions that obtain high-value information to the expert thinking chain template.

6. A high-value information mining method based on an expert thinking chain large model agent, characterized in that: Implement the high-value information mining method based on the expert thinking chain large model intelligent agent as described in any one of claims 1-5 to realize high-value information mining based on the expert thinking chain large model intelligent agent.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, the high-value information mining method based on the expert thinking chain large model intelligent agent as described in any one of claims 1 to 5 is implemented to realize high-value information mining based on the expert thinking chain large model intelligent agent.

8. A computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the high-value information mining method based on the expert thinking chain large model intelligent agent as described in any one of claims 1 to 5 is implemented to realize high-value information mining based on the expert thinking chain large model intelligent agent.

Images