Decision-making auxiliary method and device based on large model

Abstract

The invention discloses a decision-making auxiliary method and device based on a large model, and the method comprises the steps: obtaining and processing user problem data, and obtaining user problem information; processing the user question information by using a large model to obtain first answer information; processing the user question information by using a question solving model to obtain second answer information; and comprehensively processing the first answer information and the second answer information to obtain question answer information. The invention provides a decision-making auxiliary method and device based on a large model, and the method comprises the steps: calling an auxiliary tool based on the fusion of knowledge graph questions and answers and retrieval enhanced questions and answers by using knowledge graph data in a knowledge graph and real-time updated retrieval enhanced question and answer data; and the requirement of matching questions and answers by a professional computing tool in an auxiliary decision-making scene is met.

Description

Technical Field

[0001] The present invention relates to the field of question-answering technology, and in particular to a decision-making assistance method and device based on a large model. Background Art

[0002] With the rapid development of artificial intelligence technology, large models such as GPT (Generative Pre-trained Transformer, a series of natural language processing models based on the Transformer architecture developed by OpenAI) and DeepSeek have made significant progress in the field of natural language processing. Question answering technology, as one of the important applications of large models, aims to extract information from massive documents in an intelligent way and generate accurate answers. Existing question answering technologies based on knowledge graphs and retrieval-augmented generation (RAG) (a technical framework that combines information retrieval and generative models and aims to enhance the performance of generative models by retrieving relevant information from external knowledge bases or documents) have the following main shortcomings:

[0003] (1) Question answering based on knowledge graphs is difficult to update dynamically, knowledge graph updates are delayed, and it is difficult to support real-time question answering;

[0004] (2) Question-answering based on retrieval enhancement is overly dependent on the retrieval content, resulting in a high error rate in generated answers and a lack of explicit constraints on the generated answers;

[0005] (3) It cannot meet the needs of professional computing tools in decision-making support scenarios. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a decision-making assistance method and device based on a large model, which utilizes the high-quality data manually reviewed and verified in the knowledge graph and the retrieval-enhanced question-answering data updated in real time. On the basis of integrating knowledge graph question-answering and retrieval-enhanced question-answering, combined with auxiliary tool calls, it solves the problem of needing professional computing tools to cooperate with question-answering in decision-making assistance scenarios.

[0007] In order to solve the above technical problems, the first aspect of the embodiments of the present invention discloses a decision support method based on a large model, characterized in that the method includes:

[0008] S1, obtain and process user problem data to obtain user problem information;

[0009] The user question data includes user picture data and user text data;

[0010] S2, using the large model to process the user question information to obtain first answer information;

[0011] S3, using the problem-solving model to process the user question information to obtain second answer information;

[0012] S4: Comprehensively process the first answer information and the second answer information to obtain question answer information.

[0013] As an optional implementation manner, in the first aspect of the embodiment of the present invention, acquiring and processing user question data to obtain user question information includes:

[0014] S11, obtaining user question data;

[0015] S12, performing discrimination processing on the user question data to obtain question type information and question information;

[0016] S13, when the question type data is a picture, identifying and processing the user picture data to obtain first question information; otherwise, the question information is updated to second question information;

[0017] S14: Combine the first question information and the second question information to obtain user question information.

[0018] As an optional implementation, in the first aspect of the embodiment of the present invention, the problem-solving model includes: a task acquisition module, a task decomposition module, a task query module, a knowledge graph storage module, a retrieval vector storage module and a tool set module;

[0019] The task acquisition module is used to analyze and process the user question information to obtain task requirement information;

[0020] The task decomposition module is used to decompose the task requirement information to obtain subtask information;

[0021] The task query module is configured to process the subtask information using the toolset module based on the knowledge graph storage module and the retrieval vector storage module to obtain the second answer information;

[0022] The knowledge graph storage module is used to construct and store the knowledge graph;

[0023] The retrieval vector storage module is used to construct and store knowledge graph vector data;

[0024] The output end of the task acquisition module is data-connected to the input end of the task decomposition module; the output end of the task decomposition module is data-connected to the input end of the task query module; the output end of the toolset module and the input end of the task query module are data-connected respectively; the knowledge graph storage module, the retrieval vector storage module and the task query module are data-connected respectively.

[0025] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the using of the problem-solving model to process the user question information to obtain the second answer information includes:

[0026] S31, the task acquisition module analyzes and processes the user question letter to obtain task requirement information;

[0027] S32, the task decomposition module decomposes the task requirement information to obtain subtask information;

[0028] S33, parsing the subtask information to obtain subtask type information;

[0029] S34, when the subtask type is the first set value, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain second answer information; otherwise, execute S35;

[0030] S35 , based on the knowledge graph storage module, the retrieval vector storage module and the toolset module, the task query module is used to process the subtask information to obtain second answer information.

[0031] As an optional implementation, in the first aspect of the embodiment of the present invention, the processing of the subtask information by the task query module based on the knowledge graph storage module, the retrieval vector storage module, and the toolset module to obtain the second answer information includes:

[0032] S351, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain first query result information;

[0033] S352: Based on the retrieval vector storage module, the task query module processes the subtask information to obtain second query result information;

[0034] S353: Based on the toolset module, use the task query module to process the subtask information to obtain third query result information;

[0035] S354: Comprehensively process the first query result information, the second query result information, and the third query result information to obtain second answer information.

[0036] As an optional implementation, in the first aspect of the embodiment of the present invention, the subtask information is processed by the task query module based on the knowledge graph storage module to obtain first query result information, including:

[0037] S3511, performing query processing on the subtask information based on the knowledge graph storage module to obtain first knowledge graph query value information;

[0038] S3512: Determine whether the first knowledge graph query value is equal to a preset value, and obtain a first knowledge graph query result;

[0039] S3513: When the first knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3514;

[0040] S3514, performing enhancement processing on the subtask information to obtain subtask enhanced information;

[0041] S3515: Perform query processing on the subtask enhancement information based on the retrieval vector storage module to obtain second knowledge graph query value information;

[0042] S35136, determining whether the second knowledge graph query value is equal to a preset value, and obtaining a second knowledge graph query result;

[0043] S3517: When the second knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3518;

[0044] S3518: Use the large model to process the subtask enhancement information to obtain first query result information.

[0045] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the subtask information is processed by the task query module based on the toolset module to obtain third query result information, including:

[0046] S3531, extracting element information from the subtask information to obtain task category information and tool operation parameter information;

[0047] S3532, processing the task category information and tool operation parameter information to obtain a tool selection prompt;

[0048] S3533, the large model uses the tool selection prompt word to automatically determine the target tool to be called and obtain the tool path;

[0049] S3534: Based on the tool path, the large model calls the target tool to process the tool operation parameter information to obtain third query result information.

[0050] A second aspect of an embodiment of the present invention discloses a large-model-based decision-making auxiliary question-answering device, characterized in that the device comprises: a question information acquisition module, a first processing module, a second processing module, and a comprehensive processing module;

[0051] The question information acquisition module is used to acquire and process user question information to obtain user demand information; the user question information includes user picture data information and user text data information;

[0052] The first processing module is used to process the user demand information using a large model to obtain first answer information;

[0053] The second processing module is configured to process the user demand information using a problem-solving model to obtain second answer information;

[0054] The comprehensive processing module is used to comprehensively process the first answer information and the second answer information to obtain question answer information.

[0055] The third aspect of the present invention discloses another large-scale model-based decision-making auxiliary question-answering device, comprising:

[0056] a memory storing executable program code;

[0057] a processor coupled to the memory;

[0058] The processor calls the executable program code stored in the memory to execute part or all of the steps in the large model-based decision support method disclosed in the first aspect of the embodiment of the present invention.

[0059] The fourth aspect of the present invention discloses a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions, and when the computer instructions are called, they execute part or all of the steps in the large model-based decision support method disclosed in the first aspect of the embodiment of the present invention.

[0060] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0061] In an embodiment of the present invention, knowledge graph data in the knowledge graph and real-time updated retrieval-enhanced question-answering data are utilized, and auxiliary tools are called on the basis of integrating knowledge graph question-answering and retrieval-enhanced question-answering, thereby solving the need for professional computing tools to cooperate with question-answering in auxiliary decision-making scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 Schematic diagram of a scenario of a large-model-based decision-making support question-answering system provided by an embodiment of the present invention;

[0064] Figure 2 This is a flow chart of a decision support method based on a large model disclosed in an embodiment of the present invention;

[0065] Figure 3 This is a schematic diagram of the structure of a large-model-based decision-making support question-answering system disclosed in an embodiment of the present invention;

[0066] Figure 4 It is a structural diagram of another large-model-based decision-making auxiliary question-answering device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] The terms "first," "second," and so on, in the description and claims of the present invention and the accompanying drawings are used to distinguish between different objects, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or device.

[0069] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0070] In this application, the word "exemplary" is used to mean "serving as an example, illustration, or illustration." Any embodiment described in this application as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is given to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that one of ordinary skill in the art can recognize that the present application can be implemented without using these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0071] It should be noted that since the method of the embodiment of the present application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time is actually time information. It can be understood that if size, quantity, position, etc. are mentioned in subsequent embodiments, the corresponding data exist for the computer device to process. The details will not be repeated here.

[0072] It should be noted that the large model involved in this application is explained, and the large model refers to an artificial neural network model with a very large number of parameters. In the field of artificial intelligence, a large model generally refers to a model with hundreds of millions to trillions of parameters. The model usually needs to be trained on a large-scale data set and requires a large amount of computing resources to be optimized and adjusted. Large models are generally used to solve complex tasks such as natural language processing, computer vision, and speech recognition. Generative AI is an AI that can create new content and ideas, including conversations, stories, images, videos, and music. In the embodiment of the present application, the large model can be ChatGPT, BERT, XLNet, Zhipu model, Claude, Moonshot AI model, ChatGLM model, Qianyi Tongwen model, MiniMax model, Spark model, Llama model, 360GPT model, qwen model, Baichuan model, Skylark model, vivoLM model and Wenxin Yiyan and other large-scale language models, which are not limited in the embodiment of the present application.

[0073] The embodiments of the present application provide a decision support method, apparatus, computer device, and computer-readable storage medium based on a large model, which are described in detail below.

[0074] See also Figure 1 , Figure 1 This is a scenario diagram of the question decision system provided in an embodiment of the present application. The decision-making auxiliary question-answering system based on a large model may include a computer device 100, in which a decision-making auxiliary question-answering device based on a large model is integrated, such as Figure 1 Computer equipment in.

[0075] In the embodiments of the present application, the computer device 100 may be an independent server, or a server network or server cluster composed of servers. For example, the computer device 100 described in the embodiments of the present application includes, but is not limited to, a computer, a network host, a single network server, a set of multiple network servers, or a cloud server composed of multiple servers. A cloud server is composed of a large number of computers or network servers based on cloud computing.

[0076] It is understood that the computer device 100 used in the embodiments of the present application can be a device that includes both receiving and transmitting hardware, that is, a device that has receiving and transmitting hardware capable of performing two-way communication over a two-way communication link. Such a device may include: a cellular or other communication device that has a single-line display, a multi-line display, or a cellular or other communication device without a multi-line display. The specific computer device 100 can be a desktop terminal or a mobile terminal. The computer device 100 can also be a mobile phone, a tablet computer, a laptop computer, etc.

[0077] Those skilled in the art will understand that Figure 1 The application environment shown in the figure is only one application scenario of the present application solution and does not constitute a limitation on the application scenario of the present application solution. Other application environments may also include Figure 1 More or fewer computer devices as shown in Figure 1 Only one computer device is shown in the figure. It can be understood that the system can also include one or more other services, which are not limited here.

[0078] In addition, if Figure 1 As shown, the problem decision system may further include a memory 200 for storing simulation data, such as simulation process data and simulation result data.

[0079] It should be noted that Figure 1The scenario diagram of the problem decision system shown is only an example. The large model-based decision-making auxiliary question-answering device and scenario described in the embodiment of this application are intended to more clearly illustrate the technical solution of the embodiment of this application, and do not constitute a limitation on the technical solution provided by the embodiment of this application. Ordinary technicians in this field can know that with the evolution of the problem decision system and the emergence of new business scenarios, the technical solution provided by the embodiment of this application is also applicable to similar technical problems.

[0080] This invention discloses a large-scale model-based decision-making assistance method and device. This method utilizes knowledge graph data and real-time updated retrieval-enhanced question-and-answer data within a knowledge graph. By integrating knowledge graph question-and-answer with retrieval-enhanced question-and-answer, it utilizes auxiliary tools to address the need for specialized computing tools to support question-and-answer in decision-making assistance scenarios. These are described in detail below.

[0081] Example 1

[0082] See also Figure 2 , Figure 2 This is a flow chart of a decision support method based on a large model disclosed in an embodiment of the present invention. Figure 2 The described decision support method based on the large model is applied to a problem decision system, such as a local server or a cloud server used in the problem decision system, and the embodiment of the present invention does not limit this. Figure 2 As shown, the decision support method based on the large model may include the following operations:

[0083] S1, obtain and process user problem data to obtain user problem information;

[0084] It should be noted that the user question data refers to question information raised based on user needs;

[0085] The user question data includes user picture data and user text data;

[0086] S2, using the large model to process the user question information to obtain first answer information;

[0087] S3, using the problem-solving model to process the user question information to obtain second answer information;

[0088] S4, comprehensively processing the first answer information and the second answer information to obtain question answer information;

[0089] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention utilizes the knowledge graph data in the knowledge graph and the real-time updated retrieval-enhanced question-answering data. On the basis of integrating knowledge graph question-answering and retrieval-enhanced question-answering, auxiliary tools are called to solve the need for professional computing tools to cooperate with question-answering in auxiliary decision-making scenarios.

[0090] In another optional embodiment, in step S1 above, acquiring and processing user question data to obtain user question information includes:

[0091] S11, obtaining user question data;

[0092] It should be noted that, in this embodiment, the user question data includes image data "Xinjinyang ship image data" and text data "What is the ship in the image and what is its maximum range";

[0093] S12, performing discrimination processing on the user question data to obtain question type information and question information;

[0094] S13, when the question type data is a picture, pre-processing the user picture data to obtain first question information; otherwise, updating the question information to second question information;

[0095] S14, combining the first question information and the second question information to obtain user question information;

[0096] It should be noted that the combination processing means combining in order of precedence;

[0097] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention is implemented to obtain and process user question information to obtain user demand information, which provides data support for the subsequent integration of knowledge graph data in the knowledge graph and real-time updated retrieval enhanced question and answer data, and the calling of auxiliary tools to process user question information to obtain question answer information, thereby solving the need for professional computing tools to cooperate with question and answer in auxiliary decision-making scenarios.

[0098] In another optional embodiment, in step S13, the pre-processing of the user image data to obtain the first question information includes:

[0099] It should be noted that the preprocessing expression is:

[0100]

[0101] Wherein, (x, y) represents the pixel coordinates in the user image data; O(x, y) represents the pixel value of the image information corresponding to the pixel coordinates in the user image data after preprocessing; H represents the intensity value of the pixel in the red channel; L represents the intensity value of the pixel in the green channel; B represents the intensity value of the pixel in the blue channel; α represents the first coefficient; β represents the second coefficient; γ represents the third coefficient; μ represents the fourth coefficient; M represents the height value of the user image data; m represents the height index of the user image data; N represents the width value of the user image data; n represents the width index of the user image data; R mn Indicates the red channel intensity value of the pixel at coordinates (m, n) in the user image data;

[0102] It should be noted that, in this embodiment, α is set to 0.3; β is set to 0.59; γ is set to 0.11; μ is set to 0.01; the value ranges of H, L, and B are all (0-255);

[0103] It should be noted that the pre-processing method is used to enhance the processing of the user image data to improve the recognition and clarity of the image, thereby improving the accuracy of recognition;

[0104] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention is implemented to pre-process the user image data to obtain the first question information, providing data support for obtaining the answer information of the question, and solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0105] In another optional embodiment, in step S2, the processing of the user question information using the large model to obtain the first answer information includes:

[0106] S21, based on the thought chain of the large model, analyzing the user question information to obtain task information;

[0107] It should be noted that the thought chain refers to a series of typical steps that the human thinking process follows when solving reasoning tasks, that is, the process of breaking down a relatively complex logical problem and forming a complete thinking process through a series of logically related thoughts;

[0108] It should be noted that the thought chain based on the large model analyzes the user question information to obtain task information, including:

[0109] The large model decomposes logically complex user questions through a series of logically related thinking to obtain task information containing a series of sub-questions;

[0110] S22, querying the task information based on the large model to obtain first answer information;

[0111] It can be seen that the decision-making assistance method based on the big model described in the embodiment of the present invention is implemented, and the user question information is processed using the big model to obtain the first answer information, providing data support for obtaining the question answer information, and solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0112] In another optional embodiment, in the above step S2, the problem-solving model includes: a task acquisition module, a task decomposition module, a task query module, a knowledge graph storage module, a retrieval vector storage module and a tool set module;

[0113] The task acquisition module is used to analyze and process the user question information to obtain task requirement information;

[0114] The task decomposition module is used to decompose the task requirement information to obtain subtask information;

[0115] The task query module is configured to process the subtask information using the toolset module based on the knowledge graph storage module and the retrieval vector storage module to obtain the second answer information;

[0116] The knowledge graph storage module is used to construct and store the knowledge graph;

[0117] It should be noted that the knowledge graph includes but is not limited to the aircraft knowledge graph and the ship knowledge graph;

[0118] The retrieval vector storage module is used to construct and store knowledge graph vector data;

[0119] It should be noted that the knowledge graph vector data includes but is not limited to airport construction material data and ship introduction material data;

[0120] The output end of the task acquisition module is data-connected to the input end of the task decomposition module; the output end of the task decomposition module is data-connected to the input end of the task query module; the output end of the toolset module and the input end of the task query module are data-connected respectively; the knowledge graph storage module, the retrieval vector storage module and the task query module are data-connected respectively.

[0121] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention is implemented by utilizing the problem-solving model to integrate the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, calling the auxiliary tool to process the user question information and obtain the question answer information, thereby solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0122] In another optional embodiment, in step S3, the step of processing the user question information using the problem-solving model to obtain the second answer information includes:

[0123] S31, the task acquisition module analyzes and processes the user question information to obtain task requirement information;

[0124] It should be noted that the user question information represents the logical questions raised by the user based on the requirements;

[0125] It should be noted that the analysis process refers to performing intention recognition processing on the user question information to obtain task requirement information;

[0126] The task requirement information includes question type information, entity information and question intent information;

[0127] It should be noted that, in this embodiment, the user question information is "Xinjinyang ship image data" and "What is the ship in the image and what is its maximum range";

[0128] It should be noted that the task requirement information is “what kind of ship is in the image and what is its maximum range”;

[0129] S32, the task decomposition module decomposes the task requirement information to obtain subtask information;

[0130] It should be noted that, in this embodiment, the large model decomposes the task requirement information "a certain airline's Boeing 737 passenger plane takes off and lands at Anqing Airport" into the following four subtask information, and executes the subtasks respectively;

[0131]

[0132] S33, parsing the subtask information to obtain subtask type information;

[0133] It should be noted that the subtask type information is of knowledge graph type;

[0134] S34, when the subtask type is the first set value, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain second answer information; otherwise, execute S35;

[0135] It should be noted that, in this embodiment, the knowledge graph is a ship knowledge graph, including a ship concept system and a Xinjinyang knowledge graph:

[0136] Vessel Concept System

[0137]

[0138] New Jinyang Knowledge Graph

[0139]

[0140] It should be noted that this embodiment first performs subtask 1 and completes the following steps based on the ship knowledge graph and image recognition tool:

[0141]

[0142] Image recognition tools

[0143]

[0144] S35 , based on the knowledge graph storage module, the retrieval vector storage module and the toolset module, the task query module is used to process the subtask information to obtain second answer information.

[0145] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention is implemented by utilizing the problem-solving model to integrate the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, calling the auxiliary tool to process the user question information and obtain the question answer information, thereby solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0146] In another optional embodiment, in step S35, the subtask information is processed by the task query module based on the knowledge graph storage module, the retrieval vector storage module, and the toolset module to obtain the second answer information, including:

[0147] S351, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain first query result information;

[0148] It should be noted that, in general, knowledge graphs are stored in databases such as MySQL or Neo4J. To obtain results from the database, the corresponding SQL statement must be executed. For example, for the user question "How fast can the USS Ronald Reagan run?", the corresponding query statement in the Neo4J library is "match n where n.name = {USS Ronald Reagan (CVN-76) aircraft carrier} return n. speed". Since each question type has a different answer format, each question type has a corresponding query syntax. This solution combines the semantic understanding and code generation capabilities of the large model, comprehensively considers the user's question and the results of intent understanding, and outputs an executable knowledge graph query statement.

[0149] S352: Based on the retrieval vector storage module, the task query module processes the subtask information to obtain second query result information;

[0150] It should be noted that, based on the retrieval vector storage module, the task query module is used to process the subtask information to obtain the second query result information, indicating that a vector library query, i.e., retrieval enhancement, is performed. With the target document as the data basis, in the system, users can upload documents of this scenario to the system, and the system automatically performs text splitting and vectorization processing. Users can then conduct knowledge questions and retrieval related to the document content based on the large model;

[0151] It should be noted that, in this embodiment, the search enhancement vector library is stored as the document material of Xinjinyang, as shown in the following table:

[0152]

[0153]

[0154] It should be noted that, in this embodiment, to perform subtask 2 to query the name and maximum range information in the document materials of Xinjinyang, the following two steps need to be completed:

[0155]

[0156] S353: Based on the toolset module, use the task query module to process the subtask information to obtain third query result information;

[0157] It should be noted that in the decision-making support process, for complex business problems, the large model needs to call many small models or tools. Small models refer to models that can relatively accurately calculate and produce clear result data after certain situation data is input. They include calculation tools, statistical tools, analysis tools, AI models, etc. The accuracy of the small models is related to the accuracy of the final answer.

[0158] It should be noted that, in this embodiment, executing subtask 3 to call the length comparison tool to compare the result data of each task requires completing the following steps:

[0159]

[0160] S354: Comprehensively process the first query result information, the second query result information, and the third query result information to obtain second answer information;

[0161] It should be noted that, in this embodiment, executing subtask 4 and summarizing the generated results requires completing the following steps:

[0162]

[0163] It can be seen that the large model-based decision assistance method described in the embodiment of the present invention is implemented based on the knowledge graph storage module, the retrieval vector storage module and the toolset module, and the task query module is used to process the subtask information to obtain the second answer information. The knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data are integrated and the auxiliary tool is called to process the user question information to obtain the question answer information, which solves the need for professional computing tools to cooperate with question and answer in the decision assistance scenario.

[0164] In another optional embodiment, in the above step S351, the subtask information is processed by the task query module based on the knowledge graph storage module to obtain first query result information, including:

[0165] S3511, performing query processing on the subtask information based on the knowledge graph storage module to obtain first knowledge graph query value information;

[0166] It should be noted that the first knowledge graph query value information indicates whether the graph query result exists. When the query result exists, the first knowledge graph query value is set to 1, otherwise it is set to 0;

[0167] S3512: Determine whether the first knowledge graph query value is equal to a preset value, and obtain a first knowledge graph query result;

[0168] It should be noted that the preset value is 1;

[0169] S3513: When the first knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3514;

[0170] S3514, performing enhancement processing on the subtask information to obtain subtask enhanced information;

[0171] S3515: Perform query processing on the subtask enhancement information based on the retrieval vector storage module to obtain second knowledge graph query value information;

[0172] It should be noted that the second knowledge graph query value information indicates whether the vector information exists. When the subtask enhancement information exists, the second knowledge graph query value is set to 1, otherwise it is set to 0;

[0173] S3516, determining whether the second knowledge graph query value is equal to a preset value, and obtaining a second knowledge graph query result;

[0174] It should be noted that the preset value is 1;

[0175] S3517: When the second knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3518;

[0176] S3518: Use the large model to process the subtask enhancement information to obtain first query result information.

[0177] It can be seen that the decision-making assistance method based on the big model described in the embodiment of the present invention is implemented, based on the knowledge graph storage module, using the task query module to process the subtask information, to obtain the first query result information, to provide data support for the acquisition of the second answer information, and to integrate the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, and to call the auxiliary tool to process the user question information to obtain the question answer information, thereby solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0178] In another optional embodiment, in step S353, the subtask information is processed by the task query module based on the toolset module to obtain third query result information, including:

[0179] S3531, extracting elements from the subtask information to obtain task category information and tool operation parameter information;

[0180] The extraction of element information from the tool operation parameter information refers to extracting the task category information and tool operation parameter information in the task target using a large model and element extraction prompt words. Exemplarily, the following procedure is used for processing:

[0181] {

[0182] classification: category;

[0183] attributes: {

[0184] Parameter A, parameter B, ...

[0185] }

[0186] }

[0187] It should be noted that the task category information is determined based on the tools supported by the professional tool set. For example, the tool set includes but is not limited to statistical tools, analytical tools, and reasoning tools, so the task categories include but are not limited to statistical, analytical, and reasoning.

[0188] It should be noted that the toolset is designed to automatically call tools for large models. The tools in the toolset need to have the following information, for example, a simple integer addition tool:

[0189]

[0190] S3532, processing the task category information and tool operation parameter information to obtain a tool selection prompt;

[0191] It should be noted that for tool selection, tool selection prompts are first generated based on the result of feature extraction. The large model automatically determines the tool to be called based on the prompts. The prompts are as follows:

[0192]

[0193] S3533, the large model uses the tool selection prompt word to automatically determine the target tool to be called and obtain the tool path;

[0194] S3534: Based on the tool path, the large model calls the target tool to process the tool operation parameter information to obtain third query result information;

[0195] It can be seen that the decision-making assistance method based on the big model described in the embodiment of the present invention is implemented, based on the knowledge graph storage module, using the task query module to process the subtask information, to obtain the first query result information, to provide data support for the acquisition of the second answer information, and to integrate the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, and to call the auxiliary tool to process the user question information to obtain the question answer information, thereby solving the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0196] In another optional embodiment, in step S354, the first query result information, the second query result information, and the third query result information are comprehensively processed to obtain the second answer information, including:

[0197] S3541: parse the first query result information to obtain first subtask type information, first subtask number information, and first subtask query result information;

[0198] Parsing the second query result information to obtain second subtask type information, second subtask number information, and second subtask query result information;

[0199] Parsing the third query result information to obtain third subtask type information, third subtask number information, and third subtask query result information;

[0200] S3542: Perform logical processing on the first subtask query result information, the second subtask query result information, and the third subtask query result information to obtain second answer information;

[0201] It should be noted that the logic processing includes but is not limited to size comparison judgment, AND processing, or processing, which is not limited in this embodiment.

[0202] It can be seen that the decision-making assistance method based on the big model described in the embodiment of the present invention is implemented to comprehensively process the first query result information, the second query result information and the third query result information to obtain the second answer information, integrate the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, call the auxiliary tool to process the user question information, and obtain the question answer information, which solves the need for professional computing tools to cooperate with question and answer in the decision-making assistance scenario.

[0203] In another optional embodiment, in step S4, the comprehensive processing of the first answer information and the second answer information to obtain the question answer information includes:

[0204] S41, performing validity determination processing on the second answer information to obtain an answer determination result;

[0205] It should be noted that the validity judgment process means obtaining the validity flag value in the second answer information;

[0206] It should be noted that when the validity flag value is 1, it indicates that the second answer information is valid; otherwise, it is invalid;

[0207] S42, when the answer judgment result is valid, the second answer information is updated to the question answer information;

[0208] Otherwise, the first answer information is updated to the question answer information;

[0209] It can be seen that the large-model-based decision-making assistance method described in the embodiment of the present invention is implemented to comprehensively process the first answer information and the second answer information to obtain question answer information, integrate and utilize the knowledge graph data in the knowledge graph and the real-time updated retrieval enhanced question and answer data, call auxiliary tools to process the user question information, and obtain question answer information, which solves the need for professional computing tools to cooperate with question and answer in auxiliary decision-making scenarios.

[0210] Example 2

[0211] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a large model-based decision-making auxiliary question-answering device disclosed in an embodiment of the present invention. Figure 3 The described device can be applied to a problem decision system, such as a local server or a cloud server for a problem decision system, and the embodiment of the present invention does not limit this. Figure 3 As shown, the device may include: a problem information acquisition module 101, a first processing module 102, a second processing module 103 and a comprehensive processing module 104;

[0212] The question information acquisition module 101 is used to acquire and process user question information to obtain user demand information; the user question information includes user picture data information and user text data information;

[0213] The first processing module 102 is used to process the user demand information using a large model to obtain first answer information;

[0214] The second processing module 103 is configured to process the user demand information using a problem-solving model to obtain second answer information;

[0215] The comprehensive processing module 104 is used to comprehensively process the first answer information and the second answer information to obtain question answer information.

[0216] Example 3

[0217] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of a large model-based decision-making auxiliary question-answering device disclosed in an embodiment of the present invention. Figure 4 The described device can be applied to a multi-sample simulation control management system, such as a local server or cloud server for a multi-sample simulation control system, and the embodiment of the present invention does not limit this. Figure 4As shown, the device may include:

[0218] A memory 201 storing executable program code;

[0219] a processor 202 coupled to the memory 201;

[0220] The processor 202 calls the executable program code stored in the memory 201 to execute the steps of the large model-based decision support method described in the first embodiment.

[0221] Example 4

[0222] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the large model-based decision-making auxiliary question answering described in the first embodiment.

[0223] Example 5

[0224] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps in the large model-based decision support method described in Example 1.

[0225] The device embodiments described above are merely illustrative. Modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0226] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus the necessary general hardware platform, or of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, the storage medium including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0227] Finally, it should be noted that the large-scale model-based decision-making assistance method and device disclosed in the embodiments of the present invention are only preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A decision support method based on a large model, characterized in that: The method comprises: S1, obtain and process user problem data to obtain user problem information; The user question data includes user picture data and user text data; S2, using the large model to process the user question information to obtain first answer information; S3, using the problem-solving model to process the user question information to obtain second answer information; S4: Comprehensively process the first answer information and the second answer information to obtain question answer information.

2. The decision support method based on a large model according to claim 1, characterized in that: The acquiring and processing of user question data to obtain user question information includes: S11, obtaining user question data; S12, performing discrimination processing on the user question data to obtain question type information and question information; S13, when the question type data is a picture, identifying and processing the user picture data to obtain first question information; otherwise, the question information is updated to second question information; S14: Combine the first question information and the second question information to obtain user question information.

3. The decision support method based on a large model according to claim 1, characterized in that: The problem-solving model includes: a task acquisition module, a task decomposition module, a task query module, a knowledge graph storage module, a retrieval vector storage module and a tool set module; The task acquisition module is used to analyze and process the user question information to obtain task requirement information; The task decomposition module is used to decompose the task requirement information to obtain subtask information; The task query module is configured to process the subtask information using the toolset module based on the knowledge graph storage module and the retrieval vector storage module to obtain the second answer information; The knowledge graph storage module is used to construct and store the knowledge graph; The retrieval vector storage module is used to construct and store knowledge graph vector data; The output end of the task acquisition module is data-connected to the input end of the task decomposition module; the output end of the task decomposition module is data-connected to the input end of the task query module; the output end of the toolset module and the input end of the task query module are data-connected respectively; the knowledge graph storage module, the retrieval vector storage module and the task query module are data-connected respectively.

4. The decision support method based on a large model according to claim 3, characterized in that: The method of processing the user question information using the problem-solving model to obtain second answer information includes: S31, the task acquisition module analyzes and processes the user question information to obtain task requirement information; S32, the task decomposition module decomposes the task requirement information to obtain subtask information; S33, parsing the subtask information to obtain subtask type information; S34, when the subtask type is the first set value, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain second answer information; otherwise, execute S35; S35 , based on the knowledge graph storage module, the retrieval vector storage module and the toolset module, the task query module is used to process the subtask information to obtain second answer information.

5. The decision support method based on a large model according to claim 4, characterized in that: The step of processing the subtask information using the task query module based on the knowledge graph storage module, the retrieval vector storage module, and the toolset module to obtain second answer information includes: S351, based on the knowledge graph storage module, using the task query module to process the subtask information to obtain first query result information; S352: Based on the retrieval vector storage module, the task query module processes the subtask information to obtain second query result information; S353: Based on the toolset module, use the task query module to process the subtask information to obtain third query result information; S354: Comprehensively process the first query result information, the second query result information, and the third query result information to obtain second answer information.

6. The decision support method based on a large model according to claim 5, characterized in that: The task query module is used to process the subtask information based on the knowledge graph storage module to obtain first query result information, including: S3511, performing query processing on the subtask information based on the knowledge graph storage module to obtain first knowledge graph query value information; S3512: Determine whether the first knowledge graph query value is equal to a preset value, and obtain a first knowledge graph query result; S3513: When the first knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3514; S3514, performing enhancement processing on the subtask information to obtain subtask enhanced information; S3515: Perform query processing on the subtask enhancement information based on the retrieval vector storage module to obtain second knowledge graph query value information; S35136, determining whether the second knowledge graph query value is equal to a preset value, and obtaining a second knowledge graph query result; S3517: When the second knowledge graph query result is yes, obtain the first query result information and execute S33; otherwise, execute S3518; S3518: Use the large model to process the subtask enhancement information to obtain first query result information.

7. The decision support method based on a large model according to claim 5, characterized in that: The step of processing the subtask information using the task query module based on the toolset module to obtain third query result information includes: S3531, extracting element information from the subtask information to obtain task category information and tool operation parameter information; S3532, processing the task category information and tool operation parameter information to obtain a tool selection prompt; S3533, the large model uses the tool selection prompt word to automatically determine the target tool to be called and obtain the tool path; S3534: Based on the tool path, the large model calls the target tool to process the tool operation parameter information to obtain third query result information.

8. A decision support device based on a large model, characterized in that: The device includes: a problem information acquisition module, a first processing module, a second processing module and a comprehensive processing module; The question information acquisition module is used to acquire and process user question information to obtain user demand information; the user question information includes user picture data information and user text data information; The first processing module is used to process the user demand information using a large model to obtain first answer information; The second processing module is configured to process the user demand information using a problem-solving model to obtain second answer information; The comprehensive processing module is used to comprehensively process the first answer information and the second answer information to obtain question answer information.

9. A decision-making auxiliary question-answering device based on a large model, characterized in that: The device comprises: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the large model-based decision support method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the large model-based decision support method according to any one of claims 1 to 7.

Images