Big data and agent driven data analysis pushing method and system and storage medium

Abstract

The invention provides a big data and agent driven data analysis pushing method and system and a storage medium, and belongs to the technical field of big data processing, and the method comprises the steps: collecting and preprocessing multi-source heterogeneous data; data perception, deep analysis, user portrait modeling and push decision making are executed based on the multiple agents; a plurality of agents complete data modeling and strategy generation through a collaboration mechanism; personalized data pushing is carried out according to the user portrait and the prediction result; and the intelligent agent behavior model is optimized according to user feedback, so that dynamic self-learning and closed-loop adjustment are realized. The device and the system comprise a data acquisition module, an agent control module, an analysis decision module, a pushing module and a feedback learning module. The method and the system have high intelligence, high real-time performance and good expandability, and are suitable for data service scenes in multiple fields of finance, logistics, e-commerce and the like.

Description

Technical Field

[0001] This invention belongs to the field of big data processing technology, and more specifically, it relates to a data analysis and push method system and storage medium driven by big data and intelligent agents. Background Technology

[0002] With the rapid development of the Internet, the Internet of Things, and mobile devices, the amount of data generated globally is growing exponentially, and the types of data are becoming increasingly complex, including structured data (such as relational database data), semi-structured data (such as JSON and XML format data), and unstructured data (such as text, images, audio, and video). Against this backdrop, enterprises and institutions face the significant challenge of effectively integrating, processing, and utilizing this massive and diverse data to support decision-making, optimize services, and enhance user experience.

[0003] Existing big data processing technologies primarily rely on static data warehouse systems or batch-processing-based analytical frameworks such as Hadoop and Spark. While these technologies offer certain advantages in data processing capabilities, they suffer from significant shortcomings in real-time response, user behavior understanding, and personalized recommendations. Especially in application environments with diverse user needs and rapidly changing scenarios, traditional methods often lack dynamic adaptability, failing to accurately perceive user status and adjust recommendation strategies, leading to information redundancy or recommendation delays, thus impacting user experience and business efficiency.

[0004] On the other hand, with the development of artificial intelligence in recent years, agent technology has been widely applied to complex behavioral scenarios such as perceiving the environment, performing tasks, and making decisions. Multi-agent systems (MAS) possess autonomy, coordination, and learning capabilities, enabling them to operate independently and collaborate to complete tasks in distributed systems. However, the deep integration of agent technology with big data platforms for collaborative work in areas such as large-scale data processing, user modeling, and intelligent recommendation is still in the exploratory stage.

[0005] Currently, there is a lack of a universal framework that can efficiently integrate big data analytics capabilities with intelligent agent decision-making mechanisms, and simultaneously achieve: efficient collection and processing of multi-source data, dynamic modeling of user behavior, intelligent analysis and precise push based on multi-agent collaboration, and the ability to continuously optimize and learn.

[0006] Therefore, there is an urgent need for an innovative analysis and push mechanism that integrates big data platforms and intelligent agent systems to achieve more intelligent, real-time, and personalized data service capabilities. This invention is proposed based on this technical pain point and has significant technological advancement and industrial application value. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a data analysis and push method system and storage medium driven by big data and intelligent agents, thereby solving the technical problems of low data analysis efficiency, inaccurate push, untimely response, and lack of intelligent dynamic adjustment in existing traditional analysis and push methods.

[0008] The purpose and effectiveness of the big data and intelligent agent-driven data analysis and push method system and storage medium of the present invention are achieved by the following specific technical means: The data analysis and push method driven by big data and intelligent agents includes the following steps: S1: Collect heterogeneous data from multiple sources and perform preprocessing in a unified format to obtain preprocessed heterogeneous data; S2: Save the preprocessed heterogeneous data to the HDFS file system; S3: Based on the big data processing engine, preprocessed heterogeneous data is processed and analyzed, and the analysis results are saved to the file system; S4: Multiple modular intelligent agents are launched based on the analysis results and event triggering. S5: Multiple intelligent agents collaborate through predefined communication protocols to perform data modeling, trend prediction, and strategy selection to obtain user profiles and behavioral models; S6: Based on user profiles and behavior models, intelligently push the most relevant data content; S7: Based on the user's feedback on the pushed content, train and optimize the agent's behavior strategy, and repeat steps S5 to S7.

[0009] According to a preferred embodiment, the multi-source heterogeneous data includes, but is not limited to, structured data, semi-structured data, and unstructured data.

[0010] According to a preferred embodiment, the activation of multiple functional modular intelligent agents includes, but is not limited to: The data-sensing intelligent agent is responsible for data monitoring and preliminary classification. Analyze intelligent agents and use machine learning models to perform in-depth analysis of data; User-modeling intelligent agents construct user profiles and track their behavioral dynamics.

[0011] According to a preferred embodiment, the agent employs reinforcement learning or federated learning mechanisms to adapt to dynamic environmental changes.

[0012] According to a preferred embodiment, the push method includes APP notification, web page pop-up, email, or automatic API access to a third-party platform.

[0013] A big data and intelligent agent-driven data analysis and push system, used to implement the aforementioned big data and intelligent agent-driven data analysis and push methods, includes: The data acquisition module is used to collect and preprocess multi-source data; The agent management module is used to schedule multiple functional agents; The analysis and decision-making module is used to collaboratively generate user profiles, analysis results, and push strategies. The push module is used to deliver personalized data content to target users; The feedback learning module is used to receive user feedback and adjust model parameters and strategies.

[0014] According to a preferred embodiment, the agent management module supports multi-threaded concurrent execution and resource scheduling mechanisms.

[0015] According to a preferred embodiment, the system further includes: The server-side system deploys a data analysis engine, an intelligent agent engine, a task scheduling service, and a database. The client system is used to receive personalized push content and provide behavioral feedback; The communication interface is used for data interaction and security authentication between the server and the client.

[0016] A computer-readable storage medium having instructions stored thereon, which, when executed by a computing device, perform all the steps of the above-described big data and intelligent agent-driven data analysis and push method.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. By activating modular intelligent agents such as data-sensing agents, analysis agents, and user-modeling agents, and coordinating with the agent management module in the system, multi-agent scheduling is achieved. Each agent collaborates according to a predefined communication protocol. The data-sensing agent autonomously completes data monitoring and preliminary classification without manual data source screening; the analysis agent uses machine learning models to perform deep analysis on preprocessed heterogeneous data, automatically uncovering correlations and potential value; and the user-modeling agent continuously builds user profiles and tracks behavioral dynamics, autonomously updating user feature tags. Throughout this process, multiple agents can collaboratively complete data value judgment and preliminary push strategy formulation without manual commands, improving the system's automation capabilities, and is particularly suitable for efficient data processing and decision support in massive data scenarios.

[0018] 2. The feedback learning module receives user behavior feedback from the client system, adjusts the agent's behavior strategy and model parameters, and applies the optimized strategy iteratively to steps S5 to S7. Simultaneously, the agent employs reinforcement learning or federated learning mechanisms to adapt to dynamically changing user needs and the environment in real time. User feedback on pushed content triggers the feedback learning module to optimize the user behavior model. The user modeling agent adjusts the profile tags based on the updated model, and the analysis agent re-formulates the push strategy accordingly, ensuring that subsequent push content promptly matches changes in user needs and avoiding push disconnection issues caused by rigid strategies.

[0019] 3. By collaboratively constructing multi-dimensional user profiles and behavioral models through multi-agent collaboration, the system's analysis and decision-making modules combine pre-processed multi-source heterogeneous data to refine user characteristics from multiple dimensions such as user interests, behavioral habits, and demand scenarios, forming a personalized user tagging system. Based on this, the push module selects push methods adapted to user habits according to the user profiles and behavioral models, delivering the most relevant data content in a targeted manner; for users who prefer fragmented learning, short knowledge videos are pushed; for users who prefer in-depth reading, articles combining text and images are pushed, ensuring that the pushed content accurately matches the personalized needs of different users, reducing information redundancy, and improving user acceptance and satisfaction with the pushed content.

[0020] 4. The system's server-side deployment includes a data analysis engine, an intelligent agent engine, a task scheduling service, and a database. Each core module adopts a modular design, allowing for flexible functional adjustments based on the business needs of different industries. In the financial industry, the data acquisition module's data source interface can be expanded to access industry-specific data such as transaction data and risk control data. The intelligent agent management module schedules newly added risk analysis intelligent agents to work collaboratively. In the education industry, the analysis and decision-making module's model library can be expanded to include learning progress prediction models, adapting to the personalized learning content delivery needs of educational scenarios. Simultaneously, the system achieves data interaction and secure authentication between the server and client through communication interfaces, enabling integration without large-scale modifications to existing industry information systems. It can be widely applied in multiple fields such as finance, education, e-commerce, and healthcare, possessing commercial application value and scenario adaptability.

[0021] 5. This method covers the entire process from data acquisition and processing to feedback optimization, with each module of the system closely integrated to form a complete business closed loop. This closed-loop design not only ensures that the output of each link can support the next link, but also enables the timely detection of potential system problems through feedback and correction through strategy optimization. Furthermore, as the number of closed-loop cycles increases, the agent's strategy and model will be continuously optimized, giving the system continuous evolution capabilities and allowing it to adapt to the long-term development and changes in business and user needs. Attached Figure Description

[0022] Figure 1 This is a flowchart of the steps in the big data and intelligent agent-driven data analysis and push method of this invention; Figure 2 This is a structural diagram of the multi-agent framework in this invention; Figure 3 This is a diagram illustrating the working principle of the intelligent agent in this invention. Detailed Implementation

[0023] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the technical solutions of the present invention, but should not be used to limit the scope of protection of the present invention.

[0024] Example:

[0025] As attached Figures 1 to 3 As shown: This invention provides a data analysis and push method driven by big data and intelligent agents, particularly a data processing and push system that integrates a multi-agent collaborative mechanism, a memory system, a planning system, and a behavioral feedback learning mechanism. Referring to the system architecture shown in the accompanying drawings, in this embodiment, this invention achieves automated processing of multi-source data and personalized service push through modular design of the intelligent agent system, comprehensively improving data utilization efficiency and service intelligence level. Specifically, it includes the following steps: S1: Collect heterogeneous data from multiple sources and perform preprocessing in a unified format to obtain preprocessed heterogeneous data.

[0026] Specifically, multi-source heterogeneous data includes, but is not limited to, structured data, semi-structured data, and unstructured data. The system as a whole operates with an intelligent agent as the core unit, and all processing activities revolve around the intelligent agent, as shown in the example... Figure 2As shown. First, the system accesses multi-source heterogeneous data through a data acquisition module. This module supports multiple data access interfaces—connecting to relational databases to obtain structured basic user information, capturing text and image data from social media platforms via API interfaces, and receiving real-time sensor data from industrial equipment or smart terminals through a sensor gateway. Sources include user behavior logs, transaction records, social media text, sensor data, and search queries. This data undergoes preprocessing to clean it, removing invalid click records from user behavior logs, deleting duplicate entries from transaction data, unifying the format by converting timestamps of different formats to UTC standard time, converting unstructured text to JSON format word segmentation data, filling missing values, and extracting sentiment features from text comments and object contour features from image data. Then, it enters the intelligent agent processing flow. Each intelligent agent is a functional unit, with responsibilities based on the system's division of labor, encompassing perception, reasoning, decision-making, and behavior control capabilities. The data perception agent is responsible for perceiving the type and quality of the accessed data; the reasoning agent is responsible for analyzing the relationships between data; the decision-making agent is responsible for formulating data push strategies; and the behavior control agent is responsible for executing the push actions. The intelligent agent has independent data receiving, processing and feedback capabilities, and supports collaborative interaction with other intelligent agents to build complex multi-agent data analysis architectures. For example, when the data-aware intelligent agent detects abnormal data quality, it can directly trigger the analysis intelligent agent to investigate the cause of the abnormality.

[0027] Within the intelligent agent, various auxiliary analysis components, such as search engines, code parsers, and natural language processing engines, are first integrated through a tool submodule. These tools provide the agent with data understanding and context recognition capabilities, enabling it to accurately perceive input information and complete semantic modeling. The tool module connects with external search engines and code interpreters, allowing the agent to initiate retrieval tasks in knowledge blind spots, thus improving the system's knowledge coverage.

[0028] S2: Save the preprocessed heterogeneous data to the HDFS file system.

[0029] Specifically, after initial processing, the agent stores the received input information in the memory module. This module consists of two parts: short-term memory and long-term memory. Short-term memory temporarily stores current context information, task status, and interaction records, with a storage period typically ranging from a few hours to a few days, facilitating rapid retrieval. Long-term memory stores user profiles, including age, gender, interests, consumption habits, historical preferences, and domain knowledge graph information, providing support for long-term data analysis and personalized services. The memory module not only provides data support for the current task but also accumulates knowledge for subsequent tasks, exhibiting significant learning and adaptive capabilities. The agent can retrieve past experiences or update existing knowledge from the memory. For example, when formulating new promotional push plans, the agent can retrieve users' recent browsing history from short-term memory to adjust the priority of pushed products, thereby achieving prediction and personalized inference based on historical data.

[0030] In its implementation, the system is first deployed in an architecture with distributed storage and computing capabilities, such as using HDFS from the Hadoop ecosystem as the underlying data storage platform, combined with Spark and Flink big data processing engines, to provide access to and real-time processing capabilities for large-scale heterogeneous data. Data sources include, but are not limited to, user behavior logs, social media text, sensor-collected information, transaction data, and user profile tags. The system performs unified preprocessing operations through the access layer, including timestamp synchronization, format standardization, missing value imputation, and noise filtering, to ensure data quality and consistency for subsequent analysis modules.

[0031] Next, the processed data is stored in distributed file systems such as HDFS, with storage directories divided according to data type and usage frequency. Sensor data with high real-time requirements is stored in the real-time data directory of HDFS, while historical user behavior data is stored in the offline data directory, serving as the foundational information source for agent perception and analysis in structured or semi-structured form. Internally, the agent control module activates multiple agents with different functional characteristics. Each agent can be considered an autonomous computing node with independent perception, reasoning, and action capabilities. Agents are categorized into several types, including data perception agents, analysis and modeling agents, user profiling agents, push strategy agents, and feedback learning agents. Data perception agents monitor data updates in HDFS in real time; analysis and modeling agents use data from HDFS for model training; user profiling agents build dynamic profiles based on user data in HDFS; push strategy agents formulate plans by referring to historical push records in HDFS; and feedback learning agents store user feedback data in HDFS. Each agent can independently complete its specific tasks and also collaborate with other agents through well-defined communication protocols to complete complex data analysis and strategy formulation processes.

[0032] For example, the data-aware intelligent agent is responsible for real-time monitoring of the incoming data stream, identifying and marking abnormal events, changes in user behavior, or environmental changes, serving as the basis for the event triggering mechanism. When a specific event occurs, the system automatically launches the analysis and modeling intelligent agent. This agent calls upon historical air conditioner sales data and real-time search data stored in HDFS for the region, and uses machine learning algorithms to perform feature extraction, cluster analysis, and behavior prediction operations, thereby identifying the current user group's preferences and trend evolution. The user profiling intelligent agent combines user historical behavior, interest tags, geographical location, and device type information stored in HDFS to build a dynamic profile model. The model has a time-series update mechanism, which can continuously evolve based on the latest behavior, ensuring that the profile remains consistent with the user's current needs.

[0033] S3: Based on a big data processing engine, preprocessed heterogeneous data is processed and analyzed, and the analysis results are saved to the file system.

[0034] Specifically, the big data processing engine selects the appropriate processing method based on data type and analysis requirements. For historical data after batch preprocessing, offline analysis is performed using the Spark engine, structured data is queried and statistically analyzed using Spark SQL, and machine learning models are built using Spark MLlib for in-depth data mining. For preprocessed data generated in real time, stream processing analysis is performed using the Flink engine, data is aggregated in real time using Flink's window functions, and real-time user behavior is tracked through state management functions to ensure the timeliness of analysis results. During the analysis process, the big data processing engine calls domain rules from the system knowledge base to constrain and optimize the analysis process. After the analysis is completed, the generated analysis results, user group classification results, product association rule tables, and behavior prediction reports are saved to the analysis results directory of the HDFS file system according to preset formats and indexed for easy querying and retrieval by subsequent intelligent agents.

[0035] S4: Launch multiple modular intelligent agents based on the analysis results event trigger.

[0036] Specifically, multiple modular intelligent agents are activated, including but not limited to: a data-aware intelligent agent responsible for data monitoring and initial classification; an analysis intelligent agent that uses machine learning models to perform in-depth data analysis; and a user-modeling intelligent agent that builds user profiles and tracks their behavioral dynamics. Event triggering conditions are preset by the system, which automatically triggers the corresponding intelligent agent to activate, and then triggers the user-modeling intelligent agent to update the user profile.

[0037] Furthermore, the intelligent agent formulates operational strategies through a planning module. This module is responsible for breaking down the current task requirements into multiple executable sub-goals and generating action plans based on the system's knowledge base, memory data, and user profiles. Within the planning module, four sub-modules—reflection, self-reflection, thought chain, and sub-goal decomposition—work collaboratively: the reflection module quickly determines if similar historical tasks exist and reuses existing strategies; the self-reflection module evaluates the effectiveness of previous executions and adjusts strategies; the thought chain module forms a long-chain decision-making process through logical reasoning; and the sub-goal decomposition module is responsible for breaking down complex goals into multiple parallel or sequential sub-tasks. This mechanism ensures the system has excellent coping and adaptability when facing complex tasks.

[0038] The intelligent agent transmits the generated operation instructions to the behavior module, which is responsible for executing the decision-making actions. The behavior module is the execution end of the intelligent agent; after receiving the planning results, it converts the instructions and works collaboratively with other intelligent agents. The behavior module interacts directly with the external world, such as displaying recommended products through the front-end interface of an app or sending analysis reports via email, thereby closing the loop of system functions.

[0039] S5: Multiple intelligent agents collaborate through predefined communication protocols to perform data modeling, trend prediction, and strategy selection to obtain user profiles and behavioral models; Specifically, multiple intelligent agents collaborate through the interaction of behavioral modules to jointly complete cross-dimensional, cross-source, and cross-scenario data modeling and strategy generation tasks. The communication protocol adopts a standardized design, supporting unified data formats and standardized command interaction. User profile data is transmitted based on JSON format, and collaborative commands are sent based on a message structure with fixed fields, ensuring that different intelligent agents can accurately understand each other's needs and outputs. For example, in the financial field, one intelligent agent can focus on market trend analysis, while another handles user asset structure modeling. The two agents merge their results through a collaborative mechanism to provide users with accurate investment advice. In the e-commerce field, one intelligent agent focuses on user click behavior analysis; another agent is responsible for product feature analysis; and then a personalized recommendation list is generated through a collaborative mechanism. Through this collaborative mechanism, the system can achieve more efficient and accurate decision support.

[0040] Furthermore, the system incorporates a feedback learning mechanism. All push notifications accept user feedback, such as click-through rates, dwell time, likes, comments, and purchases. This feedback is collected by the feedback learning module, which obtains user operation data through behavioral tracking points in the client system, receives feedback data from third-party platforms via API interfaces, and uses this data as retraining data to continuously optimize the behavioral model. The feedback mechanism is applied to two key areas: the memory module and the planning module. In the memory module, feedback updates the short-term memory state, builds the latest behavior log, and updates the user profile in long-term memory. In the planning module, feedback is used to adjust strategy weights, optimize sub-goal priorities, or reconstruct thought processes. Through this continuous cycle, the system acquires self-learning, self-adaptive, and evolutionary capabilities, enabling continuous improvement in service quality and user satisfaction.

[0041] S6: Based on user profiles and behavior models, intelligently push the most relevant data content; Specifically, the push module first selects an appropriate push method based on the user's device usage habits and preferred push method tags in the user profile. Next, the push module combines user activity periods and interest decay cycle characteristics from the behavioral model to determine the push timing. Finally, the push module retrieves data content matching the user profile and behavioral model from the HDFS analysis results directory. For example, if the user profile is a young mother interested in baby products, and the behavioral model shows frequent recent browsing of diapers, then the module retrieves highly rated diaper recommendations and baby care knowledge content for intelligent push notifications.

[0042] S7: Based on the user's feedback on the pushed content, train and optimize the agent's behavior strategy, and repeat steps S5 to S7.

[0043] Specifically, the intelligent agent employs reinforcement learning or federated learning mechanisms to adapt to dynamic environmental changes. Reinforcement learning optimizes strategies through a reward-punishment mechanism: positive user feedback on pushed content rewards the corresponding agent, reinforcing the policy's execution tendency; negative feedback results in punishment, reducing the policy's execution probability. Policy parameters are adjusted through multiple iterations. Federated learning, on the other hand, achieves multi-system collaborative optimization while protecting data privacy. Agents from multiple e-commerce platforms, without sharing raw user data, transmit encrypted parameters to jointly train a better recommendation model, adapting to changes in user needs across platforms. After a push notification is completed, the system continues to collect subsequent user feedback behaviors, including click-through rates, browsing time, conversion behavior, and active ratings. This feedback-learning agent models and analyzes the data, converting the feedback data into model-recognizable feature vectors, which are then input into the agent's behavioral model, where gradient descent algorithms are used to adjust model parameters. This process forms a complete data-driven closed loop: from agent collaborative modeling in S5, to content push in S6, and then to feedback optimization in S7. The optimized strategy is then reapplied to the modeling process in S5, realizing the iterative cycle from step S5 to S7.

[0044] Furthermore, the system supports modular expansion, facilitating adaptation to different business scenarios: in the financial sector, it can be used for user credit risk assessment and personalized financial product recommendations; in the logistics industry, it can analyze transportation data and user needs in real time, optimize delivery routes, and push delivery notifications; on e-commerce platforms, it can identify product popularity and user preferences, accurately pushing new product, activity, and promotional information. The system is based on containerized deployment, supports distributed collaboration and elastic scaling of multiple agents, and can flexibly adapt to large-scale concurrent user requests and high-frequency data interaction needs, exhibiting excellent engineering feasibility and maintainability.

[0045] This embodiment also emphasizes the agent's sustainable learning and policy self-optimization capabilities in its behavioral model. During the feedback learning process, the system doesn't simply adjust recommendation rules; instead, it uses structured behavioral modeling and knowledge transfer techniques to allow the agent to share experiences across different user groups. For example, if a certain strategy performs well in a specific user group, the system can transfer the core parameters of that strategy to user groups with similar interests, only fine-tuning the details to adapt to that group, thus enabling rapid strategy propagation and optimization. Furthermore, the system integrates an anomaly detection mechanism. Through feedback learning, the system monitors the policy execution performance of each agent in real time. If an agent consistently generates negative feedback, the system automatically freezes its policy model and prompts developers via the client or management backend for manual intervention or retraining, ensuring the overall system's push quality and user experience.

[0046] This invention also provides a data analysis and push system driven by big data and intelligent agents, for implementing the above-mentioned data analysis and push method driven by big data and intelligent agents, including: The data acquisition module is used to collect and preprocess multi-source data; The agent management module is used to schedule multiple functional agents; The analysis and decision-making module is used to collaboratively generate user profiles, analysis results, and push strategies. The push module is used to deliver personalized data content to target users; The feedback learning module is used to receive user feedback and adjust model parameters and strategies.

[0047] The agent management module supports multi-threaded concurrent execution and resource scheduling mechanisms.

[0048] The system also includes: The server-side system deploys a data analysis engine, an intelligent agent engine, a task scheduling service, and a database. The client system is used to receive personalized push content and provide behavioral feedback; The communication interface is used for data interaction and security authentication between the server and the client.

[0049] The present invention also provides a computer-readable storage medium storing instructions thereon, which, when executed by a computing device, perform all the steps of the above-described big data and intelligent agent-driven data analysis and push method.

[0050] Specifically, the memory stores computer programs that are run by the processor. When the processor runs these programs, it executes big data technology and agent-driven data analysis and push systems.

[0051] This invention combines big data technology with a multi-agent architecture to construct a data service system with strong adaptability, self-learning capabilities, and high push accuracy. This system not only achieves a high degree of automation in data processing and analysis but also enables highly personalized and dynamic adjustments in strategy formulation and content delivery.

[0052] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A data analysis and push method driven by big data and intelligent agents, characterized in that, Includes the following steps: S1: Collect heterogeneous data from multiple sources and perform preprocessing in a unified format to obtain preprocessed heterogeneous data; S2: Save the preprocessed heterogeneous data to the HDFS file system; S3: Based on the big data processing engine, preprocessed heterogeneous data is processed and analyzed, and the analysis results are saved to the file system; S4: Multiple modular intelligent agents are launched based on the analysis results and event triggering. S5: Multiple intelligent agents collaborate through predefined communication protocols to perform data modeling, trend prediction, and strategy selection to obtain user profiles and behavioral models; S6: Based on user profiles and behavior models, intelligently push the most relevant data content; S7: Based on the user's feedback on the pushed content, train and optimize the agent's behavior strategy, and repeat steps S5 to S7.

2. The data analysis and push method driven by big data and intelligent agents according to claim 1, characterized in that, The multi-source heterogeneous data includes, but is not limited to, structured data, semi-structured data, and unstructured data.

3. The data analysis and push method driven by big data and intelligent agents according to claim 1, characterized in that, The activation of multiple functional modular intelligent agents includes, but is not limited to: The data-sensing intelligent agent is responsible for data monitoring and preliminary classification. Analyze intelligent agents and use machine learning models to perform in-depth analysis of data; User-modeling intelligent agents construct user profiles and track their behavioral dynamics.

4. The data analysis and push method driven by big data and intelligent agents according to claim 1, characterized in that, The agent employs reinforcement learning or federated learning mechanisms to adapt to dynamic environmental changes.

5. The data analysis and push method driven by big data and intelligent agents according to claim 1, characterized in that, The push methods include APP notifications, web page pop-ups, emails, or automatic API access to third-party platforms.

6. A data analysis and push system driven by big data and intelligent agents, used to implement the data analysis and push method driven by big data and intelligent agents as described in any one of claims 1 to 5, characterized in that, include: The data acquisition module is used to collect and preprocess multi-source data; The agent management module is used to schedule multiple functional agents; The analysis and decision-making module is used to collaboratively generate user profiles, analysis results, and push strategies. The push module is used to deliver personalized data content to target users; The feedback learning module is used to receive user feedback and adjust model parameters and strategies.

7. The big data and intelligent agent-driven data analysis and push system according to claim 6, characterized in that, The intelligent agent management module supports multi-threaded concurrent execution and resource scheduling mechanisms.

8. The big data and intelligent agent-driven data analysis and push system according to claim 6, characterized in that, The system also includes: The server-side system deploys a data analysis engine, an intelligent agent engine, a task scheduling service, and a database. The client system is used to receive personalized push content and provide behavioral feedback; The communication interface is used for data interaction and security authentication between the server and the client.

9. A computer-readable storage medium having instructions stored thereon, which, when executed by a computing device, perform all the steps of the big data and agent-driven data analysis and push method according to any one of claims 1 to 5.

Images