Problem-driven strategic intelligence assessment method
Through the problem-driven strategic intelligence evaluation method, the pre-trained large language model is used to construct research problem space and organizational intelligence data, which solves the problems of insufficient and biased information in strategic intelligence evaluation, and achieves an efficient and credible strategic intelligence evaluation and a closed loop of the intelligence production chain.
Patent Information
- Application Number
- CN202411389736.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-08
- Publication Date
- 2025-03-25
AI Technical Summary
There are two main challenges in current strategic intelligence assessment: one is that the target problem is highly comprehensive and abstract, and the lack of intelligence and knowledge mastered makes it difficult to form credible assessment conclusions; the other is that intelligence analysis and assessment personnel are easily affected by the professional limitations and habits of individuals or units, resulting in bias in assessment conclusions.
Using a problem-driven strategic intelligence evaluation method, we use pre-trained large language models to build a research problem space for strategic intelligence topics, organize relevant intelligence data, and use analysis tools and computing models to conduct multiple rounds of interactive intelligence analysis. This method not only realizes a complete closed loop of the intelligence analysis chain and the evidence chain, but also can recommend appropriate analysis strategies based on specific problem models.
Effectively guide intelligence analysts to solve strategic problems, comprehensively use intelligence data, analysis tools, and computing models, carry out argument accumulation and opinion refinement, enhance the credibility of strategic intelligence assessment, and strengthen the effective closed loop of the intelligence production chain.
Smart Images
Figure CN119692795A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligence analysis applications, and in particular to a problem-driven strategic intelligence assessment method. Background Art
[0002] The statements in this section merely provide background information related to the present disclosure and may not constitute prior art.
[0003] Strategic intelligence assessment work is to coordinate the intelligence resources of various strategic intelligence departments at the national level according to the needs of high-level decision-making, integrate and merge the valuable viewpoints of various strategic intelligence departments, conduct in-depth analysis, scientific judgment and prediction on fundamental, global, strategic, key and forward-looking issues, current and potential strategic threats, the strength, intentions and possible actions of major competitors, and provide important basis for serving national strategic decision-making, joint operations, national construction and management.
[0004] Strategic intelligence assessment covers a wide range of areas, including politics, military affairs, diplomacy, security, economy, science and technology. It covers physical space, network electromagnetic space, information space and cognitive space. It also covers international politics, intelligence science, information theory, game theory, logic, information technology and other fields. There are two challenges in current strategic intelligence assessment. First, the target issues of strategic intelligence assessment are often highly comprehensive and abstract, while the intelligence and knowledge available are extremely scarce and even difficult to use effectively. Therefore, the research, analysis and inference based on existing facts and logic are not sufficient to form a credible assessment conclusion. Second, when conducting strategic intelligence assessment, intelligence analysis and assessment personnel are often influenced by past experience and work habits, and are also limited by the professional limitations of individuals or units. They often make mistakes when sorting out analytical ideas, and sometimes find it difficult to objectively judge and evaluate key issues, resulting in biased assessment conclusions.
[0005] Problem-driven strategic intelligence assessment adopts the idea of starting from the end, first proposing several core issues of strategic intelligence topics, then continuously decomposing and linking the core issues layer by layer, and forming a research problem space of strategic intelligence topics in a variety of visual ways. Then, hypotheses are proposed and opinions are expressed for each problem, collaborative discussions are organized, and evidence is sought with the help of intelligence data and analytical tools. In this process, not only can the complete closed loop of the intelligence analysis chain and the evidence chain be achieved, but also the intelligence data requirements can be proposed for the missing evidence, thus strengthening the effective closed loop of the intelligence production chain. Problem-driven strategic intelligence assessment is not only a beneficial supplement to the traditional data-driven strategic intelligence assessment, but also a major innovation in the working mode and thinking mode of intelligence analysis and assessment personnel.
[0006] As the underlying support technology for problem-driven strategic intelligence assessment, the pre-trained big prediction model is a large-scale neural network based on billions of parameters. It has powerful natural language processing, computer vision and speech recognition capabilities. It can learn and extract features from massive data to achieve accurate understanding and efficient processing of input data. It plays a huge role in application fields such as machine translation, sentiment analysis, speech recognition, text and image generation, and data recommendation. The pre-trained big prediction model is considered a key technology for the next generation of artificial intelligence. The pre-trained big prediction model can generate multi-dimensional and multi-level research questions, assist users in the construction of strategic intelligence research problem space, and summarize and extend problems in the analysis process, and improve strategic intelligence analysis ideas. The pre-trained big prediction model can also recommend appropriate analysis strategies based on specific problem models during the analysis process, so that implicit knowledge such as expert experience can be externalized and reused. Summary of the invention
[0007] The purpose of the present invention is to provide a problem-driven strategic intelligence assessment method to address the problems existing in the prior art, which can effectively guide intelligence analysts to solve strategic problems, comprehensively use intelligence data, analytical tools, and computational models, accumulate arguments and refine ideas, thereby solving the above-mentioned problems.
[0008] The technical solution of the present invention is as follows:
[0009] A problem-driven approach to strategic intelligence assessment that includes:
[0010] Step S1: Based on the pre-trained large language model, construct the research problem space of strategic intelligence topics in an icon-based manner and model each research problem structure;
[0011] Step S2: For each research question, organize relevant intelligence data by manual search and recommendation based on pre-trained large language models, use intelligence messages, materials, entities, events, and feature data as input, and use analytical tools and computational models to conduct multi-round, interactive intelligence analysis;
[0012] Step S3: saving the original data, intermediate data or output data as arguments during the analysis process, and editing and saving one or more evaluation opinions based on subjective and objective arguments;
[0013] Step S4: Conduct a review of the analysis process, save all or part of the intelligence data references, analysis tool calls, computational model selections, and expected output operations as analysis strategies, combine the analysis progress and research problem models to form a thinking chain in the interactive intelligence analysis of other research problems, and make strategy recommendations based on the pre-trained large language model.
[0014] Furthermore, the icons include: a hierarchical list, a relationship diagram, a fishbone diagram, and a tree diagram.
[0015] Furthermore, the modeling includes:
[0016] Modeling is done according to the name, type, instructions, labels, assumptions, and problem relationships of each research question, and research questions can be added, merged, and deleted.
[0017] Furthermore, the problem relationship includes: inductive relationship, extended relationship, and associated relationship.
[0018] Furthermore, the relevant intelligence data includes: intelligence messages, information, entities, events, and feature data.
[0019] Furthermore, the step S2 comprises:
[0020] For each research question, according to the scope defined by its model, relevant intelligence data are organized by manual search and recommendation based on pre-trained large language models. According to the different types and characteristics of intelligence messages, materials, entities, events, and feature data, they are displayed in the form of lists, relationship diagrams, and geographic information platforms. With intelligence messages, materials, entities, events, and feature data as input, multiple rounds of interactive intelligence analysis are carried out through dragging and manual editing, with the help of analysis tools and computational models, and new research questions are summarized and derived during the analysis process.
[0021] Furthermore, the step S3 comprises:
[0022] During the analysis process, save the original data, intermediate data or output data as arguments and attach a snapshot of the analysis tool. You can also manually upload and edit arguments. Edit and save one or more evaluation viewpoints based on subjective and objective arguments, associate the evaluation viewpoints and arguments, and use the evaluation viewpoints and arguments as the results of the current research problem.
[0023] Compared with the prior art, the present invention has the following beneficial effects:
[0024] 1. The present invention uses a pre-trained large language model to construct a research problem space for strategic intelligence topics in a graphical manner; for each research question, it organizes relevant intelligence data by manual search and recommendation based on a pre-trained large language model, and uses intelligence messages, materials, entities, events, and feature data as inputs, and uses analysis tools and computing models to conduct multiple rounds of interactive intelligence analysis; during the analysis process review, all or part of the data references, analysis tools, and computing model selection operations are saved as analysis strategies, and recommended in the intelligence analysis process of other problems. Compared with the traditional data-driven strategic intelligence assessment model, it can effectively guide intelligence analysts to solve strategic problems, comprehensively use intelligence data, analysis tools, and computing models, and accumulate arguments and refine opinions.
[0025] 2. On the one hand, the present invention can realize the complete closed loop of strategic intelligence assessment analysis chain and evidence chain, effectively avoid the mindset, enhance the credibility of strategic intelligence assessment work, and also propose intelligence data requirements for missing evidence to strengthen the effective closed loop of intelligence production chain; on the other hand, it can recommend appropriate analysis strategies in the analysis process based on specific problem models, so that implicit knowledge such as expert experience can be externalized and reused. . BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 A flowchart of a problem-driven approach to strategic intelligence assessment. DETAILED DESCRIPTION
[0027] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0028] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.
[0029] Embodiment 1
[0030] See also Figure 1 , a problem-driven strategic intelligence assessment method, which includes the following steps:
[0031] Step S1: Based on the pre-trained large language model, the research problem space of the strategic intelligence topic is constructed in the form of hierarchical lists, relationship diagrams, fishbone diagrams, and tree diagrams, and each research problem is modeled according to its name, type, instructions, labels, assumptions, and problem relationships. Research problems can be added, merged, and deleted;
[0032] Step S2: For each research question, according to the scope defined by its model, organize the relevant intelligence data by manual search and recommendation based on the pre-trained large language model, and display them in the form of lists, relationship diagrams, and geographic information platforms according to the different types and characteristics of intelligence messages, materials, entities, events, and feature data. With intelligence messages, materials, entities, events, and feature data as input, use drag-and-drop and manual editing methods to carry out multiple rounds of interactive intelligence analysis with the help of analysis tools and computational models, and summarize and derive new research questions in the analysis process;
[0033] Step S3: during the analysis process, save the original data, intermediate data or output data as arguments, and attach a snapshot of the analysis tool. You can also manually upload and edit the arguments. Edit and save one or more evaluation viewpoints based on subjective and objective arguments. Associate the evaluation viewpoints and arguments, and use the evaluation viewpoints and arguments as the results of the current research problem.
[0034] Step S4: Conduct a review of the analysis process, save all or part of the intelligence data references, analysis tool calls, computational model selections, and expected output operations as analysis strategies, combine the analysis progress and research problem models to form a thinking chain in the interactive intelligence analysis of other research problems, and make strategy recommendations based on the pre-trained large language model.
[0035] In this embodiment, specifically, during the query process of the characteristic data related to the strategic intelligence research question, this example is developed using JavaScript, and the Json configuration example of the characteristic data is:
[0036] "d_feature_conf":{
[0037] "name":"Soybean import trade of country A in January 2023",
[0038] "source":"Event Library",
[0039] "start_time":"2023-1-1",
[0040] "end_time":"2023-1-31",
[0041] "locations":[
[0042] "Country A",
[0043]
[0044] Among them, name represents the name of the feature data, source represents the original source of the feature data, start_time and end_time represent the start and end time, locations represent the geographic information attributes, types represent the specific type of the selected source data, entities represent the entity attributes involved, and expression represents the expression for extracting feature data.
[0045] In this embodiment, specifically, the strategic intelligence research problem-related analysis tools include: data retrieval tools, statistical analysis tools, relationship analysis tools, time series analysis tools, geographic analysis tools, indicator calculation tools, comparison analysis tools, aggregation analysis tools, SWOT tools, ACH tools, and the specific description of the data type and visualization method carried by each analysis tool is as follows:
[0046] Table 1: Data types and visualization methods supported by each analysis tool
[0047]
[0048]
[0049] In this embodiment, specifically, the strategic intelligence research question-related analysis strategy structure includes: recommended data types or data objects, recommended analysis tools, recommended calculation models, and expected analysis results.
[0050] In this embodiment, specifically, the thought chain structure of the strategic intelligence research problem-related analysis strategy recommendation includes: current problem model, analysis progress, and analysis strategy. The Json prompt word configuration example is:
[0051]
[0052] "hypothesis":[
[0053] "Country B's soybean inventory is relatively small",
[0054] "International soybean prices continue to rise"
[0055] ],
[0056] "relations":[
[0057] {
[0058] "name":"The impact of reduced soybean imports on agriculture in country A",
[0059] "type":"Induction",
[0060] } ]
[0062] },
[0063] "progress":[
[0064] {
[0065] "index":1,
[0066] "tool":"Data search tool",
[0067] "model":"Event recommendation model",
[0068] "input":"International trade of soybeans",
[0069] "output":{
[0070] "events":[
[0071] "8e025b5e-7dec-4808-a16b-4e6d3e617a17",
[0072] "550e8400-e29b-41d4-a716-446655440000"]
[0073] }
[0074] },
[0075] {
[0076] "index":2,
[0077] "tool":"Relationship Analysis Tool",
[0078] "model":"Event Relationship Discovery Model",
[0079] "input":{
[0080] "events":[
[0081] "8e025b5e-7dec-4808-a16b-4e6d3e617a17",
[0082] "550e8400-e29b-41d4-a716-446655440000" ]
[0084] },
[0085] "output":{
[0086] "relation":{
[0087] "start_node":"8e025b5e-7dec-4808-a16b-4e6d3e617a17",
[0088] "end_node":"550e8400-e29b-41d4-a716-446655440000",
[0089] "type":"Cause and Effect",
[0090] "possibility":0.8
[0091] }
[0092] }
[0093] } ]
[0095] ],
[0096] "answer":[
[0097] {
[0098] "tool":"Indicator calculation tool",
[0099] "model":"",
[0100] "input":{
[0101] "events":[
[0102] "8e025b5e-7dec-4808-a16b-4e6d3e617a17",
[0103] "550e8400-e29b-41d4-a716-446655440000" ]
[0105] },
[0106] "output":"Serious"
[0107] } ]
[0109] }.
[0110] The above-mentioned embodiments only express the specific implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the protection scope of the present application. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the technical solution concept of the present application, and these all belong to the protection scope of the present application.
[0111] This background section is provided to generally present the context of the invention, and the work of the presently named inventors, the work to the extent described in this background section, and aspects of this section that did not constitute prior art at the time of application are neither explicitly nor implicitly admitted to be prior art to the present invention.
Claims
1. A problem-driven strategic intelligence assessment method characterized by: include: Step S1: Based on the pre-trained large language model, construct the research problem space of strategic intelligence topics in an icon-based manner and model each research problem structure; Step S2: For each research question, organize relevant intelligence data by manual search and recommendation based on pre-trained large language models, use intelligence messages, materials, entities, events, and feature data as input, and use analytical tools and computational models to conduct multi-round, interactive intelligence analysis; Step S3: saving the original data, intermediate data or output data as arguments during the analysis process, and editing and saving one or more evaluation opinions based on subjective and objective arguments; Step S4: Conduct a review of the analysis process, save all or part of the intelligence data references, analysis tool calls, computational model selections, and expected output operations as analysis strategies, combine the analysis progress and research problem models to form a thinking chain in the interactive intelligence analysis of other research problems, and make strategy recommendations based on the pre-trained large language model.
2. A problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The icons include: a hierarchical list, a relationship diagram, a fishbone diagram, and a tree diagram.
3. The problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The modeling includes: Modeling is done according to the name, type, instructions, labels, assumptions, and problem relationships of each research question, and research questions can be added, merged, and deleted.
4. A problem-driven strategic intelligence assessment method according to claim 3, characterized in that: The problem relationships include: inductive relationships, extended relationships, and associative relationships.
5. The problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The relevant intelligence data include: intelligence messages, information, entities, events, and feature data.
6. A problem-driven strategic intelligence assessment method according to claim 5, characterized in that: The step S2 comprises: For each research question, according to the scope defined by its model, relevant intelligence data are organized by manual search and recommendation based on pre-trained large language models. According to the different types and characteristics of intelligence messages, materials, entities, events, and feature data, they are displayed in the form of lists, relationship diagrams, and geographic information platforms. With intelligence messages, materials, entities, events, and feature data as input, multiple rounds of interactive intelligence analysis are carried out through dragging and manual editing, with the help of analysis tools and computational models, and new research questions are summarized and derived during the analysis process.
7. The problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The step S3 comprises: During the analysis process, save the original data, intermediate data or output data as arguments and attach a snapshot of the analysis tool. You can also manually upload and edit arguments. Edit and save one or more evaluation viewpoints based on subjective and objective arguments, associate the evaluation viewpoints and arguments, and use the evaluation viewpoints and arguments as the results of the current research problem.
8. A problem-driven strategic intelligence assessment method according to claim 7, characterized in that: The analysis tools include: data retrieval tools, statistical analysis tools, relationship analysis tools, time series analysis tools, geographic analysis tools, indicator calculation tools, comparison analysis tools, aggregation analysis tools, SWOT tools, and ACH tools.
9. The problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The structure of the analysis strategy includes: recommended data types or data objects, recommended analysis tools, recommended calculation models, and expected analysis results.
10. The problem-driven strategic intelligence assessment method according to claim 1, characterized in that: The structure of the thinking chain includes: current problem model, analysis progress, and analysis strategy.