Big language model multi-agent system for security industry in information and creative executable environment

Abstract

The invention relates to a creative executable environment security industry big language model multi-agent system, and the system comprises a first agent module and a second agent module which are used for carrying out the game playing of a first analysis result and a second analysis result obtained after the analysis of the same task based on the memory data obtained from a memory module, at least one agent module obtains events corresponding to different times from the memory module based on the index time and the event occurrence time corresponding to the data stored in the memory module for game playing; or at least one agent module obtains the memory data with the importance ranking reaching the preset threshold value from the memory module, and games based on the obtained memory data; the third agent module carries out adjustment and analysis based on the game result to obtain a decision result, and the third agent module is an advanced agent module with the level higher than that of the first agent module and that of the second agent module.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a multi-agent system of a large language model for the securities industry in an innovative executable environment. Background Art

[0002] The securities and futures industry is a complex and highly interactive field. The combination of large language models and multi-agent systems has brought new possibilities to the securities and futures industry. The universal large language model can process and generate natural language text, providing the securities and futures industry with more intelligent information processing and decision support capabilities, while providing a major breakthrough in natural language communication between multiple agents, thereby improving efficiency.

[0003] However, applying large language models and multi-agent systems to the securities and futures industry also faces some challenges, such as data privacy protection, system stability, algorithm transparency, etc. Therefore, developing an effective collaborative control algorithm and system is crucial to achieve the good application of large language models and multi-agent systems in the securities and futures industry. Summary of the invention

[0004] One purpose of this application is to propose a multi-agent system with a large language model for the securities industry in a trusted and executable environment.

[0005] According to one embodiment of the present application, a multi-agent system of a securities industry large language model in a trustworthy executable environment is provided, and the system includes:

[0006] The first agent module analyzes the current task based on the first information obtained to obtain a first analysis result; the second agent module analyzes the current task based on the second information obtained to obtain a second analysis result; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information; wherein the first information, the second information and the third information are different types of information;

[0007] The first agent module and the second agent module play a game on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data obtained from the memory module to obtain a game result, wherein at least one agent module plays a game on events corresponding to different times from the memory module based on the index time and the event occurrence time corresponding to the data stored in the memory module, wherein based on the formula Get the time events corresponding to different times, δ E Represents the time t at which the event occurred in the memory unit P With the inquiry time t EThe difference, Q l Indicates the number of days, Used to obtain time events corresponding to different required times; or, at least one of the agent modules obtains memory data whose importance ranking reaches a preset threshold from the memory module, and conducts a game based on the obtained memory data;

[0008] The third agent module performs adjustment analysis based on the game result to obtain a decision result, wherein the third agent module is a high-level agent module with a higher level than the first agent module and the second agent module.

[0009] In one embodiment, the at least one agent module obtains memory data whose importance ranking reaches a preset threshold from the memory module, and performs a game based on the obtained memory data, including:

[0010] The importance of the data stored in the memory module Determined based on a piecewise scoring function based on the weights of different events And the degradation rate α l and the time difference δ E Determine that the longer the event in the database occurs, the greater the corresponding weight, and the degradation rate α l It is used to measure the importance difference between different events. E It is the time interval between the event occurrence time and the knowledge base search query time.

[0011] In one embodiment, one or more of the first proxy module, the second proxy module, and the third proxy module indexes data from the memory module based on a vector index;

[0012] One or more of the first agent module, the second agent module and the third agent module play a game based on the data indexed from the memory module.

[0013] In one embodiment, indexing data from the memory module in a vector-based indexing manner includes:

[0014] The proxy module determines the similarity between the data to be indexed and the data stored in the memory module, and indexes the data from the memory module based on the similarity; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module; and / or,

[0015] The method of indexing data from the memory module based on vector indexing includes:

[0016] The proxy module determines the index time corresponding to the index data and the event occurrence time corresponding to the data stored in the memory module, and obtains events corresponding to different times from the memory module based on the index time and the event occurrence event; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module.

[0017] In one embodiment, the system further comprises:

[0018] sorting the memory data based on the importance of the memory data stored in the memory module;

[0019] The agent module obtains memory data whose importance ranking reaches a preset threshold from the memory module, and performs game based on the obtained memory data. The agent module includes one or more of the first agent module, the second agent module and the third agent module.

[0020] In one embodiment, the first agent module and the second agent module play a game based on the first analysis result and the second analysis result obtained after analyzing the same task, and obtain a game result, including:

[0021] The first agent module and the second agent module analyze the same task, obtain the first analysis result and the second analysis result, and reach an agreement after negotiation to obtain a game result, or,

[0022] The third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on a preset weight, and obtains a game result when the game between the first agent module and the second agent module meets a preset game condition.

[0023] In one embodiment, the first agent module and the second agent module analyze the same task, obtain the first analysis result and the second analysis result, and reach an agreement after negotiation to obtain a game result, including:

[0024] The first agent module and the second agent module bargain on the first analysis result and the second analysis result based on each other's views and the knowledge they have acquired; in response to the third agent module judging that the first agent module and the second agent module have reached an agreement, it is determined that the first agent module and the second agent module have reached an agreement through negotiation, and a bargaining result is obtained;

[0025] The third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on a preset weight, and when the game between the first agent module and the second agent module meets the preset game condition, obtains a game result, including:

[0026] Based on the importance of different opinions and factors generated by the first agent module and the second agent module during the game process, weights corresponding to the different opinions and factors are determined; the third agent module performs weighted processing on the opinions generated by the first agent module and the second agent module during the game process based on the weights, and obtains a game result when the game between the first agent module and the second agent module meets a preset number of game rounds or a preset game time.

[0027] In one embodiment, the third agent module performs adjustment analysis based on the game result to obtain a decision result, including:

[0028] The third agent module generates an initial solution based on the same task, and consults and gives feedback on suggestions for the initial solution to the first agent module and the second agent module;

[0029] The third agent module makes a decision to obtain a decision result based on the suggestions for the initial solution fed back by the first agent module and the second agent module.

[0030] In one embodiment, the agent module includes a brain module, a perception module and an action module, and the agent module includes one or more of a first agent module, a second agent module or a third agent module;

[0031] The brain module includes giving different roles to the large language model based on Prompt engineering technology, guiding the large language model to adopt procedures to solve problems;

[0032] The perception module is used to receive and understand information from the external environment or other agent modules, and process and interpret the input data so that the agent module can understand and respond accordingly, wherein the input data includes text, images, sounds or other forms of information;

[0033] The action module is used to perform corresponding actions based on the results obtained after reasoning and decision-making based on the information received by the perception module. The actions include controlling the movement of the robot, generating text replies, and performing one or more of specific tasks.

[0034] The present application also provides a computer-readable storage medium having a computer program or instruction stored thereon, characterized in that when the computer program or instruction is executed by a processor, the steps of the system provided in any one of the above embodiments are implemented, and the system includes: a first agent module analyzes a current task based on the acquired first information to obtain a first analysis result; a second agent module analyzes the current task based on the acquired second information to obtain a second analysis result; the first agent module and the second agent module, based on the memory data acquired from the memory module, play a game on the first analysis result and the second analysis result obtained after analyzing the same task to obtain a game result; the third agent module adjusts and analyzes the game result to obtain a decision result; wherein the memory module includes short-term memory data for storing real-time data, and long-term memory data for storing historical data; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information.

[0035] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0037] Figure 1 This is a flow chart of a large language model multi-agent system based on long short-term memory modules in the securities and futures industry provided by an embodiment of the present application;

[0038] Figure 2 This is a partial schematic diagram of a multi-agent brain provided by an embodiment of the present application;

[0039] Figure 3 is a schematic diagram of a memory module provided by an embodiment of the present application;

[0040] Figure 4 is a flowchart of decision-making based on a multi-agent module and a memory module provided in another embodiment of the present application;

[0041] Figure 5 It is a module diagram of a large language model multi-agent system based on long short-term memory modules in the securities and futures industry provided in another embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0044] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0045] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0046] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0047] In one embodiment, Figure 1 As shown, a multi-agent system of a large language model for the securities industry in a trustworthy executable environment is provided, and the system includes:

[0048] Step 101: A first agent module analyzes a current task based on acquired first information to obtain a first analysis result.

[0049] Step 102: The second agent module analyzes the current task based on the acquired second information to obtain a second analysis result.

[0050] The first proxy module and the second proxy module respectively include large models trained based on the first information and the second information.

[0051] The first information and the second information mentioned above refer to different types of information. For example, the first information may be public opinion information, and the second information may be market information. In this way, the general large model obtained by training with different types of information has different functions, and can analyze problems from different angles to obtain answers based on different angles.

[0052] The first information may be public opinion information. The first agent module performs public opinion analysis based on the acquired public opinion information to obtain a first analysis result, and the first agent module includes a general large model trained based on the public opinion information.

[0053] The second information may be market information. The second agent module performs market analysis based on the acquired market information to obtain a second analysis result, and the second agent module includes a general large model trained based on the market information.

[0054] The current task may be a task that currently requires a decision, for example, a task of “buy”, “sell” or “hold” for a certain stock.

[0055] Step 103: The first agent module and the second agent module conduct a game on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data obtained from the memory module to obtain a game result.

[0056] Since the large models included in the first agent module and the second agent module are obtained based on different data training, the task results obtained after the first agent module and the second agent module process the same task are obtained based on different perspectives, and the two may be different. In some embodiments, the results obtained after the first agent module and the second agent module process the same task can be gambled to obtain the game results. In some embodiments, the game can be a one-round game or a multi-round game, which is not limited here.

[0057] The memory module includes short-term memory data for storing real-time data and long-term memory data for storing historical data.

[0058] In some embodiments, the memory module is used to store the public opinion information and the market information in a structured manner; the memory module includes long-term memory and short-term memory; the long-term memory is used to store historical data, and the short-term memory is used to store real-time data.

[0059] In some embodiments, the memory module can be used to store real-time news information and information required for market analysis, so that the large language model can simulate human memory. For example, long-term memory can be used to store company reports, financial data, transaction history data, etc., and short-term memory may be used to store current news, social media public opinion, real-time market data, etc.

[0060] In some embodiments, this information may be stored in a memory module in a structured form. For example, the structured data may be in a key-value pair format such as (Q, A).

[0061] In some embodiments, the system further includes: the first agent module and the second agent module play a game on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data obtained from the memory module to obtain a game result. Since the memory module stores short-term memory and long-term memory, the first agent module and the second agent module can play a game based on the latest memory data obtained from the memory module.

[0062] Step 104: The third agent module performs adjustment analysis based on the game result to obtain a decision result. The third agent module includes a large model trained based on the third information.

[0063] The general large model included in the third agent module is trained based on the third information, and the first information, the second information and the third information are different types of information. For example, see Figure 2 Based on Prompt engineering technology, different roles can be given to large language models to guide them to use procedures to solve problems.

[0064] In some embodiments, the first agent module, the second agent module, and the third agent module are different agent modules, and the third agent module is a high-level agent module with a higher level than the first agent module and the second agent module.

[0065] In some embodiments, the big models in the first agent module, the second agent module and the third agent module are different functional roles played by the same big model or are the same big model with different functional roles.

[0066] In some embodiments, the first information, the second information and the third information are different types of information, and the large models in the first agent module, the second agent module and the third agent module are trained based on different types of information, and different large models make decisions on the same task based on the different types of information they have obtained.

[0067] After the first agent module and the second agent module have played a game on the same task, the third agent module can further analyze the game results to obtain the final decision result. In this way, the decision on the same task is jointly achieved through three different types of agent modules, which enables the task to be decided from different angles, and achieves a more intelligent and reasonable decision.

[0068] In the above embodiment, the first agent module, the second agent module and the third agent module can be understood as different agent modules, which are respectively used to act as agents for different matters. The three agent modules work together to complete the decision-making process of the business.

[0069] In some embodiments, the first information includes public opinion information, the second information includes behavior information, the first agent module includes a public opinion analysis investor Agent, the second agent module includes a market investor Agent, and the third agent module includes an advanced decision support system Agent.

[0070] In some embodiments, the first agent module can be used to process public opinion related information. For example, the first agent module can be a public opinion analysis investor Agent, which obtains public opinion information and analyzes the impact of public opinion on stock price trends.

[0071] In some embodiments, the second agent module may be a market investor Agent, which performs market analysis and transaction decisions based on market data and technical indicators.

[0072] Moreover, the first agent module and the second agent module may engage in a game based on their respective analysis results, and specifically, the public opinion analysis investor Agent may engage in a game with the market investor Agent.

[0073] In some embodiments, the third agent module may be a decision support system Agent, which extracts key information from the game results of the two agents corresponding to the first agent module and the second agent module, comprehensively considers the results of public opinion analysis and market analysis, as well as the arguments and rebuttals of the two parties in the game, to obtain the final decision result.

[0074] In the above embodiments, the multi-agent system can simulate the behaviors of multiple participants in the market, realize intelligent decision-making and resource allocation, and thus play an important role in securities and futures trading.

[0075] In some embodiments, the system further comprises: indexing data from the memory module based on a vector index. One or more of the first agent module, the second agent module and the third agent module index data from the memory module based on a vector index; and one or more of the first agent module, the second agent module and the third agent module conducts a game based on the data indexed from the memory module.

[0076] In order to quickly retrieve and analyze the stored information, indexing technology can be used to index the data, and content can be obtained through vector retrieval, so that the required data can be quickly indexed from the memory module.

[0077] In some embodiments, the vector index-based method for indexing data from the memory module includes: the proxy module determines the similarity between the data to be indexed and the data stored in the memory module, and indexes the data from the memory module based on the similarity; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module.

[0078] In order to quickly retrieve and analyze the stored information, indexing technology is used to index the data to speed up the retrieval and query efficiency of the data. Content can be obtained through vector retrieval, as shown in formula (1).

[0079]

[0080] As shown in formula (1), it represents the memory event m E The embedding vector of the text content and the hint query m p The cosine similarity between .

[0081] In some embodiments, the vector index-based method for indexing data from the memory module includes: the proxy module determines the index time corresponding to the indexing data, and the event occurrence time corresponding to the data stored in the memory module, and obtains events corresponding to different times from the memory module based on the index time and the event occurrence event; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module.

[0082] As shown in formula (2) and formula (3), it is used to obtain events at different times.

[0083] δ E =t P -t E (2)

[0084] Among them, δ EIndicates the time difference between the event occurrence time and the query time in the memory unit. P With t E They represent the occurrence and query time of events in the memory unit respectively.

[0085]

[0086] Among them, Q l Indicates the number of days, δ E It represents the difference between the time when the event occurs in the memory unit and the query time. Through this formula, the time events corresponding to the required different times can be obtained.

[0087] In the above embodiment, the knowledge required for a specific task is first stored in the database. In addition, the stored knowledge information provides indexing and retrieval functions to serve the general large model. There are two retrieval methods, see formula (1) to formula (3) for details. It is realized that the problem is embedded based on the vector model as in formula (1), and the similarity is compared with the embedding of the knowledge base content. It is also possible to query some timely or non-timely knowledge through formulas (2) and (3).

[0088] like Figure 3 FIG. 1 is a schematic diagram of a memory module provided in one embodiment of the present application. Figure 3 As shown in the left figure, the main functions of the memory module include data storage and management, content indexing and retrieval, content processing and analysis, and real-time update and synchronization functions. The memory module can be divided into long-term memory and short-term memory. Long-term memory mainly stores data including company report information, macroeconomic data, national policies, and upstream and downstream connections of enterprises. These data can provide support for investors who value market conditions. Short-term memory mainly stores real-time news, social network public opinion, stock market data, and futures market data. These data can support investors who value public opinion. In this way, agents of different roles can obtain timely information to support their own views.

[0089] In some embodiments, the long-term memory and short-term memory information stored in the memory unit may be further processed and analyzed to provide useful information for decision making.

[0090] In some embodiments, the system also includes: sorting the memory data based on the importance of the memory data stored in the memory module; the agent module obtains the memory data whose importance ranking reaches a preset threshold from the memory module, and conducts a game based on the obtained memory data, and the agent module includes one or more of the first agent module, the second agent module and the third agent module.

[0091] The processing and analysis of the above content refers to sorting the relevant data. For a large amount of data, the importance of the general large model will be sorted according to the data, and then the top-K memory events will be obtained to give the general large model. This information contains insights and emotions that are key to a specific task. The importance score is shown in formula (4).

[0092]

[0093] in, It refers to a unified piecewise scoring function (Formula 5), ​​which gives a weight to different events in the database. The larger the value, the longer the event occurred in the database, and the smaller the value, the shorter the event occurred. l is the degradation rate, α l There are two values, short and long, where α short <α long , α short You can take 0.9, α long It can be taken as 0.988. This is a measure of the importance difference between different times. E It refers to the time interval between the event occurrence time and the knowledge base search query time (as shown in Formula 2). Indicates the decreasing importance of an event over time.

[0094] Short-term memory (p1, p2, p3) can be (0.8, 0.15, 0.05), and long-term memory (p1, p2, p3) can be (0.05, 0.15, 0.8). This means that if one agent pays more attention to the data in the short-term memory, (p1, p2, p3) will be (0.8, 0.15, 0.05). The probability of taking 40 is high, and more attention is paid to events that occur in a shorter time. If one agent pays more attention to the data in long-term memory, (p1, p2, p3) can be (0.05, 0.15, 0.8). The probability of taking 80 will be higher, and more attention will be paid to events that take a longer time to occur.

[0095] The above-mentioned real-time updating and synchronization requires continuous updating of the memory data in the memory module in order to ensure the timeliness of the knowledge in the database.

[0096] In some embodiments, the system further includes real-time updating and synchronization of the first information, the second information, and the third information in the memory module. For example, public opinion information and market information are updated in real time and need to be synchronized with the actual market. In some embodiments, the memory unit has the ability to update and synchronize in real time to ensure that the information stored therein is the latest.

[0097] In some embodiments, the system further includes an execution module, which is used to formulate specific transaction decisions and convert them into actual transaction activities based on the relevant information stored in the memory unit and the results analyzed by the advanced agent. It mainly includes the following parts:

[0098] Decision making. Based on the comprehensive analysis results of the advanced agent and the information in the memory unit, the decision module can use algorithms and models to make specific trading decisions. This may involve making decisions to buy, sell or hold specific securities.

[0099] Risk management. The decision-making module needs to consider risk management factors to ensure that the trading decisions made are in line with the investor's risk preferences and fund management strategies, which may involve setting stop-loss points, diversifying the portfolio, and considering market liquidity.

[0100] Real-time monitoring. The decision-making module needs to monitor the results of transaction execution in real time to ensure that the transaction is executed as expected, and can timely optimize and adjust the decision-making process through learning and feedback.

[0101] like Figure 4 As shown, it is a flowchart of decision-making based on multi-agent modules and memory modules provided in one of the embodiments of the present application. The execution module refers to the result after the game for task decision-making. First, the data is stored in the long-term memory and short-term memory modules. The long-term memory and short-term memory data are acquired for similarity and timeliness. The acquired information is obtained according to the importance of the Top-K events and given to the general large model for game. The advanced agent combines the database knowledge and the results of the multi-agent debate to make corresponding decision operations. In the field of investment, it can decide whether to "buy", "sell" or "keep" in a certain period of time. Carry out corresponding risk management and real-time monitoring.

[0102] like Figure 5As shown in the figure, it is a module diagram of the algorithm and system based on the large language model and multi-agent collaborative control. The system includes a multi-agent brain module, a memory module and an execution module. The multi-agent brain module mainly uses the natural language processing and text generation capabilities of the general large model as the brain part of the system, which can realize the understanding, analysis and generation of complex information in the securities and futures industry, thereby providing powerful intelligent decision-making support for the multi-agent system. The memory module mainly stores the relevant information required for the task, because the general large model is trained by large-scale data, but this data has lags, and the large model cannot learn the latest information. Therefore, in order to address this shortcoming, the general large model can query the information of the memory module for understanding. The memory module includes two parts: long-term memory and short-term memory. This hierarchical structure is used to adapt to the different time sensitivities inherent in different types of securities and futures data. The execution module effectively integrates the analysis results of the general large model and the relevant content of the memory module to support wise decision-making.

[0103] In some embodiments, the memory module can memorize information related to the discussion task, which can come from the knowledge base or the search engine to enhance the timeliness of the system. This part includes two modules: long-term memory and short-term memory. Each agent can search for information related to the memory module and make corresponding responses based on this information and its own role. The execution module mainly makes task decisions based on the results of multi-agent collaboration.

[0104] In some embodiments, the third agent module performs adjustment analysis based on the game result to obtain a decision result, including: the third agent module generates an initial solution based on the task of the game, and obtains suggestions for the initial solution from the first agent module and the second agent module; the third agent module makes a decision based on the suggestions for the initial solution received from the first agent module and the second agent module to obtain a decision result.

[0105] Specifically, the third agent module can be a senior agent role, which generates an initial solution and then consults and feedbacks each other participant (such as the first agent module and the second agent module). Participants are also encouraged to evaluate the initial solution and propose amendments. Finally, the senior agent role summarizes and provides reference for decision-making. Since the third agent module includes a large model trained based on the third information, the third information has a more advanced decision-making ability and has the ability to make comprehensive decisions based on multiple suggestions. Therefore, the final decision obtained through the third agent module is more reasonable and effective.

[0106] like Figure 2As shown, it is a schematic diagram of the multi-agent brain part provided in one of the embodiments of the present application. The multi-agent brain part uses a large language model to simulate different roles. Through the prompt project, different personalities are given to the large model to play different roles. In the prompt, you can introduce the background of the task, the goal of the task, the role played in the task, what skills you have, give some examples, etc. Different roles play games for the same task, synergistically combining the advantages and knowledge of multiple brains to improve the ability to solve complex tasks. At the same time, a high-level agent is set up to conduct a specific analysis of the task, and then assigned to different agents for game play, and finally the game results are collected and summarized to help decision-making.

[0107] In some embodiments, the first agent module and the second agent module conduct a game based on the first analysis result and the second analysis result obtained after analyzing the same task to obtain a game result, including: the first agent module and the second agent module reach an agreement after negotiation based on the first analysis result and the second analysis result obtained after analyzing the same task to obtain the game result, or the third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on a preset weight, and when the game between the first agent module and the second agent module meets a preset game condition, the game result is obtained.

[0108] In some embodiments, the first agent module and the second agent module analyze the same task, obtain the first analysis result and the second analysis result, and reach an agreement after negotiation to obtain a game result, including: the first agent module and the second agent module each negotiate the first analysis result and the second analysis result based on the other party's views and the knowledge respectively acquired; in response to the third agent module judging that the first agent module and the second agent module reach an agreement, it is determined that the first agent module and the second agent module reach an agreement through negotiation to obtain a game result.

[0109] For example, the first agent module and the second agent module negotiate on a specific matter. Negotiation means that the two agents will discuss in combination with the other agent's views and their own knowledge during the discussion, and finally reach a consensus or make a compromise. When both parties meet and accept the consistent decision or plan, it means that they have reached a consensus and the negotiation is over. In some embodiments, the third agent module can be used to determine whether the two parties have reached a consensus to determine whether the game is over.

[0110] In the process of game, the knowledge acquired by the first agent module and the second agent module may be the latest real-time information, so that the first agent module and the second agent module can play the game based on the latest acquired data, making the game more reasonable and accurate. In other embodiments, the knowledge acquired by the first agent module and the second agent module may also be historically stored information, so that the corresponding knowledge can be acquired more quickly, and the game can be played based on the acquired knowledge to obtain the final decision.

[0111] In some embodiments, the third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on preset weights, and obtains a game result when the game between the first agent module and the second agent module meets preset game conditions, including: determining the weights corresponding to different opinions and factors based on the importance of different opinions and factors generated by the first agent module and the second agent module during the game; the third agent module weights the opinions generated by the first agent module and the second agent module during the game based on the weights, and obtains a game result when the game between the first agent module and the second agent module meets a preset number of game rounds or a preset game time.

[0112] For example, after the first agent module and the second agent module have gone through multiple rounds of game, the third agent module will weight the opinions discussed by both parties according to different weights. The above weights can refer to numerical values ​​that quantify the importance of different opinions and factors. The determination of weights can be determined by subjective experience or by objective decision-making based on data analysis. The above weighting is used to tell the third agent module to comprehensively consider the opinions of all parties according to the weights, so as to better consider the importance of the opinions of all parties. The end of this part of the game can determine whether the game is over by meeting a specific number of rounds or time limits.

[0113] In some embodiments, the third agent module may be a decision support system Agent, which extracts key information from the game results of the two agents corresponding to the first agent module and the second agent module, comprehensively considers the results of public opinion analysis and market analysis, as well as the arguments and rebuttals of the two parties in the game, to obtain the final decision result.

[0114] In the above embodiments, the multi-agent system can simulate the behaviors of multiple participants in the market, realize intelligent decision-making and resource allocation, and thus play an important role in securities and futures trading.

[0115] The large language model has powerful natural language processing and generation capabilities. The Agent based on the large language model in the above embodiment can use the large language model as a brain module, so the above agent module has powerful natural language processing and generation capabilities. The collaboration of multiple agents in the above embodiment can be a collaborative interaction of information between multiple intelligent entities to jointly complete a task.

[0116] In some embodiments, the agent module Agent can be an intelligent entity that can perceive the environment, make decisions and execute. The agent module includes a brain module, a perception module and an action module. The brain module includes giving different roles to the large language model based on Prompt engineering technology, and guiding the large language model to adopt procedures to solve problems. The perception module is used to receive and understand information from the external environment or other agent modules, and process and interpret the input data so that the agent module is used to understand and respond accordingly, wherein the input data includes text, images, sounds or other forms of information. The action module is used to perform corresponding actions based on the results obtained after reasoning and decision-making based on the information received by the perception module, and the actions include controlling the movement of the robot, generating text replies, and performing one or more of specific tasks.

[0117] When a large language model uses a program to solve a problem, the program can include three parts: role identification, game, and multi-round iterative collaboration. The following is an example of a decision-making scenario in the securities and futures industry. The details are as follows:

[0118] Role identification. For two different types of investors, public opinion analysis and market analysis, two agents are designed, namely public opinion analysis investor agent and market investor agent. Among them, public opinion analysis investor agent can obtain public opinion information from a large amount of text data such as news and social media, and analyze the impact of public opinion on stock price trends. Market investor agent can perform market analysis and trading decisions based on market data and technical indicators. At the same time, an advanced agent is introduced as a decision support system to extract key information from the game results of the two agents, comprehensively consider the results of public opinion analysis and market analysis, as well as the arguments and rebuttals of both parties in the game, so as to obtain the final result. Multiple large language models can play different roles for different tasks and are widely used.

[0119] Gaming. After identifying different participants, senior agents are considered to be the leaders who initiate collaboration and generate initial solutions. Before the final answer is generated, public opinion analysis investor agents and market investor agents gamble on the trend of stocks. Public opinion analysis investor agents may focus on information such as the company's public opinion, industry dynamics, and management performance, while market investor agents may pay more attention to information such as market trends, technical indicators, and trading volume.

[0120] Multi-round iterative collaboration. After multiple rounds of game, the condition for the end of the game can be that both parties reach a consensus on a specific issue, which means the game is over, or the game reaches a specific number of rounds and time limit, thus ending the game. The public opinion analysis investor agent and the market investor agent can make an investment decision based on the results and comprehensive analysis of the game. Both parties consider each other's views and their own knowledge during the game process and finally reach a joint investment decision. This decision can be "buy", "sell" or "hold" for a certain stock. This decision may be reached through negotiation or it may be based on a comprehensive consideration of the opinions of both parties according to a certain weight. The senior agent role will generate an initial solution and then consult and feedback from each other participant. Participants are also encouraged to evaluate the initial solution and propose modifications. Finally, the senior agent role summarizes it for reference in decision-making.

[0121] The securities and futures industry is a field full of complexity and high interaction. The combination of large language models and multi-agent systems has brought new possibilities to the securities and futures industry. The universal large language model can process and generate natural language text, providing the securities and futures industry with more intelligent information processing and decision support capabilities, while providing a great breakthrough in the communication of natural language between multiple agents, thereby improving efficiency. At the same time, compared with the lack of single-agent systems in terms of collaboration with other agents and the ability to acquire knowledge from social interactions. Multi-agent systems allow different intelligent agents (such as trading algorithms, risk management systems, etc.) to collaborate, compete and learn with each other to achieve more effective market transactions and risk control.

[0122] In the above embodiments, in the securities and futures industry, large language models are used in market intelligence analysis, public opinion monitoring, financial news reporting, etc., to help financial institutions better understand market dynamics and decision-making basis. At the same time, securities and futures trading involves multiple parties and complex trading decision-making processes, so multi-agent collaborative control algorithms become an important tool for optimizing trading strategies, improving trading efficiency and reducing risks. Multi-agent systems can simulate the behavior of multiple participants in the market, realize intelligent decision-making and resource allocation, and thus play an important role in securities and futures trading.

[0123] In the above embodiment, a multi-agent collaborative control algorithm is used to solve relevant tasks in the securities and futures field. Combining the natural language processing and understanding capabilities of a large language model and the decision-making negotiation mechanism of multi-agent collaborative control, the system can comprehensively consider more information and interests, and use group wisdom to improve the accuracy of decision-making and global optimization capabilities. In the task of the impact of public opinion and market conditions on investment decisions, the system not only performs well in historical data, but can also be used for future data. At the same time, the multi-agent collaborative control algorithm can define different roles for discussion for different tasks, and is widely used, such as: international crude oil price analysis, futures trading and other fields, with strong versatility and applicability.

[0124] The system provided in the above embodiments can be applied to securities and futures scenarios. For example, the system provided in this application can be a securities and futures industry large language model multi-agent collaborative control algorithm and system. The system mainly includes three core structures: a multi-agent brain module, a memory module, and an execution module. Among them, the multi-agent brain module gives different personalities to the large language model to play different roles for discussion or game. In the discussion on the impact of public opinion and market conditions on investment decisions in the securities industry, the multi-agent brain module can include two agents, representing the characteristics, decision-making preferences and behavior patterns of two different types of investors, public opinion analysis and market analysis. At the same time, a high-level agent can be set to summarize the results of the discussion.

[0125] Based on the shortcomings of the current large language model in terms of hallucinations and information not being real-time, the securities and futures industry has high requirements for stability, security, and information real-time. At the same time, a single agent system is relatively isolated and cannot communicate, exchange, and learn from other agents to improve performance. Based on the above shortcomings, a set of securities and futures industry large language model multi-agent collaborative control algorithms and systems are proposed. Give the large language model different personalities to play different roles, or use different large language models to discuss or game the same task. Finally, a more advanced agent summarizes the content of the discussion or game. At the same time, the system is given a memory module to store some information related to specific tasks, such as obtaining current mainstream industry data, company reports, news, etc. in the memory module. Multiple agents can obtain the information they need in the memory module to support their own views during the discussion and debate. The execution module mainly makes corresponding decisions based on the results of the discussion or debate of multiple agents.

[0126] The present application also provides a computer-readable storage medium having a computer program or instruction stored thereon, characterized in that when the computer program or instruction is executed by a processor, the steps of the system provided in any one of the above embodiments are implemented, and the system includes: a first agent module analyzes a current task based on the acquired first information to obtain a first analysis result; a second agent module analyzes the current task based on the acquired second information to obtain a second analysis result; the first agent module and the second agent module, based on the memory data acquired from the memory module, play a game on the first analysis result and the second analysis result obtained after analyzing the same task to obtain a game result; the third agent module adjusts and analyzes the game result to obtain a decision result; wherein the memory module includes short-term memory data for storing real-time data, and long-term memory data for storing historical data; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information.

[0127] In one embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the system provided in any one of the above embodiments are implemented, wherein the system includes: a first agent module analyzes a current task based on the acquired first information to obtain a first analysis result; a second agent module analyzes the current task based on the acquired second information to obtain a second analysis result; the first agent module and the second agent module play a game on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data acquired from the memory module to obtain a game result; the third agent module adjusts and analyzes the game result to obtain a decision result; wherein the memory module includes short-term memory data for storing real-time data, and long-term memory data for storing historical data; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information.

[0128] In one embodiment, a computer program product is also provided, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the steps of the system provided in any one of the above embodiments are implemented, and the system includes: a first agent module analyzes the current task based on the acquired first information to obtain a first analysis result; a second agent module analyzes the current task based on the acquired second information to obtain a second analysis result; the first agent module and the second agent module play a game on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data obtained from the memory module to obtain a game result; the third agent module adjusts and analyzes based on the game result to obtain a decision result; wherein the memory module includes short-term memory data for storing real-time data, and long-term memory data for storing historical data; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information.

[0129] It should be understood that although Figure 1-5 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1-5 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0130] It should also be understood that the first, second, third, fourth and various numerical numbers involved in this document are only distinctions made for the convenience of description and are not intended to limit the scope of the present application.

[0131] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0132] Unless otherwise defined, all technical and scientific terms used in the embodiments of the present application have the same meanings as those commonly understood by those skilled in the art to which the present application belongs. In the event of any inconsistency, the meanings described in this specification or derived from the contents recorded in this specification shall prevail.

[0133] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment system can be implemented by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned systems.

[0134] It should be understood that the processor mentioned in the embodiments of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0135] It should also be understood that the memory mentioned in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus random access memory (Direct Rambus RAM, DR RAM).

[0136] It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, the memory (storage module) is integrated in the processor.

[0137] It should be noted that the memory described herein is intended to include, but is not limited to, these and any other suitable types of memory. The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0138] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A multi-agent system of a large language model in the securities industry with a trustworthy executable environment, characterized in that: The system comprises: The first agent module analyzes the current task based on the first information obtained to obtain a first analysis result; the second agent module analyzes the current task based on the second information obtained to obtain a second analysis result; the first agent module, the second agent module and the third agent module respectively include large models trained based on the first information, the second information and the third information; wherein the first information, the second information and the third information are different types of information; The first agent module and the second agent module play games on the first analysis result and the second analysis result obtained after analyzing the same task based on the memory data obtained from the memory module to obtain a game result, wherein at least one agent module plays games on events corresponding to different times from the memory module based on the index time and the event occurrence time corresponding to the data stored in the memory module, wherein based on the formula Get the time events corresponding to different times, δ E Represents the time t at which the event occurred in the memory unit P With the inquiry time t E The difference, Q l Indicates the number of days, Used to obtain time events corresponding to different required times; or, at least one of the agent modules obtains memory data whose importance ranking reaches a preset threshold from the memory module, and conducts a game based on the obtained memory data; The third agent module performs adjustment analysis based on the game result to obtain a decision result, wherein the third agent module is a high-level agent module with a higher level than the first agent module and the second agent module.

2. The system according to claim 1, characterized in that The at least one agent module obtains memory data whose importance ranking reaches a preset threshold from the memory module, and performs a game based on the obtained memory data, including: The importance of the data stored in the memory module Determined based on a piecewise scoring function based on the weights of different events And the degradation rate α l and the time difference δ E Determine that the longer the event in the database occurs, the greater the corresponding weight, and the degradation rate α l It is used to measure the importance difference between different events. E It is the time interval between the event occurrence time and the knowledge base search query time.

3. The system according to claim 2, characterized in that The importance of the data stored in the memory module in in, It is a unified piecewise scoring function that gives different events in the database a weight. The larger the value, the longer the event occurred in the database, and the smaller the value, the shorter the event occurred. l is the degradation rate, δ E The time t at which the event occurs P With the inquiry time t E time interval; Indicates the decreasing importance of an event over time.

4. The system according to claim 1, characterized in that The system further comprises: One or more of the first proxy module, the second proxy module and the third proxy module index data from the memory module based on a vector index; One or more of the first agent module, the second agent module and the third agent module play a game based on the data indexed from the memory module.

5. The system according to claim 4, characterized in that The method of indexing data from the memory module based on vector indexing includes: The proxy module determines the similarity between the data to be indexed and the data stored in the memory module, and indexes the data from the memory module based on the similarity; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module; and / or, The method of indexing data from the memory module based on vector indexing includes: The proxy module determines the index time corresponding to the index data and the event occurrence time corresponding to the data stored in the memory module, and obtains events corresponding to different times from the memory module based on the index time and the event occurrence event; wherein the proxy module includes one or more of the first proxy module, the second proxy module and the third proxy module.

6. The system according to claim 1, characterized in that The system further comprises: sorting the memory data based on the importance of the memory data stored in the memory module; The agent module obtains memory data whose importance ranking reaches a preset threshold from the memory module, and performs game based on the obtained memory data. The agent module includes one or more of the first agent module, the second agent module and the third agent module.

7. The system according to claim 1, characterized in that The first agent module and the second agent module play a game based on the first analysis result and the second analysis result obtained after analyzing the same task, and obtain a game result, including: The first agent module and the second agent module analyze the same task, obtain the first analysis result and the second analysis result, and reach an agreement after negotiation to obtain a game result, or, The third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on a preset weight, and obtains a game result when the game between the first agent module and the second agent module meets a preset game condition.

8. The system according to claim 7, characterized in that The first agent module and the second agent module analyze the same task, obtain the first analysis result and the second analysis result, and reach an agreement after negotiation to obtain a game result, including: The first agent module and the second agent module bargain on the first analysis result and the second analysis result based on each other's views and the knowledge they have acquired; in response to the third agent module judging that the first agent module and the second agent module have reached an agreement, it is determined that the first agent module and the second agent module have reached an agreement through negotiation, and a bargaining result is obtained; The third agent module weights the opinions corresponding to the first analysis result and the second analysis result based on a preset weight, and when the game between the first agent module and the second agent module meets the preset game condition, obtains a game result, including: Based on the importance of different opinions and factors generated by the first agent module and the second agent module during the game process, weights corresponding to the different opinions and factors are determined; the third agent module performs weighted processing on the opinions generated by the first agent module and the second agent module during the game process based on the weights, and obtains a game result when the game between the first agent module and the second agent module meets a preset number of game rounds or a preset game time.

9. The system according to claim 1, characterized in that The third agent module performs adjustment analysis based on the game result to obtain a decision result, including: The third agent module generates an initial solution based on the same task, and consults and gives feedback on suggestions for the initial solution to the first agent module and the second agent module; The third agent module makes a decision to obtain a decision result based on the suggestions for the initial solution fed back by the first agent module and the second agent module.

10. The system according to any one of claims 1 to 9, characterized in that The agent module includes a brain module, a perception module and an action module, and the agent module includes one or more of a first agent module, a second agent module or a third agent module; The brain module includes giving different roles to the large language model based on Prompt engineering technology, guiding the large language model to adopt procedures to solve problems; The perception module is used to receive and understand information from the external environment or other agent modules, and process and interpret the input data so that the agent module can understand and respond accordingly, wherein the input data includes text, images, sounds or other forms of information; The action module is used to perform corresponding actions according to the results obtained after reasoning and decision-making based on the information received by the perception module, and the actions include one or more of controlling the movement of the robot, generating text replies, and performing specific tasks; and / or, The first agent module, the second agent module and the third agent module are different agent modules, and the third agent module is a high-level agent module with a higher level than the first agent module and the second agent module; The large models in the first agent module, the second agent module and the third agent module are different functional roles played by the same large model or the same large model with different functional roles; The first information, the second information and the third information are different types of information, and the large models in the first agent module, the second agent module and the third agent module are trained based on different types of information. When different large models make decisions on the same task, they make decisions based on the different types of information they have obtained.

Images