Disaster chain construction and disaster response decision-making method based on large model
Through a large-model-based method, real-time disaster information at the disaster area site is obtained and processed, disaster chains and response trees are built, and decision-making is made in combination with large language models. The problem of insufficient on-site monitoring in disaster areas in the existing technology is solved, and efficient and reliable Disaster response.
Patent Information
- Application Number
- CN202411065875.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-29
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology cannot achieve real-time monitoring of disaster areas, resulting in low disaster response efficiency and ineffective response to natural and man-made disasters.
A large-model-based method is adopted to obtain real-time disaster information at the disaster area site, process and build a disaster chain and response tree, and make decisions in combination with a large language model to achieve real-time monitoring and efficient response to the disaster area site.
Real-time monitoring and efficient response to disaster areas has been achieved, timeliness and reliability of disaster responses have been improved, and losses have been reduced.
Smart Images

Figure CN119047868A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a method for disaster chain construction and disaster response decision-making based on a large model. Background Art
[0002] Natural disasters and man-made disasters can cause serious casualties and economic losses, so timely and effective disaster response is crucial to reducing losses.
[0003] At present, when facing disasters, traditional disaster response decisions mainly rely on experts' decision-making experience in historical disasters and their judgment of on-site conditions. With the continuous development of artificial intelligence, the analysis of disaster sites is currently achieved by training models with historical data, and then experts give decision-making suggestions. However, due to the limitations of data sources and the uncertainty of on-site conditions, existing technologies cannot achieve real-time monitoring of disaster sites, and disaster response efficiency is low.
[0004] Therefore, how to provide an efficient technical solution for disaster chain construction and disaster response decision-making based on large models has become a technical problem that needs to be solved urgently. Summary of the invention
[0005] The purpose of some embodiments of the present application is to provide a method for disaster chain construction and disaster response decision-making based on a large model. Through the technical solutions of the embodiments of the present application, real-time monitoring of the disaster area can be achieved and the efficiency of disaster response can be improved.
[0006] In the first aspect, some embodiments of the present application provide a method for disaster chain construction and disaster response decision-making based on a large model, including: obtaining real-time disaster information at the disaster site; processing the real-time disaster information to obtain key entity information and a real-time disaster chain; retrieving a disaster chain subtree associated with the associated entity information from a target disaster chain, and retrieving a disaster response subtree corresponding to the real-time disaster chain from a target disaster response tree; obtaining response decision information in response to the real-time disaster information based at least on the disaster chain subtree, the disaster response subtree and the target large language model.
[0007] Some embodiments of the present application process real-time disaster information at the disaster site to obtain key entity information and real-time disaster chains, and then combine the target disaster chain, target disaster response tree, and target large language model to obtain real-time response decision information. The embodiments of the present application can achieve real-time monitoring of the disaster site, improve disaster response efficiency, and have high reliability.
[0008] In some embodiments, the real-time disaster information is processed to obtain key entity information and a real-time disaster chain, including: extracting the real-time disaster information to obtain the key entity information and key entity association relationships; and constructing the real-time disaster chain based on the key entity association relationships.
[0009] Some embodiments of the present application can obtain the specific situation at the disaster site in real time and improve the efficiency of disaster response by extracting key entity information from real-time disaster information and building a real-time disaster chain.
[0010] In some embodiments, obtaining response decision information for responding to the real-time disaster information at least based on the disaster chain subtree, the disaster response subtree and the target large language model includes: inputting the disaster chain subtree and the disaster response subtree into the target large language model to obtain first decision information; inputting the real-time disaster information into another large language model to obtain second decision information; integrating the first decision information and the second decision information to obtain the response decision information.
[0011] Some embodiments of the present application obtain response decision information by integrating the output results of a target large language model and another large language model, thereby improving the reliability of the decision while ensuring timeliness.
[0012] In some embodiments, the obtaining of response decision information for responding to the real-time disaster information is performed at least based on the disaster chain subtree, the disaster response subtree and the target large language model, including: inputting the disaster chain subtree, the disaster response subtree and the prompt information into the target large language model to obtain first decision information; inputting the real-time disaster information into another large language model to obtain second decision information; integrating the first decision information, the second decision information and expert advice to obtain the response decision information.
[0013] Some embodiments of the present application obtain response decision information through a disaster chain subtree, a disaster response subtree and prompt information, combined with a target large language model, another large language model and expert advice, to improve the reliability of decisions while ensuring timeliness.
[0014] In some embodiments, before obtaining real-time disaster information at the disaster site, the method further includes: obtaining historical disaster-related text corresponding to the historical disasters; and obtaining the target disaster chain and the target disaster response tree based on the historical disaster-related text.
[0015] Some embodiments of the present application construct a target disaster chain and a target disaster response tree through historical disaster-related texts, providing reliable support for subsequent real-time analysis of disasters.
[0016] In some embodiments, obtaining historical disaster-related text corresponding to the historical disaster includes: collecting disaster data related to the historical disaster; cleaning, filtering and data structure processing the disaster data to obtain the historical disaster-related text.
[0017] Some embodiments of the present application obtain historical disaster-related text by performing relevant processing on disaster data related to historical disasters, thereby reducing text noise and improving text quality.
[0018] In some embodiments, obtaining the target disaster chain and the target disaster response tree based on the historical disaster-related text includes: inputting the historical disaster-related text into a pre-set large language model to obtain disaster entities and causal relationship information between the disaster entities; after preprocessing the disaster entities and the causal relationship information, constructing the target disaster chain and the target disaster response tree; wherein the types of preprocessing include: deduplication and merging.
[0019] Some embodiments of the present application use a large language model to extract disaster entities from historical disaster-related texts and output causal relationship information between disaster entities, and construct a target disaster chain and a target disaster response tree after preprocessing, which have high reliability and provide reliable support for subsequent real-time analysis of disasters.
[0020] On the second aspect, some embodiments of the present application provide a device for disaster chain construction and disaster response decision-making based on a large model, including: an information acquisition module, used to obtain real-time disaster information at the disaster site; a processing module, used to process the real-time disaster information, obtain key entity information and a real-time disaster chain; a retrieval module, used to retrieve a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieve a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree; a response module, used to obtain response decision information for responding to the real-time disaster information based on at least the disaster chain subtree, the disaster response subtree and the target large language model.
[0021] In a third aspect, some embodiments of the present application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the method described in any embodiment of the first aspect.
[0022] In a fourth aspect, some embodiments of the present application provide an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement a method as described in any embodiment of the first aspect when executing the program.
[0023] In a fifth aspect, some embodiments of the present application provide a computer program product, wherein the computer program product comprises a computer program, wherein the computer program, when executed by a processor, can implement the method described in any embodiment of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the technical solutions of some embodiments of the present application, the drawings required for use in some embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0025] Figure 1 A system diagram of disaster chain construction and disaster response decision-making based on a large model provided for some embodiments of the present application;
[0026] Figure 2 A flow chart of a method for constructing a target disaster chain and a target disaster response tree provided in some embodiments of the present application;
[0027] Figure 3 A schematic diagram of retrieval of a target disaster chain and a target disaster response tree provided for some embodiments of the present application;
[0028] Figure 4 One of the flow charts of the method for disaster chain construction and disaster response decision-making based on a large model provided in some embodiments of the present application;
[0029] Figure 5 A second flowchart of a method for disaster chain construction and disaster response decision-making based on a large model provided in some embodiments of the present application;
[0030] Figure 6 A block diagram of the device composition for disaster chain construction and disaster response decision-making based on a large model provided in some embodiments of the present application;
[0031] Figure 7 A schematic diagram of an electronic device is provided for some embodiments of the present application. DETAILED DESCRIPTION
[0032] The technical solutions in some embodiments of the present application will be described below in conjunction with the drawings in some embodiments of the present application.
[0033] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0034] In related technologies, a disaster chain refers to a series of chain reactions and derivative events triggered by a disaster event. For example, an earthquake may trigger secondary disasters such as a tsunami and a nuclear leak. Accurately constructing a disaster chain is crucial for comprehensive risk assessment and the formulation of response measures. However, due to the complexity and variability of the disaster chain, traditional modeling methods are difficult to capture the causal relationship and dynamic development process, especially in terms of timeliness, which is far from enough.
[0035] The Large Language Model (LLM) is an emerging technology in the field of natural language processing, which has powerful text understanding and generation capabilities. It can learn knowledge from large-scale text data and generate relevant responses based on the input text dialogue. Since LLM can well capture the semantic and contextual information in the text, it is expected to be applied to disaster chain construction and disaster response decision-making.
[0036] In view of this, some embodiments of the present application provide a method for disaster chain construction and disaster response decision-making based on a large model, which combines historical disaster-related data with a large language model to obtain a target disaster chain and a target disaster response tree. At the scene of the actual disaster, real-time disaster information is obtained in real time, and after processing the real-time disaster information, the target disaster chain, the target disaster response tree and the target large language model are combined to obtain response decision information. Some embodiments of the present application can make full use of the knowledge contained in the massive historical disaster data, dynamically capture the development and evolution of the disaster chain, and provide timely and targeted suggestions to decision makers, which is expected to greatly improve the timeliness and pertinence of disaster response, thereby reducing the losses caused by disasters.
[0037] The following is combined with Figure 1 The overall composition structure of the system for disaster chain construction and disaster response decision-making based on a large model provided by some embodiments of the present application is exemplified.
[0038] like Figure 1As shown, some embodiments of the present application provide a system for disaster chain construction and disaster response decision-making based on a large model, and the system for disaster chain construction and disaster response decision-making based on a large model includes: a terminal 100 and a server 200. Among them, the terminal 100 can send real-time disaster information at the disaster site to the server 200. The server 200 can process the real-time disaster information to obtain key entity information and a real-time disaster chain; then, through the key entity information and the real-time disaster chain, the disaster chain subtree and the disaster response subtree are retrieved in the target disaster chain and the target disaster response tree respectively. Finally, according to the disaster chain subtree, the disaster response subtree and the target large language model, the response decision information for responding to the real-time disaster information is obtained.
[0039] In some embodiments of the present application, the terminal 100 may be a mobile terminal or a non-portable computer terminal, which is not specifically limited in the embodiments of the present application.
[0040] It is understandable that in order to achieve timely response decisions at the disaster site, in some embodiments of the present application, it is first necessary to process and analyze historical disaster situations to obtain a target disaster chain and a target disaster response tree.
[0041] The following is combined with Figure 2 The implementation process of constructing a target disaster chain and a target disaster response tree performed by the server 200 provided in some embodiments of the present application is exemplified.
[0042] Please see attached Figure 2 , Figure 2 A flow chart of a method for constructing a target disaster chain and a target disaster response tree is provided for some embodiments of the present application. The method for constructing a target disaster chain and a target disaster response tree may include:
[0043] S210, obtaining historical disaster related text corresponding to the historical disaster.
[0044] For example, in some embodiments of the present application, historical disaster information that has occurred in the past is first collected and summarized to obtain historical disaster related texts. The historical disasters may include earthquakes, floods, etc., which are not specifically limited in the present application.
[0045] In some embodiments of the present application, S210 may include: collecting disaster data related to the historical disaster; cleaning, filtering and data structure processing the disaster data to obtain text related to the historical disaster.
[0046] For example, in some embodiments of the present application, historical disaster reports, network data, papers and other unstructured data are collected (as a specific example of disaster data, the data format can be text, table, picture, etc.). Irrelevant content is cleaned up, and only the parts related to the disaster situation and the processing situation are retained, so as to reduce the text noise processed by the subsequent large language model and improve the quality of the subsequent generated content. Similarly, for the existing manually extracted high-quality structured disaster data (as another specific example of disaster data, the data format can be table, json, etc.), irrelevant content is filtered out and the disaster-related parts are retained. The data structure of the collected disaster data is adjusted to obtain historical disaster-related text that is convenient for subsequent input into the large language model for analysis.
[0047] S220: Based on the historical disaster related text, obtain the target disaster chain and the target disaster response tree.
[0048] For example, in some embodiments of the present application, a target disaster chain and a target disaster response tree are obtained by analyzing historical disaster-related texts.
[0049] In some embodiments of the present application, S220 may include: inputting the historical disaster-related text into a pre-set large language model to obtain disaster entities and causal relationship information between the disaster entities; after preprocessing the disaster entities and the causal relationship information, constructing the target disaster chain and the target disaster response tree; wherein the types of preprocessing include: deduplication and merging.
[0050] For example, in some embodiments of the present application, the prompt engineering instruction is tuned to better set the large language model task (as a specific example of a pre-set large language model), and the large language model is controlled to perform stable data processing. The above-mentioned historical disaster-related text is input into the large language model, and the disaster entities such as primary disasters, secondary disasters, a series of subsequent results, and the causal relationship between the disaster entities (i.e., causal relationship information) are extracted from the historical disaster-related text. Since there is a certain degree of redundancy and overlap in the extracted disaster entities and causal relationships, these disaster entities and relationships can be further deduplicated and merged to construct the target disaster chain and target disaster response tree respectively.
[0051] Specifically, the construction of the target disaster chain only requires connecting the disaster entities through the causal relationship between them. The entity nodes in the target disaster chain are disasters and consequences at different levels, while the relationship is the causal relationship from the previous level of disaster to the next level of disaster.
[0052] For the target disaster response tree, it is the response measures from primary disasters to secondary disasters and the final consequences. Among them, the characteristic of the target disaster response tree is that the final leaf node is a number of response measures for a specific disaster chain, such as Figure 3 The tree structure of the target disaster response tree shown in the figure has a root node that is a top-level disaster, such as an earthquake; there are many secondary disasters of secondary consequences, such as heavy rain, landslide, etc. In addition, there are still many secondary consequences, which are constructed step by step into a tree structure. The final leaf node is the response measure corresponding to the disaster (such as Figure 3 Response 2-1-1, response 2-1-2, etc.), and on each level of the path from the root node to the leaf node, there is the frequency of occurrence of the corresponding disaster entity, which provides a reference for disaster response decision-making (for example, giving priority to response measures with higher frequency to deal with disasters).
[0053] The following is combined with Figure 4 The implementation process of disaster chain construction and disaster response decision-making based on a large model and performed by the server 200 is exemplified in some embodiments of the present application.
[0054] Please see attached Figure 4 , Figure 4 A flowchart of a method for disaster chain construction and disaster response decision-making based on a large model is provided for some embodiments of the present application. The method for disaster chain construction and disaster response decision-making based on a large model may include:
[0055] S410, obtaining real-time disaster information at the disaster site.
[0056] For example, in some embodiments of the present application, voice, text and other disaster description content (as a specific example of disaster information) fed back from the disaster area is obtained in real time through relevant on-site equipment (such as video equipment, mobile phone terminals, drones, etc.).
[0057] S420, processing the real-time disaster information to obtain key entity information and a real-time disaster chain.
[0058] For example, in some embodiments of the present application, key entity information is quickly extracted from the disaster description content (i.e., it can correspond to nodes in the target disaster chain and the target disaster response tree, such as typhoons, landslides, etc.), and a real-time disaster chain is quickly constructed.
[0059] In some embodiments of the present application, S420 may include: extracting the real-time disaster information to obtain the key entity information and key entity association relationships; and constructing the real-time disaster chain based on the key entity association relationships.
[0060] For example, in some embodiments of the present application, by Figure 2The provided method embodiment extracts entities from the disaster description content to obtain key entity information (referred to as key entities) and key entity association relationships, and constructs a real-time disaster chain corresponding to the key entity information through the key entity association relationships.
[0061] S430: Retrieve a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieve a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree.
[0062] For example, in some embodiments of the present application, based on the key entity information obtained above and the constructed real-time disaster chain, the target disaster chain and the target disaster response tree are matched and retrieved to obtain the possible subsequent disaster trends and corresponding reference measures. Figure 3 As shown in Figure 1, based on the key entity (i.e., the disaster node in the figure) as the retrieval basis, the tree is traversed to return the subtree of the corresponding node (for example, Figure 3 The nodes in the thick frame are used as a specific example of the disaster chain subtree. The target disaster response tree can be searched through the real-time disaster chain, which can be understood as the search of the subtree of the target disaster response tree (for example, the real-time disaster chain of the actual disaster is "secondary disaster 2-2 to secondary disaster 2-2-2", then the retrieved subtree is Figure 3 The tree in the dashed box (as a specific example of a disaster response subtree).
[0063] S440, obtaining response decision information for responding to the real-time disaster information at least based on the disaster chain subtree, the disaster response subtree and the target large language model.
[0064] For example, in some embodiments of the present application, the disaster chain subtree and the disaster response subtree are analyzed through the target large language model to obtain response decision information.
[0065] In some embodiments of the present application, S440 may include: inputting the disaster chain subtree and the disaster response subtree into the target large language model to obtain first decision information; inputting the real-time disaster information into another large language model to obtain second decision information; integrating the first decision information and the second decision information to obtain the response decision information.
[0066] For example, in some embodiments of the present application, two LLM experts are set up (as a specific example of the target large language model and another large language model). Specifically, the disaster chain subtree and disaster response subtree obtained above are input into an LLM expert for retrieval to give the first decision information. The original disaster description content is input into another LLM expert to obtain faster and more direct second decision information. By integrating the two, the response decision information can be obtained. The specific integration method can be selected according to actual needs, for example, taking the similar parts of the two, or combining the two together as the final response decision information, etc. The embodiments of the present application are not specifically limited here.
[0067] In other embodiments of the present application, S440 may include: inputting the disaster chain subtree, the disaster response subtree and the prompt information into the target large language model to obtain first decision information; inputting the real-time disaster information into another large language model to obtain second decision information; integrating the first decision information, the second decision information and expert advice to obtain the response decision information.
[0068] For example, in some embodiments of the present application, the user can also provide some relevant prompt information. When making a decision, the disaster chain subtree, disaster response subtree and prompt information obtained above can be input into an LLM expert for retrieval, and a retrieval result analysis (as a specific example of the first decision information) is given. The disaster description content is analyzed by another large language model to obtain a disaster description analysis (as a specific example of the second decision information). The retrieval result analysis, disaster description analysis and expert advice are then integrated to obtain the final response decision information, and the disaster response decision is completed. The reliability of disaster response is improved by combining a large language model with manual methods.
[0069] The following is combined with Figure 5 The specific process of disaster chain construction and disaster response decision-making based on a large model provided by some embodiments of the present application is exemplified.
[0070] Please see attached Figure 5 , Figure 5 A flow chart of a method for disaster chain construction and disaster response decision-making based on a large model is provided for some embodiments of the present application.
[0071] The above process is explained below as an example.
[0072] S501, collecting disaster data related to historical disasters.
[0073] S502, cleaning, filtering and data structure processing of disaster data to obtain historical disaster related text.
[0074] S503, inputting historical disaster-related text into a pre-set large language model to obtain disaster entities and causal relationship information between disaster entities.
[0075] S504, after pre-processing the disaster entity and causal relationship information, a target disaster chain and a target disaster response tree are constructed.
[0076] S505, obtaining real-time disaster information at the disaster site.
[0077] S506, extracting the real-time disaster information to obtain key entity information and key entity association relationships.
[0078] S507, construct a real-time disaster chain based on the key entity association relationship.
[0079] S508, retrieving a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieving a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree.
[0080] S509, input the disaster chain subtree, the disaster response subtree and the prompt information into the target large language model to obtain the first decision information; input the real-time disaster information into another large language model to obtain the second decision information.
[0081] S510, integrating the first decision information, the second decision information and the expert advice to obtain response decision information.
[0082] It can be understood that the specific implementation process of S501 to S510 can refer to the method embodiment provided above, and in order to avoid repetition, the detailed description is appropriately omitted here.
[0083] Through some of the above-mentioned embodiments of the present application, it can be seen that this application benefits from the rich knowledge and powerful generation capabilities of LLM. The construction of the disaster chain is simpler, the cost is greatly reduced, and the scenarios for building the disaster chain are more numerous and wider. There is no need to re-collect and process data, and to spend manpower, material resources and financial resources to train special models to build disaster chain generation models for specific scenarios. The present application can quickly and in real time build disaster chains. It can not only generate disaster chains using historical data, but also build disaster chains in real time based on news from the disaster site, and can quickly provide support for disaster analysis and decision-making recommendations. The present application also supports multi-expert decision-making, with native LLM as an ordinary expert, RAG (retrieval enhanced generation) + LLM as a senior expert, and in conjunction with human experts, to make more scientific, reasonable, reliable and effective response decision information.
[0084] Please refer to Figure 6 , Figure 6The block diagram of the composition of the device for disaster chain construction and disaster response decision-making based on a large model provided by some embodiments of the present application is shown. It should be understood that the device for disaster chain construction and disaster response decision-making based on a large model corresponds to the above method embodiment and can execute each step involved in the above method embodiment. The specific functions of the device for disaster chain construction and disaster response decision-making based on a large model can be found in the description above. To avoid repetition, the detailed description is appropriately omitted here.
[0085] Figure 6 The device for disaster chain construction and disaster response decision-making based on a large model includes at least one software function module that can be stored in a memory in the form of software or firmware or solidified in the device for disaster chain construction and disaster response decision-making based on a large model. The device for disaster chain construction and disaster response decision-making based on a large model includes: an information acquisition module 610, which is used to obtain real-time disaster information at the disaster site; a processing module 620, which is used to process the real-time disaster information to obtain key entity information and a real-time disaster chain; a retrieval module 630, which is used to retrieve a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieve a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree; a response module 640, which is used to obtain response decision information to respond to the real-time disaster information based on at least the disaster chain subtree, the disaster response subtree and the target large language model.
[0086] In some embodiments of the present application, the processing module 620 is used to extract the real-time disaster information to obtain the key entity information and key entity association relationships; and construct the real-time disaster chain based on the key entity association relationships.
[0087] In some embodiments of the present application, the response module 640 is used to input the disaster chain subtree and the disaster response subtree into the target large language model to obtain the response decision information.
[0088] In some embodiments of the present application, the response module 640 is used to input the disaster chain subtree, the disaster response subtree and the prompt information into the target large language model to obtain the first decision information; and integrate the first decision information and the expert advice to obtain the response decision information.
[0089] In some embodiments of the present application, before the information acquisition module 610, the device for disaster chain construction and disaster response decision-making based on the large model includes: a construction module (not shown in the figure), which is used to obtain historical disaster-related texts corresponding to historical disasters; based on the historical disaster-related texts, the target disaster chain and the target disaster response tree are obtained.
[0090] In some embodiments of the present application, a construction module (not shown in the figure) is used to collect disaster data related to the historical disaster; clean, filter and process the data structure of the disaster data to obtain text related to the historical disaster.
[0091] In some embodiments of the present application, a construction module (not shown in the figure) is used to input the historical disaster-related text into a pre-set large language model to obtain disaster entities and causal relationship information between the disaster entities; after preprocessing the disaster entities and the causal relationship information, the target disaster chain and the target disaster response tree are constructed; wherein the types of preprocessing include: deduplication and merging.
[0092] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method, and will not be described in detail here.
[0093] Some embodiments of the present application further provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the operations of the method corresponding to any of the above methods provided in the above embodiments.
[0094] Some embodiments of the present application further provide a computer program product, which includes a computer program, wherein when the computer program is executed by a processor, it can implement the operations corresponding to any of the above methods provided in the above embodiments.
[0095] like Figure 7 As shown, some embodiments of the present application provide an electronic device 700, which includes: a memory 710, a processor 720, and a computer program stored in the memory 710 and executable on the processor 720, wherein the processor 720 can implement a method as described in any of the above embodiments when reading the program from the memory 710 through a bus 730 and executing the program.
[0096] Processor 720 can process digital signals and can include various computing structures, such as complex instruction set computer structure, reduced instruction set computer structure, or a structure that implements a combination of multiple instruction sets. In some examples, processor 720 can be a microprocessor.
[0097] The memory 710 may be used to store instructions executed by the processor 720 or data related to the execution of instructions. These instructions and / or data may include codes for implementing some or all functions of one or more modules described in the embodiments of the present application. The processor 720 of the disclosed embodiment may be used to execute instructions in the memory 710 to implement the method shown above. The memory 710 includes a dynamic random access memory, a static random access memory, a flash memory, an optical memory, or other memory known to those skilled in the art.
[0098] The above description is only an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.
[0099] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0100] 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 a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
Claims
1. A method for disaster chain construction and disaster response decision-making based on a large model, characterized in that: include: Obtain real-time disaster information at the disaster site; Processing the real-time disaster information to obtain key entity information and real-time disaster chain; Retrieving a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieving a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree; Response decision information for responding to the real-time disaster information is obtained at least based on the disaster chain subtree, the disaster response subtree and the target large language model.
2. The method according to claim 1, characterized in that The processing of the real-time disaster information to obtain key entity information and a real-time disaster chain includes: Extracting the real-time disaster information to obtain the key entity information and key entity association relationships; Based on the key entity association relationship, the real-time disaster chain is constructed.
3. The method according to claim 1 or 2, characterized in that The step of obtaining response decision information for responding to the real-time disaster information at least according to the disaster chain subtree, the disaster response subtree and the target large language model includes: Inputting the disaster chain subtree and the disaster response subtree into the target large language model to obtain first decision information; Inputting the real-time disaster information into another large language model to obtain second decision information; The first decision information and the second decision information are integrated to obtain the response decision information.
4. The method according to claim 1 or 2, characterized in that: The step of obtaining response decision information for responding to the real-time disaster information at least according to the disaster chain subtree, the disaster response subtree and the target large language model includes: Inputting the disaster chain subtree, the disaster response subtree and prompt information into the target large language model to obtain first decision information; Inputting the real-time disaster information into another large language model to obtain second decision information; The first decision information, the second decision information and the expert advice are integrated to obtain the response decision information.
5. The method according to claim 1 or 2, characterized in that: Before obtaining the real-time disaster information at the disaster site, the method further includes: Obtain historical disaster related text corresponding to historical disasters; Based on the historical disaster related text, the target disaster chain and the target disaster response tree are obtained.
6. The method according to claim 5, characterized in that The obtaining of historical disaster related text corresponding to the historical disaster includes: Collect disaster data related to the historical disaster; The disaster data is cleaned, filtered and data structured to obtain text related to the historical disaster.
7. The method according to claim 5, characterized in that The step of acquiring the target disaster chain and the target disaster response tree based on the historical disaster related text includes: Inputting the historical disaster-related text into a pre-set large language model to obtain disaster entities and causal relationship information between the disaster entities; After preprocessing the disaster entity and the causal relationship information, the target disaster chain and the target disaster response tree are constructed; wherein the types of the preprocessing include: deduplication and merging.
8. A device for disaster chain construction and disaster response decision-making based on a large model, characterized in that: include: Information acquisition module, used to obtain real-time disaster information at the disaster site; A processing module, used to process the real-time disaster information to obtain key entity information and a real-time disaster chain; A retrieval module, used to retrieve a disaster chain subtree associated with the associated entity information from the target disaster chain, and retrieve a disaster response subtree corresponding to the real-time disaster chain from the target disaster response tree; The response module is used to obtain response decision information for responding to the real-time disaster information based on at least the disaster chain subtree, the disaster response subtree and the target large language model.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program executes the method according to any one of claims 1 to 7 when executed by a processor.
10. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the computer program executes the method according to any one of claims 1 to 7 when being run by the processor.
Citation Information
Patent Citations
Method and device for determining earthquake disaster rescue information
CN115186052A
Disaster emergency knowledge graph construction method for aid decision-making
CN117252255A
Method and system for defending disasters in region
CN117787471A
Method, apparatus and program for providing disaster situation management demage prediction platform
KR102593646B1
Cited By
Typhoon disaster knowledge graph generation method and system based on large language model
CN119884231A
Urban disaster chain identification method, system, device, medium and product
CN121029990A
Cross-disaster-type dynamic prompt-driven disaster knowledge element extraction method and related device
CN121328538A
Disaster knowledge element extraction method driven by cross-disaster dynamic prompt and related device
CN121328538B
Disaster event analysis method and system based on robust semantic clustering and electronic equipment
CN122470797A
