Technical economic risk assessment evidence-based report generation method based on open source intelligence

Abstract

The invention provides a technical and economic risk assessment evidence-based report generation method based on open source intelligence, and relates to the technical field of risk assessment, and the method comprises the steps: carrying out the dynamic collection and preprocessing of multi-source intelligence, and outputting a data set; customizing a technical economic risk assessment framework to obtain a report framework; performing field focusing and technical point value evaluation based on the data set; performing data association organization on the technical point list; performing structured information extraction and evidence-based generation on the technical point data set to obtain an event list; filling to a report framework, and outputting a standardized report. The technical problem that technical risk assessment evidence-based requirements cannot be matched due to the fact that multi-source open-source information is difficult to integrate and a report template is rigid in the prior art, and accordingly the professionality and efficiency of report generation are affected can be solved, and the technical risk field open-source information automatic convergence is achieved, so that the technical risk assessment efficiency is improved. The technical risk assessment report is automated, and the generation efficiency of the technical risk assessment report is improved.

Description

Technical Field

[0001] This invention relates to the field of risk assessment technology, specifically to a method for generating evidence-based reports on techno-economic risk assessment based on open-source intelligence. Background Technology

[0002] Current methods for generating evidence-based technology risk assessment reports primarily rely on two approaches: manual review and analysis, and automated scanning tools and large-scale model question-and-answer methods. These methods are inefficient and risk missing key risk events. Furthermore, existing methods typically collect data from limited, static open-source information sources, lacking flexibility and adaptability. They cannot integrate multi-source intelligence in real time, nor can they personalize and adjust open-source information sources according to the specific needs of different technology risk domains, thus affecting the accuracy and comprehensiveness of technology risk assessments. In addition, most existing technology risk assessment reports rely on fixed templates, failing to provide flexible report generation based on diverse technology risk assessment needs. The generally fixed templates cannot be customized to the individual needs of different domains or users, resulting in report content and structure that do not meet specific technology risk assessment evidence-based requirements, thus impacting the professionalism and efficiency of report generation.

[0003] In summary, existing technologies suffer from technical problems such as difficulty in integrating multi-source open-source information and rigid report templates, which fail to meet the evidence-based requirements of technical risk assessment, thus affecting the professionalism and efficiency of report generation. Summary of the Invention

[0004] This application provides a method for generating evidence-based reports on technology and economic risk assessment based on open-source intelligence. This method addresses the technical problems in existing technologies, such as the difficulty in integrating multi-source open-source information and the rigidity of report templates, which makes it impossible to match the evidence-based requirements of technology risk assessment and thus affects the professionalism and efficiency of report generation.

[0005] In view of the above problems, this application provides a method for generating evidence-based reports on technology and economic risk assessment based on open-source intelligence. The method includes: dynamic collection and preprocessing of multi-source intelligence to output a dataset; customization of a technology and economic risk assessment framework to obtain a report framework; domain focusing and technology point value assessment based on the dataset to obtain a list of technology points; data association and organization of the technology point list to obtain a technology point dataset; structured information extraction and evidence-based generation of the technology point dataset to obtain an event list; and filling the event list into the report framework to output a standardized report.

[0006] Optionally, the data source can be dynamically configured through an interactive interface; the data source can be crawled and updated in a triggered manner to clean the unstructured data; the cleaned structured data can be stored in the database by field, and the dataset can be output.

[0007] Optionally, a preset template based on a basic framework is loaded, wherein the basic framework includes a service object module, an analysis subject module, and a main text module; the user-added module dynamically adjusted by the user is received, and after adjusting the module weights in conjunction with the preset template, the report framework is output.

[0008] Optionally, domain keywords are extracted based on the analysis subject in the report framework; synonyms of the domain keywords are generated through a domain knowledge graph to obtain a domain-related dataset; the frequency of technical point words is statistically analyzed in the domain-related dataset and time-decay weighted to obtain a domain-related popularity set; risk keyword co-occurrence analysis is performed on the domain-related popularity set to obtain the risk correlation degree of the domain-related popularity set; and a list of technical points after sorting the domain-related popularity set is output based on the risk correlation degree.

[0009] Optionally, technical point matching is performed on the heat data associated with each field in the technical point list, and data IDs are associated to obtain an associated dataset; the associated dataset is divided according to the risk correlation, and the technical point dataset is output, wherein the technical point dataset includes technical point name, associated data ID list and risk level.

[0010] Optionally, the technology point dataset is classified into a preset definition of technology and economic risk event types to obtain event type labels for the technology point dataset; the five elements of the event are extracted from the technology point dataset, the time format is normalized, and the event element information is integrated to obtain an event list.

[0011] Optionally, multiple event element information of the same technical point are sorted by time, and the event element information with timeline alignment is connected to form entity relationships to obtain a subject-behavior-impact closed loop; multi-source cross-validation is performed based on the subject-behavior-impact closed loop to output a structured evidence chain, which is then added to the event list.

[0012] Optionally, the event list is categorized and populated into the report framework to generate a draft report, wherein each event in the event list is accompanied by multiple source references; the draft report is then converted into a standardized report based on preset industry standards.

[0013] Optionally, the initial draft report is optimized under key optimization points to obtain an initial optimized report, wherein the key optimization points include terminology standardization, logical coherence, and redundancy removal; the initial optimized report is then adjusted in priority, supplemented with evidence, and reassessed for risk based on a manual intervention interface to output the standardized report.

[0014] Optionally, the standardized report can be exported in multiple formats, while retaining historical versions and the multi-source references; version backtracking and comparison can be performed based on the historical versions to generate a change log; and the change log can be added to the standardized report.

[0015] The technical solution provided in this application has at least the following beneficial effects:

[0016] Through dynamic collection and preprocessing of multi-source intelligence, a dataset is output; a customized technology and economic risk assessment framework is developed to obtain a report framework; based on the dataset, domain focusing and technology point value assessment are performed to obtain a technology point list; the technology point list is then organized through data association to obtain a technology point dataset; structured information extraction and evidence-based generation are performed on the technology point dataset to obtain an event list; the event list is then populated into the report framework to output a standardized report. In other words, through dynamic collection and preprocessing of multi-source intelligence, a customized report framework is developed based on the specific needs of technology and economic risk assessment. Knowledge extraction is performed on the preprocessed dataset to identify high-value technology points. Combined with large-scale modeling technology, structured descriptions of technology risk events are automatically generated, and an evidence-based technology risk assessment report is output, improving the efficiency and professionalism of technology risk assessment report generation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method for generating evidence-based reports on techno-economic risk assessment based on open-source intelligence, as proposed in this application.

[0019] Figure 2 This is a flowchart illustrating the process of obtaining a list of technical points in the evidence-based report generation method for technical and economic risk assessment based on open-source intelligence, as described in this application. Detailed Implementation

[0020] This application provides a method for generating evidence-based reports on technology and economic risk assessments based on open-source intelligence. This addresses the technical challenges of existing methods, such as the difficulty in integrating multi-source open-source information and rigid report templates, which fail to meet the evidence-based requirements of technology risk assessments, thus affecting the professionalism and efficiency of report generation. Through dynamic collection and preprocessing of multi-source intelligence, a customized report framework is developed based on the specific needs of the technology and economic risk assessment. Knowledge extraction is performed on the preprocessed dataset to identify high-value technical points. Combined with large-scale modeling techniques, structured descriptions of technology risk events are automatically generated, outputting an evidence-based technology risk assessment report. This improves the efficiency and professionalism of technology risk assessment report generation.

[0021] Examples, such as Figure 1 As shown, this application provides a method for generating an evidence-based report on technological and economic risk assessment based on open-source intelligence. Specifically, this method includes the following steps:

[0022] Dynamic collection and preprocessing of multi-source intelligence, outputting a dataset.

[0023] Furthermore, this application also includes the following steps: dynamically configuring the data source through an interactive interface; triggering the crawling and updating of the data source to clean the unstructured data; storing the cleaned structured data to the database by field, and outputting the dataset.

[0024] Specifically, through an interactive interface, users can dynamically configure data sources based on their specific technical risk assessment needs. The interface allows users to select specific open-source intelligence sources (such as news websites, academic papers, technical forums, patent databases, etc.) and adjust them according to the needs of different fields, such as adding, selecting, and adjusting the source and type of open-source intelligence data. For example, some users may only focus on risks in the new energy sector, while others may focus on the artificial intelligence sector. The interactive interface allows users to interact with the system through a graphical interface, inputting or adjusting data configurations using controls provided by the interface (such as buttons, input boxes, etc.), enabling users to easily configure, control, and manage the system according to their needs.

[0025] After configuration, data crawling and updating are triggered by specific events. This means the data crawling process is initiated based on predefined rules or event-triggered mechanisms. For example, when a data source (such as a news website or research database) publishes new information, data crawling automatically starts, promptly capturing the newly published content and updating the data. This triggering mechanism ensures data timeliness. For instance, if a new policy in a certain technical field is released at night, relevant information is automatically captured and updated within a short period (e.g., within minutes), ensuring the data is always up-to-date. Triggered crawling means that data collection is not continuous but initiated by specific events or conditions. Updating refers to re-collecting information from the data source to obtain new or changed content since the last collection, maintaining data timeliness.

[0026] The majority of the crawled data is unstructured, such as news articles, research reports, and social media posts. Cleaning this unstructured data involves removing irrelevant content (such as advertisements and noisy data), labeling key information (such as technical risk events and relevant policies), and formatting (such as segmenting long texts into concise entries and extracting important event descriptions and dates). Unstructured data is data with no fixed format or predefined pattern, such as text on web pages, images, PDF document content, and social media posts. Cleaning processes the raw data to remove noise, errors, and irrelevant information (such as HTML tags, advertisements, and irrelevant links), correct formatting errors, and standardize data representation, making the data cleaner, more standardized, and easier to process later.

[0027] Cleaned unstructured data is transformed into structured data, which is then stored in a database according to fields (such as event time, event title, event description, impact, etc.). This way, each data point has a clear format, facilitating subsequent querying and analysis. Fields are columns or attributes in structured data. This ever-growing table in the database, along with all the structured records stored within it, constitutes the dataset.

[0028] For example, the user selected data sources related to new energy technologies, including the latest technology news websites and patent databases. The user set up automatic data updates at 3:00 AM every day and added keyword filtering. After configuration, the system automatically crawled and updated information from these sources. A new file was crawled, published at 6:00 AM, containing updates to new energy battery technology standards. The unstructured text was cleaned, and key information about the event, such as time (6:00 AM), title (update of new energy battery technology standards), description (standards affecting battery technology R&D and production), and impact (high), was extracted and stored in the database.

[0029] It achieves flexibility and automation in open-source intelligence data acquisition, significantly solving the problem of ineffective integration of multi-source information in existing technologies. Through user-driven dynamic configuration, it precisely focuses on the information sources needed in specific technology risk areas, avoiding interference from irrelevant information and improving data relevance.

[0030] A technology and economic risk assessment framework was customized, resulting in a reporting framework.

[0031] Furthermore, this application also includes the following steps: loading a preset template based on a basic framework, wherein the basic framework includes a service object module, an analysis subject module, and a main text module; receiving a user-added module dynamically adjusted by the user, adjusting the module weights in conjunction with the preset template, and then outputting the report framework.

[0032] Specifically, the system loads a pre-defined template based on a basic framework, including a target audience module, an analysis body module, and a main body module. The basic framework provides a standard structure and organization for the report, ensuring it includes the necessary components. For example, the target audience module defines the report's intended audience or target reader group, specifying that the report is aimed at strategic decision-makers to help them understand the impact of technological risks. Clearly defining the target audience helps determine the report's focus, language style, and level of detail. The analysis body module defines the core object of the report's analysis, which can be a specific entity (such as an organization) or an abstract concept (such as a technological field or industry). The main body module presents the detailed technological risk analysis and is the core content of the report, containing specific analyses, findings, and conclusions. In short, the basic framework clearly defines who the report is for (target audience module), the core of the analysis (analysis body module), and the main content areas of the report (main body module).

[0033] The system accepts user-added modules that are dynamically adjusted by users, meaning that users can dynamically adjust the report framework based on their specific technical risk assessment needs. Users can add new modules to the report framework through an interactive interface, such as adding a market demand forecasting module or a technology development trend analysis module, or delete unnecessary parts, allowing the report to flexibly meet the needs of different fields or users.

[0034] After users add or adjust modules, the weights of each module are adjusted automatically using algorithms, combining preset templates and the new modules. For example, if a certain technological risk is particularly important to strategic decision-makers, the weight of that module is automatically increased, giving it more space in the report. By adjusting the module weights, the report content accurately reflects the user's needs and concerns. For example, a startup company focused on AI chip design (the analysis subject) needs to write an evidence-based technology risk assessment report, targeting the company's investors and technical decision-makers (the service recipients). First, the basic framework is loaded. The preset template includes modules for the service recipients, the analysis subject, and the main text (overview, technological risks, impact, etc.). For this startup, technological progress and potential risks are the most critical, so the framework is dynamically adjusted: the weight of the preset sub-technology point technology risk monitoring module is increased from the default 0.3 to 0.6, and a new supply chain security risk module is added, with a weight of 0.15. Following these adjustments, the technical risk monitoring module for specific technical points was placed earlier in the report body with more space allocated. The newly added supply chain security risk module was integrated into the main body, with its position and length allocated according to its weight. The final report framework clearly states that the report will focus on the specific technical risks of AI chip design (accounting for 60% of importance), followed by supply chain security risks (15%), then related impacts (10%), and finally basic information such as a report overview, ensuring that the report content is highly aligned with the company's current most pressing concerns.

[0035] By introducing a dynamically adjustable basic framework and preset templates, the system moves beyond generic and potentially impractical report structures. Instead, it allows for flexible customization of report modules and their importance based on user-specified technical risk areas, analysis targets, and service recipients. The ability to add modules and adjust weights ensures the report framework accurately reflects the core concerns of a specific task, guaranteeing the report's professionalism and relevance.

[0036] Based on the dataset, domain focus and technology value assessment are performed to obtain a list of technologies.

[0037] Further details are attached. Figure 2 As shown, this application further includes the following steps: extracting domain keywords based on the analysis subject in the report framework; generating synonyms for the domain keywords through a domain knowledge graph to obtain a domain-related dataset; statistically analyzing the frequency of technical points in the domain-related dataset and performing time-decay weighting to obtain a domain-related popularity set; performing risk keyword co-occurrence analysis on the domain-related popularity set to obtain the risk correlation degree of the domain-related popularity set; and outputting a list of technical points after sorting the domain-related popularity set based on the risk correlation degree.

[0038] Specifically, focusing on specific domains and assessing the value of technologies based on datasets is equivalent to organizing data by domain and selecting high-value, important, and closely related multi-source intelligence data. The analysis subject module in the report framework—the technical field or object of the report's focus—helps define the report's analytical scope and focus. Domain keywords, which are core terms related to a specific technical field, are extracted from the analysis subject module. These keywords typically reflect the main problems, trends, and technological advancements in that field. Domain keywords are a series of core terms or phrases closely related to the analysis subject, summarizing the core content of the field and forming the basis for subsequent information retrieval and analysis. For example, in the field of artificial intelligence chip design, keywords might include AI chips, GPUs, TPUs, neural network accelerators, and architecture design.

[0039] By leveraging domain knowledge graphs, synonyms for domain keywords are generated, thus expanding the domain keywords into a rich set of synonyms and forming a domain-related dataset. This significantly broadens the coverage of subsequent searches. The domain-related dataset contains domain keywords and their synonym sets derived from the knowledge graph, which are used for subsequent retrieval and analysis across a larger dataset to discover a wide range of information related to that domain. For example, GPU can be expanded to include graphics processing units.

[0040] In the domain-related dataset, the frequency of each technical point is analyzed, and time-decay weighting is applied. Technical point frequency refers to the frequency of each technical point or keyword in the domain-related dataset. A high frequency may indicate that the technical point is of high importance or has received considerable attention within a certain period. Time-decay weighting adjusts the technical point frequency to give higher weight to more recent events or technical points. Typically, newer data is given higher weight than older data to reflect the timeliness of the information.

[0041] After time-decay weighting, a domain-related heat set is obtained, reflecting the relative heat (importance) of each technology point within the current time period. Co-occurrence analysis refers to analyzing the frequency of two or more words appearing simultaneously in the same document or sentence. Risk keyword co-occurrence analysis specifically refers to analyzing the frequency of technology point keywords and risk keywords appearing together in the same text. If a technology point frequently co-occurs with risk keywords such as supply chain disruptions, then this technology point may have a high supply chain risk. Risk keyword co-occurrence analysis helps identify which technology points are associated with potential risks. Based on the results of contribution analysis, the risk correlation degree of each technology point is calculated. Technology points with high risk correlation degrees indicate that they may be associated with higher technological risks.

[0042] The list of technical points is sorted based on risk relevance, with high-risk technical points prioritized for user focus. This results in a list of technical points including event time, event subject, event title, event description, and event impact. For example, the current report framework analyzes LiDAR technology in autonomous vehicles. First, domain keywords are extracted, such as LiDAR, solid-state LiDAR, MEMS LiDAR, detection range, and point cloud density. Then, using a domain knowledge graph, LiDAR is expanded to include light detection and ranging, and laser scanners; solid-state LiDAR is expanded to include LiDAR without moving parts, forming a domain-related dataset. In a structured dataset containing approximately 500,000 news articles, blog posts, and technical forum posts from the past three months, word frequencies are statistically analyzed and time-decayed weighted (assuming a weight of 1.0 for the most recent month, 0.7 for the previous 1-2 months, and 0.4 for the previous 2-3 months). The total weighted word frequency is 1200 for LiDAR, 800 for solid-state LiDAR, 450 for MEMS LiDAR, and 600 for detection range. A co-occurrence analysis of risk keywords was conducted, selecting keywords such as high cost, performance bottleneck, supply chain risk, and competitor advantage. The analysis revealed that solid-state LiDAR and high cost co-occurred 28 times, MEMS LiDAR and performance bottleneck co-occurred 15 times, detection range co-occurred 10 times, and LiDAR itself and supply chain risk (such as shortage of key optical components) co-occurred 35 times. Based on the number of co-occurrences, the risk correlation was calculated: LiDAR 35, solid-state LiDAR 28, MEMS LiDAR 15, and detection range 10. A list of technologies was output, sorted by risk correlation: LiDAR (risk correlation 35), solid-state LiDAR (risk correlation 28), MEMS LiDAR (risk correlation 15), and detection range (risk correlation 10).

[0043] By combining domain knowledge graphs and intelligent analysis algorithms, this system automatically and accurately identifies high-risk technologies relevant to specific analytical subjects from massive amounts of unstructured open-source intelligence. Time decay weighting ensures the timeliness of the focus points, while risk keyword co-occurrence analysis directly links technological hotspots with potential risks. The final output is a sorted list of technologies, avoiding wasting resources on irrelevant or low-risk technologies and significantly improving the efficiency and depth of technology risk assessment.

[0044] The list of technical points is organized by data association to obtain a technical point dataset.

[0045] Furthermore, this application also includes the following steps: matching technical points with the relevant heat data for each field in the technical point list, and associating them with data IDs to obtain a related dataset; dividing the related dataset according to the risk correlation, and outputting the technical point dataset, wherein the technical point dataset includes technical point names, a list of associated data IDs, and risk levels.

[0046] Specifically, for each domain-related popularity data in the technology point list, technology point matching is performed, matching the domain-related popularity data with specific technology points to ensure that each technology point can be associated with relevant data and information. For each technology point in the list (such as LiDAR, solid-state LiDAR, etc.), a technology point matching operation is performed, traversing the previously collected, cleaned, and stored raw dataset in the database, using the technology point name and its synonyms expanded through a knowledge graph as query conditions, and matching within the dataset. Whenever a data entry containing these keywords is found, the unique ID of that data entry is recorded, creating a list of associated data IDs for each technology point. The associated dataset is the collection of data obtained after matching technology points with their associated domain popularity data. Each data item is associated with a specific technology point and contains relevant data information, such as technical background and technological development trends.

[0047] The related datasets are segmented based on risk correlation. This means that the related datasets are divided into different categories or levels according to the risk correlation of each technology point, ensuring that high-risk technologies and their related data are prioritized for processing and display. Technologies with high risk correlation will be displayed first or given higher weight, determining the risk level of each technology point. The technology point dataset is the final output dataset containing information related to each technology point, including the technology point name, a list of associated data IDs (i.e., the data entries associated with that technology point), and the risk level of that technology point. For example, for LiDAR, its associated data ID list contains 4 IDs, with a risk correlation of 0.85, and is classified as high-risk. By accurately matching technology points with domain-related heat data, the relationship between technology points and related data is effectively identified, ensuring that each technology point receives sufficient support. The risk correlation segmentation ensures that high-risk technology points and their related data are prioritized for processing, helping decision-makers focus on the most critical technological risks.

[0048] The dataset of technical points is subjected to structured information extraction and evidence-based generation to obtain an event list.

[0049] Furthermore, this application also includes the following steps: classifying the technology point dataset into a preset definition of technology and economic risk event types to obtain event type labels for the technology point dataset; extracting the five elements of an event from the technology point dataset, normalizing the time format, and integrating the event element information to obtain an event list.

[0050] Furthermore, this application also includes the following steps: sorting multiple event element information of the same technical point by time, connecting the event element information with timeline alignment to form entity relationships, and obtaining a subject-behavior-impact closed loop; performing multi-source cross-validation based on the subject-behavior-impact closed loop, outputting a structured evidence chain, and adding it to the event list.

[0051] Specifically, the predefined definitions of technological and economic risk event types are categories of potential technological and economic risk events, such as technology regulation and blockade events, technology supply chain risk events, technology competition and contest events, sanctions events, risk events, industry control and monopoly events, and technological innovation breakthrough events. Based on the events in the technology point dataset, they are assigned to these predefined technological and economic risk event types. Event type labels are obtained from the technology point dataset, and each event in the dataset is labeled according to the predefined risk event types, allowing for the classification and subsequent processing of different types of risk events.

[0052] The five key elements of an event are extracted from a technical data set: event time, event subject, event title, event description, and event impact. Event time is the time the event occurred; the event subject is the leading party, such as a company; the event title is a brief description or heading used to quickly identify the core content; the event description is a detailed account of the event, including its background, causes, and process; and the event impact is the consequences or effects of the event, typically related to the market, technology, etc. Unifying time data from different sources into a standard format enables unified processing and sorting of time data. A common standard time format is YYYY-MM-DD.

[0053] For multiple events related to the same technology, the information of each event element is sorted chronologically to ensure that each event has a clear position in the timeline of the technology. After sorting the events, it's crucial to ensure their alignment along the time dimension. For example, if event B is a market demand change triggered by the technological innovation of event A, align the times of events A and B to better illustrate the evolution of the technology. Carefully examine the sorted information, identifying those closely connected and interrelated information fragments on the timeline, and piece them together like a jigsaw puzzle. Figure 1In this way, the entities in these fragments are connected according to their logical relationships in the real world. By connecting the entity relationships, a logically self-consistent event chain is formed, a subject-behavior-impact closed loop, clearly showing: a certain subject (who) performed a certain behavior (what did they do), and this behavior led to a specific result or impact (what consequences occurred). For example, if information fragment A describes company X announcing the development of a new quantum key distribution device, information fragment B describes the successful passing of the device prototype through security testing, and information fragment C describes the industry's concerns about its security caused by the test results, connecting A, B, and C forms a closed loop: Company X develops a new device (subject-behavior) → device passes testing (behavior result) → industry concerns (impact).

[0054] After completing the closed-loop construction, a multi-source cross-validation mechanism is initiated to collect information from multiple different sources and verify the events to ensure their accuracy and consistency. Based on the results of multi-source cross-validation, a structured evidence chain is generated for each event, consisting of multiple supporting data sources. For example, the evidence chain for event A might include company announcements (ID1), industry news (ID2), and reports (ID3). The structured evidence chains are integrated and added to the event list, forming a list containing all technical risk events. This list includes not only basic event information (such as time and impact) but also verification evidence for each event, ensuring the credibility and logical coherence of the report content. A structured evidence chain refers to the chain of evidence obtained through multi-source cross-validation. It demonstrates the event's occurrence process, impact, and source through logical and hierarchical data support. A structured evidence chain is event information that has been cross-validated and has a subject-behavior-impact closed loop. It is organized according to a predefined structure (such as tables or specific fields) and includes the data source IDs or references supporting the event information, forming traceable and verifiable evidence records.

[0055] Populate the event list into the report frame to output a standardized report.

[0056] Furthermore, this application also includes the following steps: classifying and filling the event list into the report framework to generate a draft report, wherein each event in the event list is accompanied by multiple source citations; and converting the draft report into a standardized report based on a preset industry standard.

[0057] Furthermore, this application also includes the following steps: optimizing the initial draft report under key optimization points to obtain an initial optimized report, wherein the key optimization points include terminology standardization, logical coherence, and redundancy removal; adjusting the priority, supplementing evidence, and reassessing risks of the initial optimized report based on a manual intervention interface, and outputting the standardized report.

[0058] Furthermore, this application also includes the following steps: exporting the standardized report in multiple formats and retaining historical versions and the multi-source references; performing version backtracking and comparison based on the historical versions to generate a change log; and adding the change log to the standardized report.

[0059] Specifically, based on the pre-defined report framework, each event in the event list is categorized and populated into its corresponding section. For example, a technology risk assessment report might include sections on technology background, technological progress, risk event analysis, and impact. Each event in the event list (such as a technological innovation breakthrough or a change in market demand) is categorized according to its event type (such as technological innovation, market risk, etc.) and populated into the corresponding module in the report framework. Each event also includes multi-source citation information to ensure that the description of each event in the report has corresponding data support and verification sources. Multi-source citations typically include news reports, academic papers, industry reports, etc., to ensure the authenticity and completeness of the event information.

[0060] Based on the report framework populated with the event list, a preliminary draft of the report is generated. This draft includes descriptions and analyses of all events, and its structure is initially complete. However, the draft may have inconsistencies in terminology, unclear logical structure, or redundant content, thus requiring further optimization. That is, the event list is categorized and populated into the report framework; for example, the IonQ collaboration event is placed in the "Technological Innovation Breakthrough" section, the export control event in the "Technology Control and Blockade" section, and the university breakthrough event also in the "Technological Innovation Breakthrough" section, generating a preliminary draft. This draft may have a acceptable structure, but the terminology is inconsistent.

[0061] The initial draft report undergoes key optimizations, including terminology standardization, logical coherence, and redundancy removal. Terminology standardization involves unifying and standardizing the terminology used in the report, ensuring consistency and compliance with industry standards. Logical coherence involves structurally optimizing the report content, ensuring clear logical relationships between different sections and smooth flow. Redundancy removal involves eliminating redundant content, such as repetitive analyses and irrelevant details, making the report concise, efficient, and highlighting key information. The initial optimized report is the version after preliminary optimization (terminology standardization, logical coherence, and redundancy removal), which is more readable and structured than the initial draft, but may still require further adjustments.

[0062] Through a user-intervention interface, expert input is fully integrated, allowing experts or users to participate in report optimization. This enables users to prioritize reports, supplement evidence, and reassess risks, enhancing the report's operability and accuracy. Experts can participate in the optimization process, adjusting priorities to ensure the most critical technical risks are addressed first. Experts can also supplement evidence, such as adding relevant market data, research reports, or expert opinions, to enhance the report's credibility. If experts believe that the risk assessments for certain technical points are overly optimistic or pessimistic, they can adjust the risk reassessment through the interface, ensuring that the risk assessments for each technical point in the report are more accurate.

[0063] After all optimizations and adjustments are completed, the final standardized report is output, containing a complete, accurate, and clear technical risk assessment, using industry-standard terminology and structure, and with all events and analyses supported by ample evidence. Standardized reports are typically provided in WORD or PDF format for easy viewing and archiving by users.

[0064] For report generation across different technical fields, the above steps are repeated iteratively to generate a detailed risk assessment report for each technical point. The content of the risk assessment report for each technical point is customized based on its specific risks and technical characteristics. After assessing one technical point, assessments are then conducted for technical points in other technical fields, ensuring that the entire technical risk assessment system comprehensively covers all fields and provides holistic risk assessment support. Through automated input, optimization, and manual intervention, technical risk assessment reports are generated quickly and accurately, improving the efficiency and quality of report generation.

[0065] The final standardized report is a modifiable and updatable document format, exported as an editable office document. The exported document can be in Word or PDF format, allowing users to edit and adjust the report while retaining structured information such as technical point names, risk-related data, and chains of evidence, ensuring no data loss during editing. The standardized report can be exported in multiple formats; for example, users may need a PDF for formal distribution, while also requiring a Word document for further modifications. During export or saving, historical versions of the report are preserved, and the multiple data sources referenced for each information point in the report are fully documented. Each report export creates a new version, and all historical versions are saved. Each version has a unique version number and timestamp for backtracking. All data sources referenced in the report (such as documents, industry reports, academic research, etc.) are retained and recorded in the report, ensuring clear data support for each event and analysis.

[0066] All saved historical versions are reviewed, allowing users to view changes between different versions. Version review enables users to see specific modifications and optimizations made at different points in time. Users can select two different versions for comparison, displaying added, modified, and deleted content. Based on the comparison results, a change log is generated, recording the details of each modification, including the modification date, the modified part, and the reason for the modification. The change log is automatically added to the end of the standardized report to ensure a complete modification history. The change log typically includes: version number (e.g., version 1.0, version 1.1, version 2.0, etc.), modification date, modified content, reason for modification, and the person who made the modification.

[0067] Attaching a changelog to the report ensures that all modifications to the report are documented in detail. The changelog provides transparency into the report's modification history, helping users understand the modification and optimization process. Exporting in an editable office document format ensures that the report content can be easily modified and updated while retaining structured information and preventing data loss. Support for multiple report export formats allows reports to be adapted to different user needs. Version rollback and comparison enable users to clearly understand the changes to the report content; the generated changelog ensures the transparency and traceability of the report's modification history.

[0068] In summary, the method for generating evidence-based reports on techno-economic risk assessment based on open-source intelligence provided in this application has the following beneficial effects:

[0069] Through dynamic collection and preprocessing of multi-source intelligence, a dataset is output; a customized technology and economic risk assessment framework is developed to obtain a report framework; based on the dataset, domain focusing and technology point value assessment are performed to obtain a technology point list; the technology point list is then organized through data association to obtain a technology point dataset; structured information extraction and evidence-based generation are performed on the technology point dataset to obtain an event list; the event list is then populated into the report framework to output a standardized report. In other words, through dynamic collection and preprocessing of multi-source intelligence, a customized report framework is developed based on the specific needs of technology and economic risk assessment. Knowledge extraction is performed on the preprocessed dataset to identify high-value technology points. Combined with large-scale modeling technology, structured descriptions of technology risk events are automatically generated, and an evidence-based technology risk assessment report is output, improving the efficiency and professionalism of technology risk assessment report generation.

[0070] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0071] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for generating evidence-based reports on techno-economic risk assessment based on open-source intelligence, characterized in that, include: Dynamic collection and preprocessing of multi-source intelligence, outputting datasets; A technology and economic risk assessment framework was customized to obtain a reporting framework. Based on the dataset, domain focus and technology value assessment are performed to obtain a list of technology points; The list of technical points is organized by data association to obtain a technical point dataset; Structured information extraction and evidence-based generation are performed on the aforementioned technical point dataset to obtain an event list; Populate the event list into the report frame to output a standardized report.

2. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 1, characterized in that, Dynamic collection and preprocessing of multi-source intelligence, outputting a dataset including: Dynamically configure the data source through the interactive interface; Triggered crawling and updating of the data source to clean unstructured data; Store the cleaned structured data to the database by field, and output the dataset.

3. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 1, characterized in that, The technology and economic risk assessment framework was customized, resulting in a reporting framework, including: Load a preset template based on a basic framework, wherein the basic framework includes a service object module, an analysis subject module, and a main text module; The system receives dynamically adjusted user addition modules, adjusts module weights based on the preset template, and outputs the report framework.

4. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 3, characterized in that, Based on the dataset, domain focusing and technology value assessment are performed to obtain a list of technologies, including: Extract domain keywords based on the analytical subject in the report framework; Synonyms for the domain keywords are generated using a domain knowledge graph to obtain a domain-related dataset; The frequency of technical terms in the domain-related dataset is statistically analyzed and time-decayed to obtain the domain-related popularity set. A risk keyword co-occurrence analysis was performed on the aforementioned domain-related popularity set to obtain the risk correlation degree of the domain-related popularity set; The list of technical points is generated based on the risk correlation degree and the sorted domain correlation heat set.

5. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 4, characterized in that, The list of technical points is organized by data association to obtain a technical point dataset, including: For each field of the technology point list, perform technology point matching on the associated popularity data and associate data IDs to obtain the associated dataset; Based on the risk correlation, the associated dataset is divided and the technical point dataset is output. The technical point dataset includes the technical point name, the list of associated data IDs, and the risk level.

6. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 1, characterized in that, Structured information extraction and evidence-based generation are performed on the aforementioned technology point dataset to obtain an event list, including: The technology point dataset is classified into a preset technology and economic risk event type definition to obtain the event type label of the technology point dataset; The five elements of an event are extracted from the dataset of the technical points, the time format is normalized, and the event element information is integrated to obtain an event list.

7. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 6, characterized in that, By integrating the event element information, a list of events is obtained, including: By sorting multiple event element information of the same technical point by time, connecting the event element information with timeline alignment to form entity relationships, a subject-behavior-impact closed loop is obtained; Based on the subject-behavior-influence closed loop, multi-source cross-validation is performed to output a structured evidence chain, which is then added to the event list.

8. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 1, characterized in that, Populate the event list into the report framework to output a standardized report, including: The event list is categorized and populated into the report framework to generate a draft report, wherein each event in the event list is accompanied by multiple source references; The initial draft of the report is transformed into a standardized report based on pre-defined industry standards.

9. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 8, characterized in that, Transforming the initial draft report into a standardized report that conforms to pre-defined industry standards includes: The initial draft report is optimized by focusing on key optimization points to obtain an initial optimized report. These key optimization points include terminology standardization, logical coherence, and redundancy removal. The initial optimization report is adjusted in priority, supplemented with evidence, and reassessed in risk based on the human intervention interface, and the standardized report is output.

10. The method for generating an evidence-based report on techno-economic risk assessment based on open-source intelligence as described in claim 8, characterized in that, The output of standardized reports also includes: Export the standardized report in multiple formats, and retain historical versions and the multiple source references; Based on the historical versions, version backtracking and comparison are performed to generate a change log; Add the change log to the standardization report.

Images