Information analysis method and related device

Abstract

The invention provides an intelligence analysis method and a related device, and relates to the technical field of data processing. The intelligence analysis method can comprise the steps of determining an analysis task of target intelligence; wherein the analysis task comprises at least one of the following items: directly extracting intelligence key information from the target intelligence, and performing logical reasoning on the target intelligence to obtain the intelligence key information; according to the determined analysis task, inputting the target intelligence into a corresponding analysis module in an intelligence analysis model, and outputting intelligence key information through the intelligence analysis model; wherein the intelligence analysis model is obtained based on big language model training, and different analysis modules are used for executing different analysis tasks. The technical scheme provided by the invention is used for solving the problem of low accuracy of an information key information extraction mode in the prior art.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to an intelligence analysis method and related devices. Background Art

[0002] In the current era of information explosion, intelligence, as an important basis for decision-making, has become increasingly valuable. Whether in the economic or technological fields, accurate and timely intelligence analysis is crucial for grasping the situation and formulating strategies.

[0003] Currently, in the field of intelligence analysis, a commonly used analysis method is the regular matching method. This method is based on a series of predefined rules and patterns, and by means of string matching, it searches for specific keywords or phrases in the intelligence text to extract the required key information.

[0004] However, the regular matching method has strong limitations. It is mainly for simple intelligence. When facing ultra-long intelligence, multi-intent intelligence, intelligence with complex tables, intelligence that requires reasoning to draw conclusions, and intelligence where the key information involved has many description methods in the original intelligence text and cannot exhaust regular expressions, the accuracy is relatively low. Summary of the Invention

[0005] Based on the above defects and deficiencies of the prior art, this application proposes an intelligence analysis method and related devices to solve the problem of low accuracy in the extraction method of key intelligence information in the prior art.

[0006] According to the first aspect of this application, an intelligence analysis method is provided. The method includes:

[0007] Determine the analysis task of the target intelligence; wherein, the analysis task includes at least one of the following: directly extracting the key intelligence information from the target intelligence, obtaining the key intelligence information after logical reasoning on the target intelligence;

[0008] According to the determined analysis task, input the target intelligence into the corresponding analysis module in the intelligence analysis model, and output the key intelligence information through the intelligence analysis model; wherein, the intelligence analysis model is trained based on a large language model, and different analysis modules are used to perform different analysis tasks.

[0009] Optionally, determining the analysis task of the target intelligence includes:

[0010] According to the content in the target intelligence, determine the event type and / or intelligence intent of the target intelligence;

[0011] Determine a parsing task corresponding to the event type and / or intelligence intention of the target intelligence according to the reference mapping relationship; wherein, the reference mapping relationship includes at least one of the following: the corresponding relationship between the event type and the parsing task, the corresponding relationship between the intelligence intention and the parsing task.

[0012] Optionally, the output of the intelligence key information by the intelligence parsing model includes:

[0013] Execute the same parsing task multiple times on the target intelligence through the intelligence parsing model to obtain multiple parsing results; wherein, each parsing result includes the parsed intelligence key information.

[0014] Compare the multiple parsing results and determine the final parsing result based on the comparison result.

[0015] Optionally, the comparing the multiple parsing results and determining the final parsing result based on the comparison result includes:

[0016] Determine the similarity between the multiple parsing results based on the intelligence key information included in each parsing result.

[0017] Perform clustering grouping based on the similarity between the multiple parsing results.

[0018] Select a parsing result in the clustering group with the most parsing results as the determined final parsing result.

[0019] Optionally, after the output of the intelligence key information by the intelligence parsing model, the method further includes:

[0020] Determine whether the intelligence key information output by the intelligence parsing model can be directly used for the target service according to the service data required by the target service.

[0021] In the case where the intelligence key information output by the intelligence parsing model cannot be directly used for the target service, convert the intelligence key information output by the intelligence parsing model into service data that can be directly used for the target service.

[0022] Optionally, the converting the intelligence key information output by the intelligence parsing model into key information that can be directly used for the target service includes:

[0023] In the case where the intelligence key information output by the intelligence parsing model is process type information, convert the intelligence key information into conclusive information required by the target service based on the relevant knowledge information of the target service.

[0024] Optionally, the target intelligence is navigation intelligence or maritime intelligence.

[0025] According to the second aspect of the present application, an intelligence analysis device is provided. The device includes:

[0026] A task determination module for determining the analysis task of the target intelligence; wherein, the analysis task includes at least one of the following: directly extracting the key intelligence information from the target intelligence, and obtaining the key intelligence information after performing logical reasoning on the target intelligence;

[0027] An intelligence analysis module for inputting the target intelligence into the corresponding analysis module in the intelligence analysis model according to the determined analysis task, and outputting the key intelligence information through the intelligence analysis model; wherein, the intelligence analysis model is trained based on a large language model, and different analysis modules are used to perform different analysis tasks

[0028] According to the third aspect of the present application, an electronic device is provided, including: a memory and a processor;

[0029] The memory is connected to the processor and is used to store programs;

[0030] The processor is used to implement the intelligence analysis method as described in the first aspect by running the program in the memory.

[0031] According to the fourth aspect of the present application, a storage medium is provided. A computer program is stored on the storage medium, and when the computer program is run by a processor, the intelligence analysis method as described in the first aspect is implemented.

[0032] According to the fifth aspect of the present application, a computer program product or a computer program is provided. The computer program product includes the computer program, and when the processor of the computer device executes the computer program, the steps in the intelligence analysis method as described in the first aspect are implemented. Optionally, the computer program can be stored in the readable storage medium or the cloud of the computer device; the processor of the computer device reads the computer program from the readable storage medium or the cloud.

[0033] In the technical solution provided by the present application, the intelligence analysis model trained based on the large language model is used to analyze the intelligence to obtain the key intelligence information. Since the large language model has strong text understanding ability, logical reasoning ability and generalization ability, it can better face various types of intelligence, thereby improving the accuracy of extracting the key intelligence information. Moreover, the technical solution provided by the present application can also perform different analysis tasks on the intelligence according to different requirements, such as directly extracting the key intelligence information from the intelligence, or obtaining the key intelligence information after performing logical reasoning on the intelligence. By performing targeted intelligence analysis, it is beneficial to further improve the accuracy of extracting the key intelligence information. Brief Description of the Drawings

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained according to the provided drawings.

[0035] Figure 1 It is a schematic flowchart of an information analysis method provided by an embodiment of the present application;

[0036] Figure 2 It is a schematic overall flowchart of an information analysis provided by an embodiment of the present application;

[0037] Figure 3 It is a block diagram of an information analysis device provided by an embodiment of the present application;

[0038] Figure 4 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed Description of the Embodiments

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0040] Application Overview

[0041] As an important basis for decision-making, information analysis is of crucial significance for grasping the situation and formulating strategies, whether in the economic or scientific and technological fields.

[0042] For example, in the aviation field, information analysis plays a key role in aspects such as flight scheduling, route planning, and flight safety. By analyzing information such as meteorological information, airspace usage, and airport operating conditions, airlines and air traffic control departments can reasonably arrange flight departure and arrival times, adjust routes, and ensure flight safety and efficiency. For example, when encountering bad weather or temporary airspace restrictions, timely and accurate information analysis can help pilots make preparations in advance, reducing flight delays and passenger inconvenience.

[0043] For another example, in the field of navigation, intelligence analysis is also indispensable for ship navigation safety, maritime traffic management, marine resource development, etc. Navigators need to rely on the analysis of charts, weather reports, hydrological data and other information to plan the best routes and avoid dangerous areas. At the same time, port management departments also need to analyze ship in and out port information, cargo loading and unloading conditions, etc. to ensure the normal operation of ports and the smooth flow of maritime traffic.

[0044] In addition to the aviation and navigation fields, intelligence analysis is also widely applied in many fields such as finance, healthcare, and environmental protection. In the financial field, by analyzing information such as market dynamics and corporate finance, investors can make wise investment decisions; in the healthcare field, the analysis of information such as disease monitoring data and medical research results helps to improve the diagnosis and treatment levels of diseases; in the environmental protection field, the analysis of information such as environmental monitoring data and pollution source distribution can provide a scientific basis for environmental protection and sustainable development.

[0045] Currently, in the field of intelligence analysis, a commonly used analysis method is the regular matching method. This method is based on a series of predefined rules and patterns, and searches for specific keywords or phrases in the intelligence text through string matching, so as to extract the required key information.

[0046] However, the regular matching method has strong limitations. It is mainly aimed at simple intelligence. When facing ultra-long intelligence, multi-intent intelligence, intelligence with complex tables, intelligence that requires reasoning to draw conclusions, and intelligence where the description methods of key information in the original text are numerous and cannot be exhausted by regular expressions, etc., the accuracy rate is relatively low. For example, ultra-long intelligence has the situations of complicated information, high probability of non-standard expression, and messy format. Since regular matching is a full-match-based matching method, the regular matching method will affect the accuracy of key information extraction when facing ultra-long intelligence, and subsequent manual correction is required. For another example, for intelligence that requires reasoning to draw conclusions, it is obviously difficult to obtain the required key information through the regular matching method.

[0047] To solve the foregoing technical problems, the present application provides an intelligence analysis technology. An intelligence analysis model trained based on a large language model is used to analyze intelligence to obtain key intelligence information. Since the large language model has strong text understanding ability, logical reasoning ability, and generalization ability, it can better face various types of intelligence, thereby improving the accuracy of key intelligence information extraction. Moreover, the technical solution provided by the present application can also perform different analysis tasks on intelligence according to different requirements. For example, directly extract key intelligence information from the intelligence, or obtain key intelligence information after logical reasoning of the intelligence. Through targeted intelligence analysis, it is beneficial to further improve the accuracy of key intelligence information extraction.

[0048] Exemplary Method

[0049] An embodiment of the present application provides an intelligence analysis method, and the execution entity can be a computer device. The computer device can be a terminal device, such as a mobile phone, a tablet computer, a desktop computer, a personal digital assistant device, etc., or a device such as a server.

[0050] The following describes this method in detail through some embodiments. The following several embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. As Figure 1 shown, the intelligence analysis method may include step 101 to step 102, which are specifically described as follows:

[0051] Step 101: Determine the analysis task of the target intelligence.

[0052] The analysis task described here may include but is not limited to at least one of the following: directly extracting key intelligence information in the target intelligence (abbreviated as element extraction task), and obtaining key intelligence information after logical reasoning on the target intelligence (abbreviated as logical reasoning task).

[0053] Since there are many types of intelligence in the same field and the description methods are not unified, for some intelligence, key information can be directly extracted from it, while for some intelligence, key information can only be obtained after logical reasoning. Therefore, the embodiments of the present application can perform different analysis tasks on intelligence according to different requirements, so as to achieve targeted analysis and better extract key information.

[0054] The target intelligence described here can be intelligence to be analyzed in any industry, such as aviation intelligence, navigation intelligence, financial intelligence, etc.

[0055] Step 102: According to the determined analysis task, input the target intelligence into the corresponding analysis module in the intelligence analysis model, and output the key intelligence information through the intelligence analysis model.

[0056] The intelligence analysis model described here can be an industry large model trained based on a large language model, that is: using the large language model as the base model, and training and optimizing the base model with industry-specific data and tasks to form a model with industry-specific knowledge and capabilities.

[0057] The intelligence analysis model has better performance in the field of intelligence analysis and can more accurately meet the requirements of intelligence analysis business. In addition, since the large language model has strong text understanding ability, logical reasoning ability and generalization ability, it can better face various types of intelligence and improve the accuracy of extracting key intelligence information.

[0058] Optionally, the large language model described herein can be a multimodal large language model, which is capable of processing intelligence in different formats, such as text intelligence, image intelligence, audio and video intelligence, etc. In this way, even if the intelligence includes complex tables, pictures, etc., it can identify and understand the content in the tables, pictures, etc., so as to judge whether it includes the required key information, and then achieve more comprehensive extraction of key information from intelligence.

[0059] Among them, the intelligence parsing model can include multiple intelligence parsing modules, and different intelligence parsing modules are used to perform different parsing tasks. For example, the intelligence parsing model can include a first parsing module and a second parsing module. The first parsing module is used to perform requirement extraction tasks, and the second parsing module is used to perform logical reasoning tasks.

[0060] In the embodiments of the present application, multi-task combined training can be performed on the intelligence parsing model, that is, multiple different parsing tasks are trained simultaneously in one model, so that when the model performs different tasks, it can independently call the capabilities of each module. In this way, even if a piece of intelligence corresponds to multiple parsing tasks, they can be executed simultaneously, which is beneficial to improving the intelligence parsing efficiency and at the same time improving the model utilization rate.

[0061] In some alternative embodiments, the training process of the intelligence parsing model can include: a training data preparation stage and a model training stage.

[0062] As Figure 2 shown, the training data preparation stage can include processes such as data collection and generalization, data cleaning, data annotation, and data quality inspection, so as to obtain a database for model training.

[0063] Among them, data collection refers to collecting data for model training. For example, intelligence data used to improve the accuracy of the output results of the model element extraction task, intelligence data used to improve the accuracy of the output results of the model logical reasoning task, relevant field knowledge documents, and prompt engineering information (such as thought chain information), etc. Taking aviation intelligence parsing as an example, the data that needs to be collected can include, but is not limited to: historical aviation intelligence data generated by airlines themselves, intelligence data provided by third-party agencies with aviation intelligence-related services, intelligence parsing training materials, intelligence parsing-related knowledge manuals, etc.

[0064] Among them, data generalization refers to using some technical means and strategies to increase the effective quantity and coverage of intelligence data when the amount of collected intelligence data is limited. For example, by transforming the original intelligence data, new samples that are similar to but not exactly the same as the original intelligence data can be generated, so as to increase the quantity and diversity of intelligence data; virtual samples similar to real intelligence data can also be generated through a generative model, so as to supplement the intelligence data set. The specific generalization method can be selected according to actual needs.

[0065] Among them, data cleaning refers to the overall cleaning of the data collected and generalized, removing data that affects model training such as duplicates, poor quality, overly complex and low-frequency data, to form a high-quality data set. For intelligence in complex formats, such as intelligence including tables, intelligence including pictures, etc., format conversion can be performed first to convert it into formats such as jason, markdown, or other formats that can be better parsed by the model, facilitating the subsequent model training work.

[0066] Among them, data annotation refers to the true value annotation of the cleaned intelligence data to meet the subsequent training of the model.

[0067] In the embodiments of this application, the training tasks of the model mainly include: element extraction ability, understanding and logical reasoning ability, etc.

[0068] For the training of the element extraction ability, true value annotation needs to be carried out in the intelligence data, that is, the key information that the model needs to extract is annotated in the intelligence data.

[0069] For the training of the understanding and reasoning ability, the reasoning result as the true value, the description of the reasoning process (that is, the chain of thought), and the knowledge point documents involved in the reasoning need to be set.

[0070] Among them, before data annotation, scenario analysis can be carried out to understand the specific scenarios in intelligence work and the key information that needs to be obtained in each scenario, so as to carry out data annotation. For example, taking aviation intelligence as an example, under types of notices such as navigational notice, snow condition notice, volcanic ash notice, airline notice, and PIB (Pre-Flight Information Bulletin), the following secondary scenarios can be included but are not limited to: terminal area notice (including but not limited to channels such as airports, fire conditions, navigation facilities, lights, runways, taxiways, etc.), airspace notice. In the embodiments of this application, the key information fields involved in each scenario can be determined, so as to carry out model design and chain of thought design, etc.

[0071] Among them, data quality inspection refers to the quality inspection of the annotated data to judge the accuracy of data annotation. Optionally, a hierarchical sampling inspection method can be used for sampling inspection. For example, the data set can be stratified based on event type, intelligence intention, or intelligence complexity, etc., and samples can be randomly selected from different layers based on a preset ratio to judge the accuracy of data annotation.

[0072] Repeat the above steps to ensure that the training tasks involved are covered by relevant data in each scenario, and finally form a data set for model training.

[0073] As Figure 2 shown, after accurately preparing the model training data in the accurate stage of training data, model training can be carried out on the basis of the large language model.

[0074] Among them, the intelligence analysis tasks mainly include: element extraction tasks and logical reasoning tasks.

[0075] For the element extraction task, the embodiments of the present application can adopt the training method of supervised fine-tuning (SFT). Data is constructed based on each element scenario to be extracted, with no less than 1000 pieces / scenario, for model training.

[0076] SFT is a deep learning strategy that can be used on pre-trained large language models. The specific approach is to fine-tune the model using labeled data to make it adapt to specific tasks or domains. In the embodiments of the present application, it enables the model to better perform the element extraction task. For this training method, the input data of the model is the original intelligence text, and the output data is the key intelligence information to be extracted based on business requirements.

[0077] For the logical reasoning task, in the embodiments of the present application, the training method of prompt engineering can be adopted to train the model's chain of thought. The main method is to use the method of distributed reasoning to organize and exemplify the process from input information to an output result of a scenario in the form of prompt engineering to teach the large model. A large number of prompt engineering, combined with relevant domain knowledge (such as aviation domain knowledge), enables the model to not only execute teaching content but also generalize in the analysis of intelligence-related problems, analyze horizontally similar scenarios, and vertically more complex scenarios, achieving the effect of drawing inferences from one instance. Among them, the prompt engineering dataset is pre-stored in the database.

[0078] During the training process, the input data of the model is the original intelligence text, the description of the reasoning process, and the knowledge point documents involved in the reasoning. The output data of the model is the logical reasoning result. The described reasoning process description is used to inform the model how to perform logical reasoning.

[0079] Among them, the training methods of SFT and prompt engineering utilize the knowledge that already exists at the bottom layer of the model (i.e., the learned knowledge). If the model effect still does not meet the usage requirements after training the general large language model with SFT and prompt engineering, the model itself needs to learn more knowledge, that is, perform pre-training on the large language model to enable the large language model to structurally learn the knowledge of relevant fields and improve the reasoning and generalization abilities of the large language model.

[0080] Finally, a multi-task combined training needs to be carried out, that is, multiple analysis tasks are trained simultaneously in one model, enabling the model to share knowledge between different analysis tasks. At the same time, when performing different analysis tasks, each analysis module's ability can be independently invoked to merge the abilities of performing different analysis tasks into one model.

[0081] In some alternative embodiments, the parsing task corresponding to the target intelligence may be pre-identified for the target intelligence by a human, and the computer device may recognize the identification information to determine the parsing task corresponding to the target intelligence.

[0082] To improve convenience and intelligence, the determination of the parsing task corresponding to the target intelligence may also be implemented by the computer device, and the computer device may automatically determine it based on some intelligence characteristics, as described below.

[0083] Step 101: Determine the parsing task of the target intelligence, which may include Step A1 and Step A2, as described below:

[0084] Step A1: Determine the event type and / or intelligence intention of the target intelligence according to the content in the target intelligence.

[0085] Step A2: Determine the parsing task corresponding to the event type and / or intelligence intention of the target intelligence according to the reference mapping relationship.

[0086] In the embodiments of the present application, the parsing task corresponding to the intelligence may be determined based on the event type and / or intelligence intention of the intelligence.

[0087] Generally, the intelligence will include event type information, which is used to indicate what event the intelligence is mainly issued for. For example, for a NOTAM (a type of aeronautical information), it may specifically be intelligence issued for airport events, fire events, aircraft runway events, or taxiway events, etc.

[0088] And there are also certain rules to follow regarding whether key information can be directly extracted from the intelligence of different events. For example, based on historical experience, it is found that for the intelligence of some events, key information can be directly extracted. For example, for the intelligence issued for taxiway events, it mainly describes the situation of the taxiway, and such intelligence can directly extract the required key information, such as whether the taxiway is available and the reason for unavailability. While for the intelligence of some events, it is difficult to directly extract the required key information. Therefore, in the embodiments of the present application, the parsing task corresponding to the target intelligence may be determined based on the event type.

[0089] Similarly, there are also certain rules to follow regarding whether key information can be directly extracted from the intelligence of different intentions. Therefore, in the embodiments of the present application, the parsing task corresponding to the target intelligence may also be determined based on the intelligence intention.

[0090] In the embodiments of the present application, the mapping relationship between each event type and the parsing task, and / or the mapping relationship between various intelligence intentions and the parsing task can be preset, so as to quickly determine the parsing task of the target intelligence according to the event type and / or intelligence intention.

[0091] Optionally, the parsing task of the target intelligence can be determined only according to the event type, or only according to the intelligence intention; it is also possible to first determine the parsing task of the target intelligence according to the event type, and if no corresponding parsing task is matched, then determine the parsing task of the target intelligence according to the intelligence intention; it is also possible to determine the parsing task of the target intelligence according to the event type and the intelligence intention at the same time. It can be specifically set according to actual needs.

[0092] Optionally, on the basis of the event type, the intelligence sending agency, the sending time, etc. can also be combined to assist in determining the parsing task corresponding to the intelligence.

[0093] In some alternative embodiments, in the embodiments of the present application, engineering script matching, small model natural language processing (Natural Language Processing) intention matching or the classification ability of a large model can be used to match the target intelligence to the corresponding parsing module in the intelligence parsing model.

[0094] Among them, engineering script matching mainly associates the pre-written script with a specific project or task according to the requirements and characteristics of the project, so that the script can run automatically under specific conditions to complete various repetitive, regular or complex work tasks. In the embodiments of the present application, through engineering script matching, the parsing task of the target intelligence is determined based on the event type and / or intelligence intention of the target intelligence, and the target intelligence is automatically input into the corresponding parsing module.

[0095] Among them, small model NLP intent matching is a technology in the field of natural language processing that uses small language models to identify and understand the intent expressed in text. In the embodiments of this application, through small model NLP intent matching, the intelligence intent of the target intelligence is determined, and based on the intelligence intent, the parsing task of the target intelligence is determined, and the target intelligence is automatically input into the corresponding parsing module. Since the structure of the small model is relatively simple, its decision-making process is relatively easy to understand and explain, which can help users better understand the prediction results and decision-making basis of the model. In addition, the small model has high flexibility and can be quickly customized and adjusted according to different application scenarios and requirements. Developers can collect relevant data to train the small model for a specific business domain to make it better adapt to the language characteristics and intent recognition requirements of that domain. Further, compared with large pre-trained language models, the training and deployment costs of small models are lower. They do not require large-scale computing resources and expensive hardware devices and can be trained and used in ordinary computing environments, with low development costs.

[0096] Of course, in the embodiments of this application, the classification ability of large models can also be used to determine the event type and / or intelligence intent of the target intelligence, so as to determine the parsing task of the target intelligence and input the target intelligence into the corresponding parsing module.

[0097] In some alternative embodiments, for the same parsing task, the intelligence parsing model supports performing multiple processes on the intelligence and comparing and verifying the results of multiple parses, so as to screen out relatively accurate output results, as described below.

[0098] "Outputting the intelligence key information through the intelligence parsing model" in step 102 may include step B1 and step B2, as described below:

[0099] Step B1: Perform multiple identical parsing tasks on the target intelligence through the intelligence parsing model to obtain multiple parsing results.

[0100] Step B2: Compare the multiple parsing results and determine the final parsing result based on the comparison result.

[0101] Among them, each parsing result includes the intelligence key information obtained by parsing.

[0102] Large language models are stochastic. For the same parsing task of the same piece of intelligence, the results output each time may be different, which can be understood as having different people handle the same parsing task of the same piece of intelligence. In the embodiments of the present application, in order to obtain relatively accurate parsing results, after inputting the target intelligence into the intelligence parsing model, the intelligence parsing model can perform the same parsing task on the target intelligence multiple times, so as to obtain multiple parsing results for the same parsing task. Then, the multiple parsing results are compared, and the final parsing result of this parsing task is determined based on the comparison result. This is beneficial to obtaining more accurate key information of the intelligence and can also save the workload of manual multiple reviews.

[0103] Optionally, the similarity between multiple parsing results can be determined based on the key information of the intelligence included in each parsing result. Then, clustering grouping is performed based on the similarity between the multiple parsing results; finally, in a clustering group with more parsing results, one parsing result is selected as the determined final parsing result.

[0104] In the embodiments of the present application, the similarity between multiple parsing results can be determined by comparing the similarity between the key information of the intelligence in different parsing results. The similarity between the key information of the intelligence here mainly refers to the similarity of the meanings expressed by the key information of the intelligence. For example, if the same situation is described in different descriptive statements in the intelligence, then it is sufficient for the parsing result to match one of the descriptive statements. Therefore, in the embodiments of the present application, the similarity of the meanings expressed by the key information of the intelligence is judged.

[0105] If the similarity of the meanings expressed by the key information of the intelligence included in two parsing results is greater than or equal to the similarity threshold, it can be considered that the similarity between the two parsing results is high; conversely, if the similarity of the meanings expressed by the key information of the intelligence included in two parsing results is less than the similarity threshold, it can be considered that the similarity between the two parsing results is low. The similarity threshold described here can be set according to actual needs, and the embodiments of the present application do not specifically limit this.

[0106] In the embodiments of the present application, clustering grouping can be performed based on the similarity between the parsing results, and the clustering group with the most parsing results is determined. This group represents the most likely output result of the model, and the parsing results in it are relatively accurate. Therefore, one parsing result can be selected from this group as the final parsing result.

[0107] It can be understood that a relatively accurate parsing result can also be manually selected from multiple parsing results as the final parsing result. Moreover, for the output final parsing result, manual verification can also be performed, and when the parsing result is incorrect, manual correction can be made to further improve the accuracy of the intelligence parsing result.

[0108] It can be understood that in the embodiments of the present application, multiple intelligence analysis models can also be used to process the same intelligence respectively, so as to obtain multiple analysis results for the same analysis task of the same intelligence. Then, the multiple analysis results are compared and analyzed, and a relatively accurate analysis result is selected as the final output result to improve the processing efficiency. For such an embodiment, a comparison and analysis module can be set up, which is connected to multiple intelligence analysis models, and each intelligence analysis model can send the output result to the comparison and analysis module for comparison and analysis.

[0109] Optionally, for the output results with low accuracy (i.e., bad cases), they can be analyzed to optimize the model. For example, as Figure 2 shown, for bad cases, the intelligence analysis model can be optimized from aspects such as training data, SFT training strategy, and instruction engineering training strategy. The output result with low accuracy can be obtained through the previous clustering method (such as the analysis result in the group with the least number of analysis results), or can be determined manually.

[0110] In some alternative embodiments, the analysis results output by the intelligence analysis model may not be directly applicable to the business system, that is, they cannot be directly used for the corresponding business. In this case, the analysis results need to be understood and processed before being used in the actual business. Therefore, after the intelligence key information is output by the intelligence analysis model, the method may further include step C1 and step C2, as described below:

[0111] Step C1: Determine whether the intelligence key information output by the intelligence analysis model can be directly used for the target business according to the business data required by the target business.

[0112] Step C2: In the case where the intelligence key information output by the intelligence analysis model cannot be directly used for the target business, convert the intelligence key information output by the intelligence analysis model into business data that can be directly used for the target business.

[0113] The business data required for different businesses has rules and specifications to follow, such as which information needs to be included, and what types of information are required. The embodiments of the present application can judge whether the intelligence key information output by the intelligence analysis model can be directly used for the corresponding business based on these rules and specifications.

[0114] Such as Figure 2As shown, if the key information of the intelligence parsed by the intelligence parsing model can be directly used for the corresponding business, it is sent to the corresponding business system for business application; if the key information of the intelligence parsed by the intelligence parsing model cannot be directly used for the corresponding business, it can be sent to the intelligence parsing agent (i.e., the intelligence parsing Agent), and this intelligence parsing Agent processes the key information of the intelligence that cannot be directly used for the business system and converts it into business data that can be directly used for the business system. This method is more convenient and intelligent compared to the manual processing method.

[0115] Optionally, the key information of the intelligence that cannot be directly used for the target business may be process information. This type of information can be understood as an intermediate value of the process and may require querying relevant knowledge bases and other methods to obtain conclusive information. Therefore, this type of information needs to be processed to convert it into conclusive information. Therefore, the foregoing step: converting the key information of the intelligence parsed by the intelligence parsing model into business data that can be directly used for the target business may include:

[0116] When the key information of the intelligence parsed by the intelligence parsing model is process information, based on the relevant knowledge information of the target business, the key information of the intelligence is converted into conclusive information required by the target business.

[0117] For example, when the intelligence parsing model outputs the key information of the intelligence based on the target intelligence: "After 18:00 every day at Airport A, lights a, b, and c on Runway 11 are turned off", at this time, it is not known whether there are other lights on Runway 11 at Airport A, and conclusive information has not been obtained. At this time, the intelligence parsing Agent can be used to query that there are only lights a, b, and c on Runway 11 at Airport A. Therefore, the conclusive information required by the business can be obtained: "All lights on Runway 11 at Airport A are turned off after 18:00".

[0118] Optionally, the reasons why the key information of the intelligence cannot be directly used for the target business may still include, but are not limited to: including redundant information, information format mismatch, lack of information, etc. These situations can all be processed by the intelligence parsing Agent to obtain business data that can be directly used for the target business.

[0119] Among them, as Figure 2 shown, the intelligence parsing Agent may include a data and business logic configuration management module, a document knowledge management module, a structured information extraction module, etc. Among them, the data and business logic configuration management module can be used to determine the conversion strategy of the key information of the intelligence after receiving the key information of the intelligence. The document knowledge management module can be used to provide the relevant knowledge required in the process of converting the key information of the intelligence. The structure information extraction module is used to extract the required relevant knowledge from the document including the relevant knowledge required in the process of converting the key information of the intelligence.

[0120] In some alternative embodiments, there is multi-intent information among existing intelligence, that is, a piece of intelligence describes multiple things, which may be events of multiple types, or different notifications of the same type of event based on different conditions. In the description of each event, there may be hit information of regular matching, but for intelligence parsing, it may not be necessary to have the key information of all events. Therefore, regular matching will affect the extraction effect of key information, resulting in redundant information in the extracted key information.

[0121] For such intelligence, in the embodiments of the present application, the corresponding parsing task can be determined based on the event type, rather than based on the intelligence intent, so as to avoid performing redundant parsing tasks and improve the accuracy of the model output result.

[0122] Of course, even if the parsing task of the target intelligence is determined according to the intelligence intent, if no corresponding parsing task is set for an unimportant intelligence intent, it is also not necessary to perform redundant parsing tasks.

[0123] In some alternative embodiments, general intelligence messages have a certain structure. In the embodiments of the present application, items with strong structure that do not need to be parsed with the help of a model can be screened out by means of an engineering script, and the intelligence text that cannot be processed by the script can be screened out and sent to the intelligence parsing model.

[0124] Taking aeronautical intelligence as an example, it generally includes items such as a header, Q line, A, B, C, D, E, etc. Among them, items with strong structure include: header, Q line, A, B, C, D items, which can be screened out by means of an engineering script, while item E cannot be processed by the script, so it is input into the intelligence parsing model.

[0125] Among them, the header part includes information such as the telegram level, receiving address, issuing date and time, sending address, etc. The Q line is the content code of the navigational notice, item A is the place of occurrence, item B is the effective time, item C is the termination time, item D is the limitation line, including flight information region (FIR), navigational notice content code (Q-CODE), type of flight (TRAFFIC), purpose of issuing the navigational notice (PURPOSE), scope of influence (SCOPE), lower and upper limits (LOWER / UPPER), coordinates, radius (COORDINATES, RADIUS), etc. Item E is the body of the navigational notice, which is a specific description and detailed explanation of the theme of the navigational notice, including the specific situation of the event, relevant data, etc.

[0126] The above is the description of the intelligence parsing method provided by the embodiments of the present application.

[0127] In summary, in the embodiments of the present application, an intelligence parsing model trained based on a large language model is used to parse intelligence to obtain key intelligence information. Since the large language model has strong text understanding ability, logical reasoning ability, and generalization ability, it can better handle various types of intelligence. The parsing result is not limited to the literal content, but can also reason and display information, improving the accuracy of extracting key intelligence information. Moreover, the technical solution provided by the present application can also perform different parsing tasks on intelligence according to different requirements, such as directly extracting key intelligence information from the intelligence, or obtaining key intelligence information after logical reasoning of the intelligence. Through targeted intelligence parsing, it is beneficial to further improve the accuracy of extracting key intelligence information. In addition, in the embodiments of the present application, batch processing of intelligence is also supported, that is, multiple intelligence parsing models can be used to parse different intelligence simultaneously, thereby improving the intelligence parsing efficiency.

[0128] Exemplary Device

[0129] Correspondingly, the embodiments of the present application also provide an intelligence parsing device, and the execution subject can be a computer device. The computer device can be a terminal device, such as a mobile phone, a tablet computer, a desktop computer, a personal digital assistant device, etc., or a server device, etc.

[0130] As Figure 3 shown, the device may include:

[0131] A task determination module 301, configured to determine the parsing task of the target intelligence.

[0132] Wherein, the parsing task includes at least one of the following: directly extracting key intelligence information from the target intelligence, and obtaining key intelligence information after logical reasoning of the target intelligence.

[0133] An intelligence parsing module 302, configured to input the target intelligence into the corresponding parsing module in the intelligence parsing model according to the determined parsing task, and output key intelligence information through the intelligence parsing model.

[0134] Wherein, the intelligence parsing model is trained based on a large language model, and different parsing modules are used to perform different parsing tasks.

[0135] In some optional embodiments, the task determination module 301 may include:

[0136] A first determination unit, configured to determine the event type and / or intelligence intention of the target intelligence according to the content in the target intelligence.

[0137] A second determination unit, configured to determine a parsing task corresponding to the event type and / or intelligence intention of the target intelligence according to a reference mapping relationship; wherein, the reference mapping relationship includes at least one of the following: a correspondence between an event type and a parsing task, and a correspondence between an intelligence intention and a parsing task.

[0138] In some alternative embodiments, the intelligence parsing module 302 may include:

[0139] A first parsing unit, configured to perform the same parsing task on the target intelligence multiple times through the intelligence parsing model to obtain multiple parsing results.

[0140] Wherein, each of the parsing results includes parsed intelligence key information.

[0141] A second parsing unit, configured to compare the multiple parsing results and determine a final parsing result based on the comparison result.

[0142] In some alternative embodiments, the second parsing unit may include:

[0143] Based on the intelligence key information included in each of the parsing results, determine the similarity between the multiple parsing results; based on the similarity between the multiple parsing results, perform clustering grouping; in a clustering group with the largest number of parsing results, select one parsing result as the determined final parsing result.

[0144] In some alternative embodiments, after the intelligence parsing model outputs the intelligence key information, the device may further include:

[0145] A judgment module, configured to determine whether the intelligence key information output by the intelligence parsing model can be directly used for the target service according to the intelligence key information required by the target service.

[0146] A conversion module, configured to convert the intelligence key information output by the intelligence parsing model into key information that can be directly used for the target service in the case where the intelligence key information output by the intelligence parsing model cannot be directly used for the target service.

[0147] In some alternative embodiments, the conversion module may include:

[0148] A conversion unit, configured to, in the case where the intelligence key information output by the intelligence parsing model is process type information, convert the intelligence key information into conclusive information required by the target service based on the relevant knowledge information of the target service.

[0149] In some alternative embodiments, the target intelligence is navigation intelligence or maritime intelligence.

[0150] The information parsing device provided in this embodiment belongs to the same inventive concept as the information parsing method provided in the above embodiments of the present application. It can execute the information parsing method provided in any of the above embodiments of the present application and has the corresponding functional modules and beneficial effects for executing the method. For the technical details not described in detail in this embodiment, reference can be made to the specific processing content of the information parsing method provided in the above embodiments of the present application, which will not be elaborated here.

[0151] It should be understood that the modules in the above information parsing device can be implemented in the form of a processor calling software. For example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or the functions of each unit of the device. The processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory inside the device or a memory outside the device. Alternatively, the units in the device can be implemented in the form of a hardware circuit. By designing the hardware circuit, the functions of some or all of the units can be realized. The hardware circuit can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are realized by designing the logical relationship of the components in the circuit. Another example, in another implementation, the hardware circuit can be implemented by a PLD. Taking an FPGA as an example, it can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured through a configuration file to realize the functions of some or all of the above units. All units of the above device can be all implemented in the form of a processor calling software, or all implemented in the form of a hardware circuit, or some are implemented in the form of a processor calling software, and the remaining part is implemented in the form of a hardware circuit.

[0152] In the embodiments of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor can be a circuit with the ability to read and execute instructions, such as a CPU, a microprocessor, a GPU, or a DSP, etc. In another implementation, the processor can realize certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or can be reconstructed. For example, the processor is a hardware circuit implemented by an ASIC or a PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to realize the configuration of the hardware circuit can be understood as the process of the processor loading instructions to realize the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as a kind of ASIC, such as an NPU, a TPU, a DPU, etc.

[0153] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method. For example: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0154] In addition, each unit in the above device can be integrated in whole or in part, or can be independently implemented. In one implementation, these units are integrated together and implemented in the form of an SOC. The SOC can include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The types of the at least one processor can be different. For example, it includes a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0155] Exemplary Electronic Device

[0156] The embodiments of the present application also provide an electronic device, such as Figure 4 shown, the electronic device includes: a memory 400 and a processor 410.

[0157] The memory 400 is connected to the processor 410 and is used to store programs.

[0158] The processor 410 is used to implement the intelligence parsing method in the above embodiments by running the programs stored in the memory 400.

[0159] Specifically, the above electronic device may further include: a communication interface 420, an input device 430, an output device 440, and a bus 450.

[0160] The processor 410, the memory 400, the communication interface 420, the input device 430, and the output device 440 are interconnected through the bus. Among them:

[0161] The bus 450 may include a path for transmitting information between various components of the computer system.

[0162] The processor 410 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present invention. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0163] The processor 410 may include a main processor, and may also include a baseband chip, a modem, etc.

[0164] The memory 400 stores a program for implementing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include program code, and the program code includes computer operation instructions. More specifically, the memory 400 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk memory, a flash memory, etc.

[0165] The input device 430 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor, etc.

[0166] The output device 440 may include a device for allowing information to be output to a user, such as a display screen, a printer, a speaker, etc.

[0167] The communication interface 420 may include a device of any transceiver type for communicating with other devices or communication networks, such as an Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc.

[0168] The processor 410 executes the program stored in the memory 400 and calls other devices, and can be used to implement each step of the intelligence analysis method provided in the above embodiments of the present application.

[0169] Exemplary Computer Program Product and Storage Medium

[0170] In addition to the above methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, and the computer program instructions, when run by a processor, cause the processor to execute the steps in the intelligence analysis method described in the embodiments of the present application.

[0171] Among them, the above computer program product may be specifically implemented in a manner of hardware, software, or a combination thereof. In an alternative embodiment, the computer program product is specifically embodied as a computer storage medium, and in another alternative embodiment, the computer program product is specifically embodied as a software product, such as a Software Development Kit (SDK), etc.

[0172] The computer program product can be written in any combination of one or more programming languages for executing the program code of the operations of the embodiments of the present application. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, executed as an independent software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0173] In addition, an embodiment of the present application can also be a storage medium, on which a computer program is stored, and the computer program is executed by a processor to perform the steps in the intelligence analysis method described in the embodiments of the present application.

[0174] For the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of combinations of actions. However, those skilled in the art should know that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0175] It should be noted that the embodiments in this specification are all described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0176] The steps in the methods of the embodiments of the present application can be adjusted, combined, and deleted according to actual needs. The technical features recorded in each embodiment can be replaced or combined.

[0177] The modules and sub-modules in the devices and terminals in the embodiments of the present application can be combined, divided, and deleted according to actual needs.

[0178] In several embodiments provided by this application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the terminal embodiments described above are merely illustrative. For example, the division of modules or sub-modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple sub-modules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or modules can be in electrical, mechanical or other forms.

[0179] The modules or sub-modules described as separate components may or may not be physically separated. The components as modules or sub-modules may or may not be physical modules or sub-modules, that is, they can be located in one place, or can be distributed to multiple network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0180] In addition, in each embodiment of this application, each functional module or sub-module can be integrated in a processing module, or each module or sub-module can exist physically alone, or two or more modules or sub-modules can be integrated in one module. The above-mentioned integrated modules or sub-modules can be implemented in the form of hardware or in the form of software functional modules or sub-modules.

[0181] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but this implementation should not be considered to exceed the scope of this application.

[0182] The steps of the methods or algorithms described in combination with the embodiments disclosed in this article can be directly implemented by hardware, software units executed by a processor, or a combination of the two. The software units can be placed in a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0183] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used 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 the 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 one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligence analysis method, characterized in that, The method includes: Determining an analysis task for the target intelligence; wherein, the analysis task includes at least one of the following: directly extracting key intelligence information from the target intelligence, obtaining key intelligence information after performing logical reasoning on the target intelligence; According to the determined analysis task, inputting the target intelligence into the corresponding analysis module in the intelligence analysis model, and outputting key intelligence information through the intelligence analysis model; wherein, the intelligence analysis model is trained based on a large language model, and different analysis modules are used to perform different analysis tasks.

2. The intelligence analysis method according to claim 1, wherein The determining of the analysis task for the target intelligence includes: Determining the event type and / or intelligence intention of the target intelligence according to the content in the target intelligence; Determining the analysis task corresponding to the event type and / or intelligence intention of the target intelligence according to the reference mapping relationship; wherein, the reference mapping relationship includes at least one of the following: the correspondence between the event type and the analysis task, the correspondence between the intelligence intention and the analysis task.

3. The intelligence analysis method according to claim 1, characterized in that, The outputting of the key intelligence information through the intelligence analysis model includes: Performing the same analysis task on the target intelligence multiple times through the intelligence analysis model to obtain multiple analysis results; wherein, each analysis result includes the key intelligence information obtained through analysis; Comparing the multiple analysis results and determining the final analysis result based on the comparison result.

4. The intelligence analysis method according to claim 3, characterized in that, The comparing of the multiple analysis results and determining the final analysis result based on the comparison result includes: Determining the similarity between the multiple analysis results based on the key intelligence information included in each analysis result; Performing clustering grouping based on the similarity between the multiple analysis results; Selecting one analysis result in the clustering group with the most analysis results as the final analysis result.

5. The intelligence analysis method according to claim 1, wherein After the outputting of the key intelligence information through the intelligence analysis model, the method further includes: Determining whether the key intelligence information output by the intelligence analysis model can be directly used for the target service according to the service data required by the target service; In the case where the key intelligence information output by the intelligence analysis model cannot be directly used for the target service, converting the key intelligence information output by the intelligence analysis model into service data that can be directly used for the target service.

6. The intelligence analysis method according to claim 5, characterized in that The converting of the key intelligence information output by the intelligence analysis model into key information that can be directly used for the target service includes: In the case where the key intelligence information output by the intelligence analysis model is intermediate information, converting the key intelligence information into conclusive information required by the target service based on the relevant knowledge information of the target service.

7. The intelligence analysis method according to any one of claims 1 to 6, characterized in that, The target intelligence is navigation intelligence or maritime intelligence.

8. An intelligence analysis device, characterized in that, The device includes: A task determination module for determining an analysis task for the target intelligence; wherein, the analysis task includes at least two of the following: directly extracting key intelligence information from the target intelligence, obtaining key intelligence information after performing logical reasoning on the target intelligence; An intelligence analysis module, configured to input the target intelligence into the corresponding analysis module in the intelligence analysis model according to the determined analysis task, and output the key intelligence information through the intelligence analysis model; wherein, the intelligence analysis model is trained based on a large language model, and different analysis modules are used to perform different analysis tasks.

9. An electronic device, characterized in that, It includes: A memory and a processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the intelligence analysis method according to any one of claims 1 to 7 by running the program in the memory.

10. A storage medium, characterized in that, A computer program is stored on the storage medium, and when the computer program is run by the processor, the intelligence analysis method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is run by the processor, the intelligence analysis method according to any one of claims 1 to 7 is implemented.

Images