Methods and apparatus for generating decisions, electronic equipment
By acquiring multi-source heterogeneous data and utilizing a data computing engine and multi-agent team collaboration for analysis, the problem of slow response of decision-making systems in complex environments is solved, achieving rapid and efficient decision generation and optimization.
Patent Information
- Application Number
- CN202510623725.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-14
- Publication Date
- 2026-11-17
AI Technical Summary
When faced with complex and ever-changing dynamic environments, existing technologies cannot make decision-making systems respond quickly, resulting in low efficiency in decision generation.
By acquiring heterogeneous data from multiple sources, using a data computing engine to extract features, generating initial decision-making schemes, and then conducting collaborative analysis through a multi-agent team to optimize the decision-making process.
It improves the efficiency and quality of decision generation, enabling the rapid provision of feasible decision solutions in complex environments, adapting to environmental changes, and reducing data processing time.
Smart Images

Figure CN123115884A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, such as a method and apparatus for generating decisions, and electronic devices. Background Technology
[0002] With the development of big data and artificial intelligence technologies, an increasing number of decision support systems are incorporating data-driven approaches to improve the accuracy and efficiency of decision-making through real-time data analysis and machine learning algorithms. Multi-agent systems have been widely applied in complex task processing in recent years. Through the collaboration of multiple agents, task decomposition, parallel processing, and dynamic optimization can be achieved, improving the overall performance of the system.
[0003] In related technologies, task processing and decision support systems rely on fixed rule bases or historical data analysis and depend on human experience for decision-making.
[0004] In the process of implementing the embodiments of this disclosure, at least the following problems were found in the related art:
[0005] When faced with complex and ever-changing dynamic environments, the decision-making systems in related technologies cannot respond quickly, resulting in decision delays and affecting the efficiency of decision generation.
[0006] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention
[0007] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.
[0008] This disclosure provides a method, apparatus, and electronic device for generating decisions, thereby improving the efficiency of decision generation.
[0009] In some embodiments, the method for generating a decision includes: acquiring multi-source heterogeneous data; extracting features from the multi-source heterogeneous data using a data computing engine to obtain target feature data; generating an initial decision scheme based on the target feature data; and conducting collaborative analysis of the initial decision scheme using a multi-agent team to obtain a target decision scheme.
[0010] Optionally, the multi-source heterogeneous data includes user requirements, sensor data, business system logs, and external API data; acquiring multi-source heterogeneous data includes: acquiring data streams through a distributed message queue; performing standardized interfaces with external systems to acquire multi-source heterogeneous data from the data streams; and preprocessing the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data.
[0011] Optionally, feature extraction is performed on multi-source heterogeneous data based on a data computing engine, including one or more of the following: calculating the statistical features of multi-source heterogeneous data based on a streaming computing framework; aggregating and reducing the dimensionality of multi-source heterogeneous data based on edge computing nodes to obtain lightweight features; and using principal component analysis and time series segmentation to reduce the amount of redundant data in multi-source heterogeneous data and extract key features.
[0012] Optionally, generating an initial decision scheme based on target feature data includes: generating candidate actions in the current environmental state corresponding to the target feature data based on a reinforcement learning model; evaluating the long-term benefits of the candidate actions in the current environmental state; adjusting model parameters based on the evaluation results, selecting a target decision path, and generating an initial decision scheme based on the target decision path.
[0013] Optionally, the method for generating decisions may further include: updating model parameters in real time through incremental training, retaining the latest data through a sliding window to adapt to sudden data changes; and / or, triggering retraining of the reinforcement learning model based on a concept drift detection algorithm.
[0014] Optionally, a multi-agent team can be constructed as follows: Based on a preset framework, a working group is created; an organizer agent and multiple participant agents are created in the working group, and the job descriptions and system messages of each agent are set to obtain the multi-agent team; the termination conditions of the multi-agent team are defined.
[0015] Optionally, the multi-agent team collaboratively analyzes the initial decision scheme to obtain the target decision scheme in the following manner: the organizer agent is responsible for decomposing the initial decision scheme into multiple analysis tasks, and sending these tasks to the corresponding participant agents according to a preset selection strategy and logic; each participant agent generates response information based on its own knowledge and reasoning ability, according to the received analysis tasks, and broadcasts the response information to all participant agents; after receiving the response information broadcast by other participant agents, each participant agent modifies its own response information and broadcasts the modified response information to all participant agents again; upon reaching the termination condition, the organizer agent collects the response information from each participant agent and generates the target decision scheme.
[0016] In some embodiments, the apparatus for generating decisions includes: a data sensing module configured to acquire multi-source heterogeneous data; a data computing module configured to extract features from the multi-source heterogeneous data based on a data computing engine to obtain target feature data; a decision generation module configured to generate an initial decision scheme based on the target feature data; and a team collaboration module configured to perform collaborative analysis on the initial decision scheme based on a multi-agent team to obtain a target decision scheme.
[0017] In some embodiments, the means for generating a decision includes a processor and a memory storing program instructions, the processor being configured to execute the method for generating a decision as described above when the program instructions are executed.
[0018] In some embodiments, the electronic device includes: an electronic device body; and the means for generating decisions as described above, which is mounted on the electronic device body.
[0019] The method, apparatus, and electronic device for generating decisions provided in this disclosure can achieve the following technical effects:
[0020] In this embodiment, feature extraction from multi-source heterogeneous data using a data computing engine reduces data processing time and improves system response speed. Generating an initial decision scheme based on target feature data quickly provides a feasible starting point for decision-making, laying the foundation for subsequent optimization. Collaborative analysis of the initial decision scheme by a multi-agent team compresses data processing time. Through mutual discussion and analysis among multiple agents, time compression and rapid trial and error are achieved, improving the efficiency of generating the target decision scheme. Furthermore, the multi-agent team can analyze and optimize the initial decision scheme from different perspectives, achieving multi-faceted optimization and improving decision quality.
[0021] The above general description and the description below are exemplary and illustrative only and are not intended to limit this application. Attached Figure Description
[0022] One or more embodiments are illustrated by way of example with reference to the accompanying drawings. These illustrations and drawings do not constitute a limitation on the embodiments. Elements having the same reference numerals in the drawings are shown as similar elements. The drawings are not to be scaled. And wherein:
[0023] Figure 1 This is a schematic diagram of a method for generating decisions provided in an embodiment of this disclosure;
[0024] Figure 2 This is a schematic diagram of another method for generating decisions provided in an embodiment of this disclosure;
[0025] Figure 3This is a schematic diagram of another method for generating decisions provided in an embodiment of this disclosure;
[0026] Figure 4 This is a schematic diagram of a method for generating target decision schemes by a multi-agent team according to an embodiment of this disclosure;
[0027] Figure 5 This is a schematic diagram of an apparatus for generating decisions provided in an embodiment of this disclosure;
[0028] Figure 6 This is a schematic diagram of another apparatus for generating decisions provided in an embodiment of this disclosure. Detailed Implementation
[0029] To provide a more detailed understanding of the features and technical content of the embodiments of this disclosure, the implementation of the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this disclosure. In the following technical description, for ease of explanation, several details are used to provide a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be simplified in their depiction to simplify the drawings.
[0030] The terms "first," "second," etc., used in the technical solutions described in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.
[0031] Unless otherwise stated, the term "multiple" means two or more.
[0032] In this embodiment of the disclosure, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.
[0033] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.
[0034] The term "correspondence" can refer to an association or binding relationship. The correspondence between A and B means that there is an association or binding relationship between A and B.
[0035] Combination Figure 1 As shown, this disclosure provides a method for generating decisions, wherein the executing entity of the method may be a processor, and the method includes:
[0036] S101, the processor acquires heterogeneous data from multiple sources.
[0037] S102, the processor uses a data computing engine to extract features from multi-source heterogeneous data to obtain target feature data.
[0038] S103, the processor generates an initial decision scheme based on the target feature data.
[0039] S104, the processor performs collaborative analysis of the initial decision scheme based on a multi-agent team to obtain the target decision scheme.
[0040] In this embodiment, feature extraction from multi-source heterogeneous data using a data computing engine reduces data processing time and improves system response speed. Generating an initial decision scheme based on target feature data quickly provides a feasible starting point for decision-making, laying the foundation for subsequent optimization. Collaborative analysis of the initial decision scheme by a multi-agent team compresses data processing time. Through mutual discussion and analysis among multiple agents, time compression and rapid trial and error are achieved, improving the efficiency of generating the target decision scheme. Furthermore, the multi-agent team can analyze and optimize the initial decision scheme from different perspectives, achieving multi-faceted optimization and improving decision quality.
[0041] Optionally, multi-source heterogeneous data includes user requirements, sensor data, business system logs, and external API (Application Programming Interface) data.
[0042] In this embodiment, multi-source heterogeneous data refers to a collection of data from different sources with different formats and characteristics, which can provide comprehensive and real-time information support for the decision support system. By effectively integrating and processing this data, more efficient and accurate decision generation can be achieved. Sensor data refers to physical quantity or environmental parameter data collected through various sensor devices. These sensors can be deployed in IoT devices, industrial automation systems, environmental monitoring equipment, etc. Business system logs refer to log data generated by various business systems within an enterprise during operation. These logs record information such as system operations, user behavior, and transaction processing. External API data refers to data obtained through external application programming interfaces (APIs). This data can come from market data providers, weather services, news media, social media, etc.
[0043] Optionally, acquiring multi-source heterogeneous data includes: acquiring data streams through a distributed message queue; performing standardized interfaces with external systems to acquire multi-source heterogeneous data from the data streams; and preprocessing the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data.
[0044] Combination Figure 2As shown in the embodiments of this disclosure, another method for generating decisions is provided, including:
[0045] S201, the processor obtains data streams through a distributed message queue.
[0046] S202, the processor interfaces with external systems in a standardized manner to acquire multi-source heterogeneous data from the data stream.
[0047] S203, the processor preprocesses the multi-source heterogeneous data to obtain the preprocessed multi-source heterogeneous data.
[0048] S204, the processor uses a data computing engine to extract features from multi-source heterogeneous data to obtain target feature data.
[0049] S205, the processor generates an initial decision scheme based on the target feature data.
[0050] S206, the processor is based on the collaborative analysis of the initial decision scheme by a multi-agent team to obtain the target decision scheme.
[0051] In this embodiment, a distributed message queue is first used to acquire the data stream, enabling efficient processing of high-throughput data streams and ensuring real-time data access. The message queue, acting as middleware, decouples the data source from the data processing system, improving system scalability and stability. Then, through standardized interfaces, it seamlessly connects with various external systems such as sensor networks, business systems, and external data services, ensuring smooth data access. Standardized interfaces unify data of different formats into a standardized format, facilitating subsequent data processing and analysis. Finally, multi-source heterogeneous data is preprocessed to remove invalid data, ensuring the accuracy and reliability of the input data.
[0052] Optionally, Kafka can be used as a distributed message queue.
[0053] Optionally, data streams can be acquired using real-time reading or offline fetching based on a preset period.
[0054] Alternatively, standardized integration can be achieved using RESTful API (Representational State Transfer API) and MQTT (Message Queuing Telemetry Transport) protocols.
[0055] Optionally, preprocessing of multi-source heterogeneous data includes one or more of the following: parsing data types, converting unstructured data of multi-source heterogeneous data into structured data; filtering invalid data; and segmenting multi-source heterogeneous data using time windows.
[0056] In this embodiment, preprocessing heterogeneous source data significantly improves data quality and usability, providing a more reliable data foundation for subsequent analysis and decision-making. Dynamic parsing of data types via Schema-on-Read allows for automatic identification and classification of data from different sources, such as sensor data (numerical), log data (text), and API data (JSON format). Since unstructured data such as text and images are often difficult to use directly for data analysis and modeling, converting them to structured data provides a unified format and standard, allowing for more effective utilization, such as for feature extraction and training machine learning models. Rule engines like Drools are used to filter invalid data, including null values and outliers. Filtering invalid data removes noise and errors, improving overall data quality. This helps reduce errors and biases during data processing, enhancing the accuracy of subsequent analysis. Time windows ensure the continuity and consistency of data over time. Segmenting data through time windows breaks down large-scale data into smaller, manageable blocks, which is particularly important for processing time-series data (such as sensor data and business system logs), ensuring that the temporal relationship of the data is not disrupted.
[0057] Optionally, feature extraction is performed on multi-source heterogeneous data based on a data computing engine, including one or more of the following: calculating the statistical features of multi-source heterogeneous data based on a streaming computing framework; aggregating and reducing the dimensionality of multi-source heterogeneous data based on edge computing nodes to obtain lightweight features; and using principal component analysis and time series segmentation to reduce the amount of redundant data in multi-source heterogeneous data and extract key features.
[0058] In this embodiment, multi-source heterogeneous data is processed using data computing engines such as streaming computing and edge computing, enabling millisecond-level feature extraction. The Apache Flink streaming computing framework is used to perform real-time feature computation, obtaining statistical features such as mean, variance, maximum, and minimum values. The streaming computing framework can handle real-time data streams, ensuring the timeliness of feature extraction and making it suitable for application scenarios requiring rapid response. By calculating various statistical features, the characteristics of the data can be comprehensively captured, providing rich information for subsequent analysis and decision-making. Lightweight computing units such as NVIDIA Jetson are deployed at the data source to implement edge computing node settings, performing preliminary data aggregation and dimensionality reduction, and lightweight feature extraction to obtain lightweight features such as moving averages and FFT spectral analysis. Principal component analysis and time series segmentation are used to effectively reduce the dimensionality of the data, extract key features, and reduce redundant data. By extracting key features, important information in the data can be captured more accurately, improving the accuracy of subsequent analysis and decision-making.
[0059] Optionally, generating an initial decision scheme based on target feature data includes: generating candidate actions in the current environmental state corresponding to the target feature data based on a reinforcement learning model; evaluating the long-term benefits of the candidate actions in the current environmental state; adjusting model parameters based on the evaluation results, selecting a target decision path, and generating an initial decision scheme based on the target decision path.
[0060] Combination Figure 3 As shown in the embodiments of this disclosure, another method for generating decisions is provided, including:
[0061] S301, the processor acquires heterogeneous data from multiple sources.
[0062] S302, the processor uses a data computing engine to extract features from multi-source heterogeneous data to obtain target feature data.
[0063] The S303 processor generates candidate actions based on a reinforcement learning model, using the current environment state corresponding to the target feature data.
[0064] S304, the processor evaluates the long-term benefits of candidate actions in the current environment state.
[0065] S305, the processor adjusts the model parameters based on the evaluation results, selects the target decision path, and generates an initial decision scheme based on the target decision path.
[0066] The S306 processor is based on a multi-agent team that collaboratively analyzes the initial decision-making scheme to obtain the target decision-making scheme.
[0067] In this embodiment, a reinforcement learning model is used to generate candidate actions in the current environmental state, adapting to environmental changes and making flexible decisions in dynamic environments. By exploring and utilizing policies, the reinforcement learning model can find the global optimum in complex environments. Then, Monte Carlo tree search or other evaluation methods are used to assess the long-term benefits of candidate actions in the current environmental state, considering the long-term impact of decisions and generating more forward-looking and strategic decisions. The evaluation of long-term benefits can take into account the risks and uncertainties of different actions, thus selecting a more robust decision path. The evaluation of candidate actions can also be dynamically adjusted according to environmental changes, ensuring the timeliness and adaptability of decisions. Finally, the model parameters are automatically adjusted based on the evaluation results, continuously optimizing the decision-making strategy. This adaptive learning capability can continuously improve the accuracy of decisions over time. By selecting the optimal target decision path, high-quality initial decision schemes can be generated.
[0068] Optionally, the reinforcement model is constructed based on deep deterministic policy gradients.
[0069] Alternatively, the methods for generating decisions may also include: updating model parameters in real time through incremental training, and retaining the latest data through a sliding window to adapt to sudden data changes.
[0070] In this embodiment, incremental training allows the model to learn and adapt to new data in real time during operation, ensuring that the model can quickly respond to changes in the environment. By updating model parameters in real time, it can always maintain sensitivity to the latest data, ensuring the timeliness and accuracy of decisions. The sliding window mechanism ensures that the model is always trained with the latest data, avoiding the impact of outdated data on model performance. By retaining the latest data, the system can quickly respond to sudden data changes, adjust model parameters in a timely manner, and ensure the accuracy of decisions.
[0071] Optionally, the method for generating decisions also includes: triggering retraining of the reinforcement learning model based on a concept drift detection algorithm.
[0072] In this embodiment, a concept drift detection algorithm is used to monitor changes in data distribution. When a significant change is detected, model retraining is triggered. The concept drift detection algorithm can promptly identify changes in data distribution, such as shifts in market trends or changes in user behavior patterns, ensuring the model can adapt to these changes in a timely manner. By triggering retraining when concept drift is detected, the model's robustness can be maintained, avoiding performance degradation caused by changes in data distribution.
[0073] Optionally, a multi-agent team can be constructed as follows: Based on a preset framework, a working group is created; an organizer agent and multiple participant agents are created in the working group, and the job descriptions and system messages of each agent are set to obtain the multi-agent team; the termination conditions of the multi-agent team are defined.
[0074] In this embodiment, the AutoGen framework can be used as a preset framework to provide a standardized collaboration mode, ensuring a smooth collaboration process. Then, a SelectorGroupChat working group is created. Within the working group, an organizer agent and multiple participant agents are created. A job description is set for each agent, clearly defining their responsibilities and tasks, avoiding role conflicts and task overlaps during collaboration. For example, the organizer agent is responsible for task allocation and coordination, ensuring efficient and orderly collaboration across the entire team; participant agents focus on their respective tasks, leveraging their professional strengths. Through a system messaging mechanism, agents can share information in real time, ensuring each agent receives the latest data and decision status, maintaining consistency in team collaboration. Finally, termination conditions for the multi-agent team need to be defined, such as achieving predetermined decision quality, completing task objectives, or reaching the maximum number of iterations. Defining termination conditions sets clear boundaries for team collaboration, avoiding endless discussions and analysis, and ensuring tasks are completed within a reasonable timeframe.
[0075] The following is a specific example illustrating the construction of a multi-intelligent team. An organizer agent is created within the workgroup and defined as: "As the project organizer, responsible for coordinating the work of all parties. Responsibilities include: receiving user requirements and distributing them to relevant participant agents; coordinating the work progress of each participant agent; summarizing the opinions of each participant agent to form a final solution; terminating the dialogue when receiving a 'end discussion' instruction from the user. Please coordinate team work in a professional but friendly manner." Then, participant agents are created, which may include: a senior requirements evaluator, a cost control expert, and a technical engineer. The participant agents are defined as follows: "As a senior requirements evaluator, skilled in analyzing project requirements. Responsibilities include: assessing the rationality and feasibility of user requirements; identifying potential requirement risks; and proposing requirements optimization suggestions. Please express the assessment results in a professional but easily understandable way." "As a cost control expert, skilled in project budgeting and cost control. Responsibilities include: assessing project costs; proposing cost optimization solutions; and monitoring budget execution. Please support cost analysis with data and facts." "As a technical engineer, responsible for the technical implementation of the project. Responsibilities include: assessing technical feasibility; designing technical solutions; and solving technical problems. Please express technical solutions in a technical but clearly structured way." Finally, termination conditions are defined, including: "Terminate the conversation when the message contains 'End discussion'"; and / or, "Terminate the conversation when the number of message rounds exceeds 20".
[0076] Optionally, the multi-agent team collaboratively analyzes the initial decision scheme to obtain the target decision scheme in the following manner: the organizer agent is responsible for decomposing the initial decision scheme into multiple analysis tasks, and sending these tasks to the corresponding participant agents according to a preset selection strategy and logic; each participant agent generates response information based on its own knowledge and reasoning ability, according to the received analysis tasks, and broadcasts the response information to all participant agents; after receiving the response information broadcast by other participant agents, each participant agent modifies its own response information and broadcasts the modified response information to all participant agents again; upon reaching the termination condition, the organizer agent collects the response information from each participant agent and generates the target decision scheme.
[0077] Combination Figure 4 As shown in the embodiments of this disclosure, a method for a multi-agent team to generate a target decision scheme is provided, including:
[0078] S401, the organizer agent is responsible for decomposing the initial decision scheme into multiple analysis tasks, and sending the multiple analysis tasks to the corresponding participant agents according to the preset selection strategy and logic.
[0079] S402, each participant agent generates response information based on its own knowledge and reasoning ability according to the received analysis task, and broadcasts the response information to all participant agents.
[0080] S403: After receiving the response information broadcast by other participant agents, each participant agent modifies its own response information and broadcasts the modified response information to all participant agents again.
[0081] S404, the organizer agent determines whether the termination condition has been met. If yes, then execute S405; otherwise, execute S403.
[0082] S405, the organizer agent collects the response information from each participant agent and generates a target decision plan.
[0083] In this embodiment, the organizer agent decomposes the initial decision-making scheme into multiple analysis tasks and assigns these tasks to participating agents based on their complexity and capabilities. Participating agents can leverage their expertise and abilities to handle specific analysis tasks, improving the quality of the analysis. Each participant agent generates response information based on its own knowledge and reasoning abilities, and a broadcast mechanism ensures that all agents can access the analysis results of other agents in real time, promoting knowledge sharing and information flow. By sharing information, agents can learn from each other, optimize their responses, and improve the overall decision-making quality. Each agent can dynamically adjust its response based on feedback from other agents, gradually approaching the optimal solution. Through multiple iterations, the agents' responses are continuously optimized, helping to converge viewpoints among agents, progressively improving the quality of the decision-making scheme, and ultimately forming a consistent decision. The termination condition ensures that the generation of the decision-making scheme is completed within a reasonable timeframe while guaranteeing decision quality.
[0084] The following is a specific example illustrating the collaborative analysis of a multi-agent team. In a practical application scenario of stock research and analysis, the multi-agent team used includes an organizer agent and three participant agents: a financial analyst, a news analyst, and a writer. The user's requirement is to improve the depth and efficiency of stock research and predict stock market trends through the professional collaboration of each agent. In generating the initial decision-making plan, the organizer agent, as the team's central hub, is responsible for task allocation and coordination. Based on specific research needs, it systematically assigns tasks to each participant agent, ensuring seamless integration and efficient execution of the entire process. The financial analyst focuses on in-depth analysis of stock data and financial indicators, providing detailed data support and laying a solid foundation for investment decisions. The news analyst extensively collects and analyzes the latest stock-related news, extracting key information and market insights to help understand market dynamics. The writer is responsible for integrating the work of the financial analyst and the news analyst, writing a logically clear and content-rich research report to provide investors with a comprehensive perspective. After each participant agent completes its specific task, it returns control to the organizer agent. The organizer agent assesses whether further task allocation is needed or terminates the process based on the current progress. This cyclical and interactive mechanism fully utilizes the expertise of each agent, enabling efficient collaboration and information sharing in the stock research process.
[0085] Combination Figure 5 As shown in the illustration, this disclosure provides an apparatus 500 for generating decisions, including a data sensing module 501, a data computing module 502, a decision generation module 503, and a team collaboration module 504. The data sensing module 501 is configured to acquire multi-source heterogeneous data. The data computing module 502 is configured to extract features from the multi-source heterogeneous data based on a data computing engine to obtain target feature data. The decision generation module 503 is configured to generate an initial decision scheme based on the target feature data. The team collaboration module 504 is configured to perform collaborative analysis on the initial decision scheme based on a multi-agent team to obtain a target decision scheme.
[0086] The decision-making apparatus provided in this disclosure, from data perception and feature extraction to initial decision generation and team collaboration optimization, aims to improve the quality and efficiency of decision-making at every step. The final generated target decision scheme has high consistency and authority, providing users with efficient and reliable decision support, helping them quickly make optimal decisions in complex and ever-changing environments.
[0087] Combination Figure 6As shown, this disclosure provides another apparatus 600 for generating decisions, including a processor 601 and a memory 602. Optionally, the apparatus may further include a communication interface 603 and a bus 604. The processor 601, communication interface 603, and memory 602 can communicate with each other via the bus 604. The communication interface 603 can be used for information transmission. The processor 601 can invoke logical instructions in the memory 602 to execute the decision-generating method of the above embodiments.
[0088] Furthermore, the logic instructions in the aforementioned memory 602 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.
[0089] The memory 602, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this disclosure. The processor 601 executes functional applications and data processing by running the program instructions / modules stored in the memory 602, that is, it implements the method for generating decisions in the above embodiments.
[0090] The memory 602 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 602 may include high-speed random access memory and may also include non-volatile memory.
[0091] This disclosure provides an electronic device, including: an electronic device body, and the aforementioned means for generating decisions. The means for generating decisions is installed in the electronic device body. The installation relationship described herein is not limited to placement within the electronic device, but also includes installation connections with other components of the electronic device, including but not limited to physical connections, electrical connections, or signal transmission connections. Those skilled in the art will understand that the means for generating decisions can be adapted to feasible electronic device bodies to achieve other feasible embodiments.
[0092] This disclosure provides a computer-readable storage medium storing computer-executable instructions configured to perform the above-described method for generating decisions.
[0093] The technical solutions of this disclosure can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this disclosure. The aforementioned storage medium can be a non-transitory storage medium, including: a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and other media capable of storing program code.
[0094] The foregoing description and accompanying drawings fully illustrate embodiments of this disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terminology used in this application is for describing embodiments only and is not intended to limit the technical solutions described herein. As used in the technical solutions described herein, the singular forms “a,” “an,” and “the” are intended to equally include the plural forms unless the context clearly indicates otherwise. Similarly, the term “and / or” as used herein means including one or more of the associated listed items and all possible combinations thereof. Additionally, when used in this application, the term "comprise" and its variations "comprises" and / or "comprising" refer to the presence of stated features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof. Without further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes said element. In this document, each embodiment may focus on the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, the relevant parts can be referred to the description of the method section.
[0095] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this disclosure. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0096] The methods and products (including but not limited to devices and equipment) disclosed in the embodiments herein can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may be merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the shown or discussed units may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to implement this embodiment according to actual needs. Furthermore, the functional units in the embodiments of this disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0097] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
Claims
1. A method for generating decisions, characterized in that, include: Acquire multi-source heterogeneous data; Based on the data computing engine, feature extraction is performed on multi-source heterogeneous data to obtain target feature data; Generate an initial decision-making scheme based on the target feature data; The target decision solution is obtained by the collaborative analysis of the initial decision scheme by a multi-agent team.
2. The method according to claim 1, characterized in that, Multi-source heterogeneous data includes user requirements, sensor data, business system logs, and external API data; acquiring multi-source heterogeneous data includes: Data streams are obtained through a distributed message queue; Standardized interfaces with external systems are used to acquire multi-source heterogeneous data from data streams. Preprocessing is performed on multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data.
3. The method according to claim 1, characterized in that, Feature extraction based on multi-source heterogeneous data using a data computing engine, including one or more of the following: Based on the streaming computing framework, the statistical characteristics of multi-source heterogeneous data are calculated. Based on edge computing nodes, multi-source heterogeneous data is aggregated and dimensionality reduced to obtain lightweight features; Principal component analysis and time series segmentation are used to reduce redundant data from multi-source heterogeneous data and extract key features.
4. The method according to claim 1, characterized in that, An initial decision-making scheme is generated based on the target feature data, including: Based on the reinforcement learning model, candidate actions are generated in the current environment state corresponding to the target feature data; Evaluate the long-term benefits of candidate actions in the current environmental state; Adjust the model parameters based on the evaluation results, select the target decision path, and generate an initial decision scheme based on the target decision path.
5. The method according to claim 4, characterized in that, Also includes: The model parameters are updated in real time through incremental training, and the latest data is retained through a sliding window to adapt to sudden data changes; And / or, Based on the concept drift detection algorithm, the reinforcement learning model is retrained.
6. The method according to any one of claims 1 to 5, characterized in that, Construct a multi-agent team as follows: Create a workgroup based on the preset framework; Create an organizer agent and multiple participant agents in the workgroup, and set the job descriptions and system messages for each agent to obtain a multi-agent team; Define the termination conditions for a multi-agent team.
7. The method according to claim 6, characterized in that, The multi-agent team collaboratively analyzes the initial decision-making scheme to obtain the target decision-making scheme in the following manner: The organizer agent is responsible for breaking down the initial decision-making scheme into multiple analysis tasks, and sending these analysis tasks to the corresponding participant agents according to the preset selection strategy and logic. Each participating agent generates a response based on its own knowledge and reasoning ability, according to the received analysis task, and broadcasts the response to all participating agents. After receiving the response information broadcast by other participant agents, each participant agent modifies its own response information and then broadcasts the modified response information to all participant agents again. After the termination condition is met, the organizer agent collects the responses from each participant agent and generates a target decision plan.
8. An apparatus for generating decisions, characterized in that, include: The data sensing module is configured to acquire multi-source heterogeneous data; The data computing module is configured to extract features from multi-source heterogeneous data based on the data computing engine to obtain target feature data; The decision generation module is configured to generate an initial decision scheme based on the target feature data; The team collaboration module is configured to perform collaborative analysis of the initial decision-making scheme based on a multi-agent team to obtain the target decision-making scheme.
9. An apparatus for generating decisions, comprising a processor and a memory storing program instructions, characterized in that, The processor is configured to, when executing the program instructions, perform the method for generating decisions as described in any one of claims 1 to 7.
10. An electronic device, characterized in that, include: The electronic device itself; The apparatus for generating decisions as described in claim 8 or 9 is mounted on the electronic device body.
