Dynamic business decision-making system driven by data assets
By constructing a data assetization layer and a dynamic decision-making system, integrating multi-source data and performing closed-loop optimization, the problem of data silos has been solved, enabling real-time, agile, and efficient decision-making, improving the accuracy and flexibility of decision-making, and reducing maintenance costs.
Patent Information
- Application Number
- CN202511803804.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-03-03
AI Technical Summary
Existing business decision-making systems suffer from data silos, making it difficult to integrate cross-departmental and cross-business data into unified and reusable data assets. This results in a one-sided decision-making perspective, and the decision-making process is mostly static or semi-static, failing to respond agilely to real-time changes in the business environment. The system also lacks an effective closed-loop learning mechanism, causing the accuracy of decisions to decline over time, and resulting in high maintenance and update costs.
The data assetization layer integrates multi-source heterogeneous data, builds and deploys machine learning models through feature engineering and model layer, generates optimal decisions using dynamic decision engine, and performs closed-loop optimization through decision execution and feedback loop. Combined with asset management and performance monitoring center for system management and evaluation, the decision strategy is optimized using multi-armed gambling machine algorithm and A/B testing framework.
It enables unified integration and flexible management of multi-source data, improves the real-time nature and agility of decision-making, ensures the adaptability and trustworthiness of the decision-making process, significantly improves the accuracy and flexibility of decision-making, and reduces maintenance and update costs.
Smart Images
Figure CN121599774A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing and intelligent decision-making technology, specifically a dynamic business decision-making system driven by data assets. Background Technology
[0002] In today's digital economy, data has become a core strategic resource for enterprises. Enterprises accumulate massive amounts of market, operational, customer, and production data through various business systems. To extract insights from this data to support decision-making, various business intelligence systems and decision support systems have been developed. These systems typically rely on technologies such as data warehousing and online analytical processing (OLAP) to extract, clean, and integrate historical data, generating fixed reports or dashboards to provide managers with static views based on historical patterns. Furthermore, with the development of machine learning technology, some recommendation systems or risk control systems based on predictive models have also emerged, capable of providing automated decision-making suggestions in specific scenarios.
[0003] Existing business decision-making systems suffer from significant shortcomings, including severe data silos. Data from across departments and business units is difficult to integrate into unified, reusable data assets, leading to biased decision-making perspectives. Furthermore, decision-making processes are often static or semi-static, heavily reliant on historical data and pre-defined rules, failing to respond agilely to real-time changes in the business environment. Simultaneously, these systems generally lack effective closed-loop learning mechanisms; once deployed, decision models become rigid, unable to utilize feedback data after decision execution for self-optimization and adjustment. This results in declining decision accuracy over time and high maintenance and update costs. Therefore, these systems do not meet current needs. To address this, we propose a data asset-driven dynamic business decision-making system. Summary of the Invention
[0004] The purpose of this invention is to provide a data asset-driven dynamic business decision-making system to address the significant shortcomings of the business decision-making systems mentioned in the background art. These shortcomings include severe data silos, difficulty in integrating cross-departmental and cross-business data into unified and reusable data assets, leading to a one-sided decision-making perspective. Furthermore, the decision-making process is often static or semi-static, heavily reliant on historical data and preset rules, failing to respond agilely to real-time changes in the business environment. Additionally, these systems generally lack effective closed-loop learning mechanisms; once deployed, the decision-making model becomes fixed, unable to utilize feedback data after decision execution for self-optimization and adjustment, resulting in a decline in decision accuracy over time and high maintenance and update costs.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a data asset-driven dynamic business decision-making system, comprising:
[0006] The data assetization layer is used to connect to multi-source heterogeneous data sources, process the raw data, and form a standardized data asset library, which includes historical data assets and real-time data stream assets.
[0007] The feature engineering and model layer, connected to the data assetization layer, is used to extract features from the data asset library and to build, train, and deploy multiple business decision machine learning models.
[0008] The dynamic decision engine, connected to the feature engineering and model layer, is used to receive business decision requests, call the corresponding machine learning models and real-time features, and generate and output the optimal decision.
[0009] The decision execution and feedback loop, connected to the dynamic decision engine, is used to send the optimal decision to the target business system for execution and collect the effect feedback data after the decision execution. The effect feedback data is sent back to the data assetization layer.
[0010] The Asset Management and Performance Monitoring Center connects to the Data Assetization Layer and the Feature Engineering and Model Layer to manage, monitor, and evaluate the performance of data assets, model assets, and decision pipelines.
[0011] Preferably, the dynamic decision engine includes:
[0012] The decision orchestration module is used to organize multiple models for collaborative decision-making based on the decision-making scenario;
[0013] The multi-armed gambling machine algorithm is used to balance the exploration of new decision-making strategies with the utilization of known optimal strategies;
[0014] The decision interpretation module is used to provide interpretability analysis for the generated decisions.
[0015] Preferably, the decision execution and feedback loop integrates an A / B testing framework for conducting online comparative experiments between the new decision-making strategy and the old strategy.
[0016] Preferably, the feature engineering and model layer includes an online feature service platform for providing low-latency real-time feature query services for the dynamic decision engine.
[0017] Preferably, after the effect feedback data is fed back to the data assetization layer, it is used to trigger incremental learning or retraining of the business decision machine learning model.
[0018] A decision-making method for a data asset-driven dynamic business decision-making system includes the following steps:
[0019] S1: Build and maintain a unified data asset library through the data assetization layer;
[0020] S2: Feature extraction, model training, and deployment are performed based on a data asset library through feature engineering and the model layer;
[0021] S3: Receives business decision requests through the dynamic decision engine, obtains real-time features, calls the decision model, and outputs the final decision result;
[0022] S4: Implement the decision-making process through a decision execution and feedback loop, and collect feedback data.
[0023] S5: The feedback data is fed back to the data assetization layer to update data assets and trigger model optimization;
[0024] S6: Through the Asset Management and Performance Monitoring Center, continuously evaluate the performance of each link in the system and make adjustments.
[0025] Preferably, the step of outputting the final decision result in step S3 includes: using a multi-armed gambling machine algorithm to weigh between exploring new strategies and using known optimal strategies.
[0026] Preferably, after the step of collecting feedback data in step S4, the method includes: using an A / B testing framework to evaluate the performance improvement of the new decision-making strategy relative to the old strategy.
[0027] Compared with the prior art, the beneficial effects of the present invention are:
[0028] 1. This invention integrates multi-source heterogeneous data into standardized, reusable, high-quality data assets by constructing a unified data assetization layer and a modular system architecture. It also achieves loosely coupled management of features, models, and decision-making processes, solving the data silo problem and significantly improving the reuse value of data and the flexibility and scalability of the system.
[0029] 2. This invention enables the system to respond to the business environment in milliseconds and continuously self-optimize through the cooperation of a dynamic decision engine and decision execution and feedback loop. At the same time, by using decision interpretation and A / B testing framework, it ensures the real-time performance, adaptability and reliability of the decision-making process, thereby comprehensively improving the accuracy and agility of business decisions. Attached Figure Description
[0030] Figure 1 This is a system framework diagram of the present invention;
[0031] Figure 2 This is a schematic diagram of the overall process of the present invention. Detailed Implementation
[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0033] Please see Figures 1 to 2 One embodiment of the present invention provides a data asset-driven dynamic business decision-making system, comprising:
[0034] The data assetization layer is used to connect to multiple heterogeneous data sources, process the raw data, and form a standardized data asset library, which includes historical data assets and real-time data stream assets.
[0035] The feature engineering and model layer, connected to the data assetization layer, is used to extract features from the data asset repository and to build, train, and deploy multiple business decision-making machine learning models.
[0036] The dynamic decision engine, connected to the feature engineering and model layers, is used to receive business decision requests, call the corresponding machine learning models and real-time features, and generate and output the optimal decision.
[0037] The decision execution and feedback loop, connected to the dynamic decision engine, is used to send the optimal decision to the target business system for execution and to collect the effect feedback data after the decision execution. The effect feedback data is sent back to the data assetization layer.
[0038] The Asset Management and Performance Monitoring Center connects to the Data Assetization Layer and the Feature Engineering and Model Layer to manage, monitor, and evaluate the performance of data assets, model assets, and decision pipelines.
[0039] The dynamic decision engine includes:
[0040] The decision orchestration module is used to organize multiple models for collaborative decision-making based on the decision-making scenario;
[0041] The multi-armed gambling machine algorithm is used to balance the exploration of new decision-making strategies with the utilization of known optimal strategies;
[0042] The decision interpretation module is used to provide interpretability analysis for the generated decisions.
[0043] The decision execution and feedback loop integrates an A / B testing framework for conducting online comparative experiments between new and old decision strategies.
[0044] The feature engineering and model layer includes an online feature service platform that provides low-latency, real-time feature query services for the dynamic decision engine.
[0045] After the feedback data is fed back to the data assetization layer, it is used to trigger incremental learning or retraining of the machine learning model for business decision-making.
[0046] A decision-making method for a data asset-driven dynamic business decision-making system includes the following steps:
[0047] S1: Build and maintain a unified data asset library through the data assetization layer;
[0048] S2: Feature extraction, model training, and deployment are performed based on a data asset library through feature engineering and the model layer;
[0049] S3: Receives business decision requests through the dynamic decision engine, obtains real-time features, calls the decision model, and outputs the final decision result;
[0050] S4: Implement the decision-making process through a decision execution and feedback loop, and collect feedback data.
[0051] S5: Feedback data is fed back to the data assetization layer to update data assets and trigger model optimization;
[0052] S6: Through the Asset Management and Performance Monitoring Center, continuously evaluate the performance of each link in the system and make adjustments.
[0053] Step S3, which outputs the final decision result, involves using the multi-armed gambling machine algorithm to weigh the trade-off between exploring new strategies and utilizing known optimal strategies.
[0054] Following the step of collecting feedback data in step S4, the following steps are included: using an A / B testing framework to evaluate the performance improvement of the new decision-making strategy relative to the old strategy.
[0055] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
Claims
1. A data asset-driven dynamic business decision-making system, characterized in that, include: The data assetization layer is used to connect to multi-source heterogeneous data sources, process the raw data, and form a standardized data asset library, which includes historical data assets and real-time data stream assets. The feature engineering and model layer, connected to the data assetization layer, is used to extract features from the data asset library and to build, train, and deploy multiple business decision machine learning models. The dynamic decision engine, connected to the feature engineering and model layer, is used to receive business decision requests, call the corresponding machine learning models and real-time features, and generate and output the optimal decision. The decision execution and feedback loop, connected to the dynamic decision engine, is used to send the optimal decision to the target business system for execution and collect the effect feedback data after the decision execution. The effect feedback data is sent back to the data assetization layer. The Asset Management and Performance Monitoring Center connects to the Data Assetization Layer and the Feature Engineering and Model Layer to manage, monitor, and evaluate the performance of data assets, model assets, and decision pipelines.
2. The data asset-driven dynamic business decision-making system according to claim 1, characterized in that, The dynamic decision engine includes: The decision orchestration module is used to organize multiple models for collaborative decision-making based on the decision-making scenario; The multi-armed gambling machine algorithm is used to balance the exploration of new decision-making strategies with the utilization of known optimal strategies; The decision interpretation module is used to provide interpretability analysis for the generated decisions.
3. The data asset-driven dynamic business decision-making system according to claim 1, characterized in that, The decision execution and feedback loop integrates an A / B testing framework for conducting online comparative experiments between new and old decision-making strategies.
4. The data asset-driven dynamic business decision-making system according to claim 1, characterized in that, The feature engineering and model layer includes an online feature service platform for providing low-latency, real-time feature query services for the dynamic decision engine.
5. The data asset-driven dynamic business decision-making system according to claim 1, characterized in that, After the feedback data is fed back to the data assetization layer, it is used to trigger incremental learning or retraining of the business decision machine learning model.
6. A decision-making method for a data asset-driven dynamic business decision-making system according to any one of claims 1-5, characterized in that, Includes the following steps: S1: Build and maintain a unified data asset library through the data assetization layer; S2: Feature extraction, model training, and deployment are performed based on a data asset library through feature engineering and the model layer; S3: Receives business decision requests through the dynamic decision engine, obtains real-time features, calls the decision model, and outputs the final decision result; S4: Implement the decision-making process through a decision execution and feedback loop, and collect feedback data. S5: The feedback data is fed back to the data assetization layer to update data assets and trigger model optimization; S6: Through the Asset Management and Performance Monitoring Center, continuously evaluate the performance of each link in the system and make adjustments.
7. The decision-making method of the data asset-driven dynamic business decision-making system according to claim 6, characterized in that, The step of outputting the final decision result in step S3 includes: using the multi-armed gambling machine algorithm to weigh between exploring new strategies and using known optimal strategies.
8. The decision-making method of the data asset-driven dynamic business decision-making system according to claim 6, characterized in that, Following the step of collecting feedback data in step S4, the following steps are included: evaluating the performance improvement of the new decision-making strategy relative to the old strategy using an A / B testing framework.