A data asset-driven small and medium-sized enterprise intelligent decision support system and implementation method

Abstract

A data asset-driven intelligent decision support system for SMEs and its implementation method are disclosed. The system includes a data access and governance module, a data asset encapsulation module, a data asset value calculation module, a decision input construction module, a decision reasoning module, a decision result evaluation module, a parameter self-updating module, and a decision release and display module. By standardizing and encapsulating multi-source business data into assets, and constructing a data asset value evaluation model based on data quality, business relevance, timeliness, and historical decision contribution, high-value data assets are prioritized for participation in the decision model reasoning process. Simultaneously, through feedback evaluation of decision execution results and parameter self-updating, closed-loop optimization control of data asset weights and decision model parameters is achieved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent decision support technology, specifically referring to a data asset-driven intelligent decision support system for small and medium-sized enterprises and its implementation method. Background Technology

[0002] With the development of information and digital technologies, SMEs have gradually deployed business systems such as Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and Financial Management System (FMS), accumulating a large amount of operational data. However, the current data utilization methods of SMEs are still mainly based on report statistics and experience-based decision-making. The data has not formed a systematic asset management mechanism, making it difficult to support the needs of high-frequency, dynamic, and scenario-based intelligent decision-making.

[0003] In existing technologies, some enterprises use data warehouses or business intelligence (BI) tools to centrally analyze business data. However, such solutions mainly focus on the statistical display of historical data and lack systematic quantitative modeling of data quality, timeliness and business relevance. It is difficult to distinguish between high-value data and low-value data, resulting in large noise in the input data of decision-making models and insufficient prediction accuracy and stability.

[0004] Furthermore, existing intelligent decision-making systems are typically centered around algorithmic models, lacking a unified governance and value assessment mechanism at the data asset level. The selection of model input features relies on human experience or static configuration, making it difficult to dynamically optimize as the business environment changes. This results in insufficient generalization ability and poor adaptability of the decision-making models. At the same time, most systems lack the ability to perform closed-loop feedback correction of data value weights and model parameters based on decision execution results, thus failing to achieve continuous self-learning and performance evolution. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention provides a data asset-driven intelligent decision support system for small and medium-sized enterprises and its implementation method, so as to at least partially solve the above-mentioned technical problems.

[0006] The technical solution adopted in this invention is as follows: This invention proposes a data asset-driven intelligent decision support system for SMEs, comprising the following modules: a data access and governance module, used to collect business data from enterprise resource planning (ERP) systems, customer management systems, and financial management systems, and to perform field standardization, unit unification, missing value imputation, and outlier removal on the business data to generate structured business data; a data asset encapsulation module, used to encapsulate the structured business data into data asset objects, each data asset object including at least a data identifier field, a data source field, a quality score field, a timeliness weight field, and a business association weight field; a data asset value calculation module, used to calculate the data asset value score of each data asset object based on the quality score field, timeliness weight field, business association weight field, and historical decision contribution field; and a decision input construction module, used to construct the decision input according to the data asset-driven intelligent decision support system for SMEs. The system ranks data asset objects based on their value scores and selects those with scores higher than a preset threshold as the decision input set. The decision reasoning module inputs the data asset-driven intelligent decision support system for SMEs into the prediction model unit and the rule reasoning unit, respectively, and generates at least one candidate decision based on the outputs of the two models. The decision result evaluation module collects the actual business results after the candidate decision is executed and calculates the deviation between the predicted and actual results. The parameter self-updating module adjusts the weight parameters in the data asset value scoring model based on the deviation value. The decision release and display module visualizes the candidate decision, prediction performance indicators, and risk assessment indicators, and pushes the confirmed decision to the business execution system.

[0007] Furthermore, the data access and governance module of the data asset-driven intelligent decision support system for SMEs includes: a field mapping unit, a data completion unit, and an anomaly detection unit. The field mapping unit is used to map synonymous fields in different business systems to unified field names. The data completion unit is used to fill in missing fields based on historical averages, medians, or business rules. The anomaly detection unit is used to identify and remove or correct abnormal data based on threshold interval methods or distribution offset detection methods.

[0008] Furthermore, the quality scoring fields of the data asset-driven intelligent decision support system for SMEs include completeness rate, consistency rate, and timeliness rate subfields; the timeliness weight field of the data asset-driven intelligent decision support system for SMEs is calculated based on the interval between the data collection time and the current time; and the business relevance weight field of the data asset-driven intelligent decision support system for SMEs is generated based on the ranking of the feature importance of this data field in the historical decision model.

[0009] Furthermore, the data asset value calculation module of the data asset-driven intelligent decision support system for SMEs calculates the data asset value score according to the following formula: Where Q represents the quality score field, R represents the business association weight field, T represents the timeliness weight field, H represents the historical decision contribution field, and a, b, c, and d are adjustable weight parameters stored in the parameter configuration table.

[0010] Furthermore, the predictive model unit of the data asset-driven intelligent decision support system for SMEs is used to generate business indicator prediction results based on regression models, classification models, or time series models; the rule reasoning unit of the data asset-driven intelligent decision support system for SMEs is used to generate constraint judgment results based on threshold judgment rules, logical combination rules, or compliance verification rules; and the decision reasoning module of the data asset-driven intelligent decision support system for SMEs generates decision candidate solutions containing recommended action values ​​and risk level identifiers based on the output results of the predictive model and the output results of the rule reasoning.

[0011] Furthermore, the decision result evaluation module of the data asset-driven intelligent decision support system for SMEs calculates decision performance evaluation parameters, including mean square error, absolute deviation, or hit rate, by comparing the predicted results with the actual business results.

[0012] Furthermore, the parameter self-updating module of the data asset-driven intelligent decision support system for SMEs updates the weight parameters according to the following process: when the decision effect evaluation parameter of the data asset-driven intelligent decision support system for SMEs is lower than the preset performance threshold, the business association weight field corresponding to the data asset object participating in the decision is increased; when the decision effect evaluation parameter of the data asset-driven intelligent decision support system for SMEs is higher than the preset performance threshold, the current weight parameter is kept unchanged or the weight field corresponding to the data asset object with low participation is decreased.

[0013] This invention discloses a data asset-driven intelligent decision support method for SMEs, characterized by the following steps: S1, acquiring business data from multiple business systems and performing field standardization, missing value imputation, and outlier removal; S2, encapsulating the processed business data into a data asset object containing a quality score field, a timeliness weight field, and a business relevance weight field; S3, calculating the data asset value score based on the data asset-driven intelligent decision support system fields and filtering to form a decision input set; S4, inputting the decision input set of the data asset-driven intelligent decision support system into the prediction model unit and the rule inference unit respectively to generate decision candidate schemes; S5, collecting the actual business results after the decision candidate schemes are executed and updating the weight parameters.

[0014] Furthermore, when selecting and forming the decision input set, only data asset objects with a data asset value score higher than the threshold corresponding to the current decision scenario are selected to participate in the model calculation.

[0015] Furthermore, by accumulating the statistical values ​​of prediction errors from multiple decision execution results, the weight parameters in the data asset value scoring model are periodically recalibrated.

[0016] Compared with the prior art, the present invention has the following advantages: By constructing a data asset management system and a multi-model collaborative decision-making engine, the scattered, heterogeneous, and low-quality data resources of SMEs are transformed into measurable, assessable, circulated, and reusable data assets. This achieves a transformation and upgrade from "experience-driven decision-making" to "data asset-driven decision-making," solving the technical problems commonly found in SMEs, such as low data accumulation value, reliance on human experience in decision-making, and lagging response to market changes. This significantly improves the scientific nature and real-time performance of enterprise management.

[0017] By introducing data quality assessment, asset value scoring, business scenario tagging modeling, and adaptive model orchestration mechanisms, the system can automatically select the optimal analysis path and prediction model based on different enterprise sizes, industry characteristics, and business scenarios. This reduces the threshold for algorithm deployment and maintenance, enabling SMEs to achieve intelligent decision-making functions such as sales forecasting, inventory optimization, cash flow risk warning, and customer churn prevention without having a professional data team. This also improves the system's scalability, generalization ability, and engineering feasibility.

[0018] By constructing a closed-loop decision feedback mechanism and a self-learning model, the dynamic evolution of data asset value and continuous iterative optimization of the decision model are realized. This enables the system to continuously improve prediction accuracy and strategy matching as business operation status changes, thereby significantly enhancing the operational resilience, resource allocation efficiency, and risk resistance of SMEs while ensuring system stability. It has good value for large-scale deployment and promising prospects for industrial applications. Attached Figure Description

[0019] Figure 1 This is an architecture diagram of the data asset-driven intelligent decision support system for SMEs proposed in Embodiment 1 of the present invention.

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0021] 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. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0023] like Figure 1As shown, this embodiment provides a data asset-driven intelligent decision support system for SMEs, including a data access and governance module, a data asset encapsulation module, a data asset value calculation module, a decision input construction module, a decision reasoning module, a decision result evaluation module, a parameter self-updating module, and a decision release and display module. Each module is connected to communicate through an internal enterprise data bus or message queue, forming a data asset-driven closed-loop intelligent decision system.

[0024] The data access and governance module establishes data interface connections with Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and Financial Management System (FMS) to periodically or in real-time collect order data, inventory data, customer transaction behavior data, accounts receivable and payable data, and cash flow data. This module performs field standardization, unit unification, missing value imputation, and outlier removal on the collected business data. Specifically, it uses a field mapping table to unify semantically identical but differently named fields from different systems to standard field names; it imputes missing fields using historical averages, medians, or business rule models; and it identifies and removes or corrects outlier data using threshold interval detection or distribution offset detection methods, thereby generating structured business data and writing it into a unified business data pool.

[0025] The data asset encapsulation module encapsulates the structured business data into data asset objects and writes them into the data asset catalog table according to a preset asset metadata structure. Each data asset object includes at least a data identifier field, a data source field, a quality score field, a timeliness weight field, a business association weight field, and a historical decision contribution field. The quality score field includes completeness rate, consistency rate, and timeliness rate subfields; the timeliness weight field is calculated based on the time interval between the data collection timestamp and the current time, with a lower weight value for longer intervals; the business association weight field is generated based on the feature importance ranking of this data field in the historical prediction model; and the historical decision contribution field records the proportion of the data asset object's contribution to the reduction of prediction errors during historical decision-making processes.

[0026] The data asset value calculation module calculates the value of the data asset object based on the quality score field, timeliness weight field, business relevance weight field, and historical decision contribution field, using the following formula:

[0027] Wherein, Q represents the weighted composite value of the quality score field, R represents the business association weight field, T represents the timeliness weight field, H represents the historical decision contribution field, and a, b, c, and d are adjustable weight parameters stored in the parameter configuration table. The quality score field Q is obtained by weighting the completeness rate, consistency rate, and timeliness rate according to a preset ratio; the business association weight field R is obtained by normalizing the feature importance values ​​in the historical model; the timeliness weight field T is obtained by calculating the time difference between the data generation time and the current time using an exponential decay function; and the historical decision contribution field H is obtained by statistically analyzing the contribution ratio of the data asset object to the reduction of prediction error when participating in decision-making. The data asset value calculation module periodically or according to the decision trigger method recalculates the value score of the data asset object and writes the result into the asset value field for subsequent modules to call.

[0028] The decision input construction module sorts data asset objects in descending order according to the data asset value score, and selects only data asset objects with a value score higher than the threshold corresponding to the current decision scenario to form the decision input set. For example, when the decision scenario is inventory replenishment, the asset value score threshold is set to 0.65, and only data asset objects with scores higher than this threshold are included in the input feature set of the inventory demand forecasting model; when the decision scenario is cash flow risk warning, the asset value score threshold is set to 0.7, and only data asset objects with scores higher than this threshold are included in the input feature set of the risk prediction model, thereby reducing the interference of low-quality or low-relevance data on the output results of the decision model.

[0029] The decision-making reasoning module comprises a prediction model unit and a rule-based reasoning unit. The prediction model unit, based on regression, classification, or time series models, predicts operational indicators such as inventory demand, sales trends, and cash flow risk levels, outputting predicted values ​​and corresponding confidence intervals. The rule-based reasoning unit constrains and verifies the prediction results based on threshold judgment rules, logical combination rules, and compliance verification rules. For example, it triggers replenishment suggestions when inventory turnover falls below a preset threshold, triggers risk warnings when accounts receivable aging exceeds a set number of days, and triggers approval restrictions when purchase amounts exceed authorized limits. The decision-making reasoning module integrates the prediction model output and the rule-based reasoning output to generate decision candidate solutions that include recommended action values, risk level indicators, and execution priority indicators.

[0030] The decision outcome evaluation module establishes an interface connection with the business execution system to obtain actual business outcome data after the execution of candidate decision solutions, and compares and analyzes the predicted results with the actual results. When the decision solution is an inventory replenishment strategy, it obtains the actual inventory consumption and sales completion volume, and calculates the mean square error or absolute error value between the predicted inventory consumption value and the actual consumption value. When the decision solution is a customer churn warning strategy, it obtains the actual customer retention results and calculates the prediction hit rate indicator. The above error indicators and hit rate indicators are written into the decision evaluation result table as decision effectiveness evaluation parameters.

[0031] The parameter self-updating module dynamically adjusts the weight parameters in the data asset value scoring model based on the decision performance evaluation parameters. When the decision performance evaluation parameters are lower than a preset performance threshold, the value of the business association weight field corresponding to the data asset object participating in the decision is increased, while the value of the weight field corresponding to the data asset object with a low contribution to prediction deviation is decreased. When the decision performance evaluation parameters are higher than the preset performance threshold, the current weight parameters remain unchanged, or the weight fields corresponding to the data asset objects with low participation are slightly attenuated. The above weight adjustment results are written into the parameter configuration table and take effect immediately for subsequent data asset value scoring calculations, realizing the system's self-learning and evolution capabilities. In addition, the parameter self-updating module performs weighted summaries of prediction error indicators from multiple historical decision scenarios on a weekly or monthly basis, and periodically recalibrates the weight values ​​of a, b, c, and d in the parameter configuration table using a proportional adjustment method or gradient descent method, so that the quality score field, business association weight field, timeliness weight field, and historical decision contribution field maintain the optimal weight ratio in the overall decision accuracy.

[0032] The decision release and display module is used to visualize decision candidate solutions, predicted effect indicators and risk level indicators, and push the confirmed decision solutions to the inventory management system, procurement management system, sales management system or capital management system for execution, so as to realize end-to-end closed-loop control from data collection, asset modeling, decision generation to execution feedback.

[0033] Through the above structure and process, this embodiment realizes the transformation of multi-source business data of enterprises into standardized data asset objects, and drives the construction and inference execution of decision model input based on the data asset value assessment mechanism. At the same time, the asset weights and model parameters are continuously optimized through the decision result feedback mechanism, so that the system can form a stable and convergent self-learning ability in the actual business operation process, thereby significantly improving the accuracy, interpretability and execution reliability of business decisions of SMEs.

[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0035] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A data asset-driven intelligent decision support system for SMEs, characterized in that, include: The data access and governance module is used to collect business data from the enterprise resource planning system, customer management system and financial management system, and perform field standardization, unit unification, missing value filling and outlier removal on the business data to generate structured business data. A data asset encapsulation module is used to encapsulate the structured business data into data asset objects. Each data asset object includes at least a data identifier field, a data source field, a quality score field, a timeliness weight field, and a business association weight field. The data asset value calculation module is used to calculate the data asset value score of each data asset object based on the quality score field, the timeliness weight field, the business association weight field, and the historical decision contribution field. A decision input construction module is used to sort data asset objects according to the data asset value score, and select data asset objects with a value score higher than a preset threshold in the sorting results as the decision input set. The decision reasoning module is used to input the decision input set into the prediction model unit and the rule reasoning unit respectively, and generate at least one decision candidate scheme based on the output results of the two types of models. The decision result evaluation module is used to collect the actual business results after the decision candidate scheme is executed, and to calculate the deviation value between the predicted result and the actual result. A parameter self-updating module is used to adjust the weight parameters in the data asset value scoring model based on the deviation value; The decision release and display module is used to visualize the decision candidate schemes, predicted effect indicators and risk assessment indicators, and push the confirmed decision schemes to the business execution system.

2. The data asset-driven intelligent decision support system for SMEs according to claim 1, characterized in that, The data access and governance module includes: Field mapping unit, data completion unit, and anomaly detection unit; The field mapping unit is used to map synonymous fields in different business systems to a unified field name; the data completion unit is used to fill in missing fields based on historical mean, median or business rules; the anomaly detection unit is used to identify abnormal data and remove or correct it based on threshold interval method or distribution offset detection method.

3. The data asset-driven intelligent decision support system for SMEs according to claim 2, characterized in that: The quality scoring field includes a completeness rate subfield, a consistency rate subfield, and a timeliness rate subfield; the timeliness weight field is calculated based on the interval between the data collection time and the current time; the business association weight field is generated based on the feature importance ranking result of this data field in the historical decision model.

4. The data asset-driven intelligent decision support system for SMEs according to claim 3, characterized in that, The data asset value calculation module calculates the data asset value score according to the following formula: Where Q represents the quality score field, R represents the business association weight field, T represents the timeliness weight field, H represents the historical decision contribution field, and a, b, c, and d are adjustable weight parameters stored in the parameter configuration table.

5. The data asset-driven intelligent decision support system for SMEs according to claim 4, characterized in that: The prediction model unit is used to generate business indicator prediction results based on regression models, classification models, or time series models; the rule reasoning unit is used to generate constraint judgment results based on threshold judgment rules, logical combination rules, or compliance verification rules; the decision reasoning module generates decision candidate solutions containing recommended action values ​​and risk level identifiers based on the prediction model output results and the rule reasoning output results.

6. The data asset-driven intelligent decision support system for SMEs according to claim 5, characterized in that: The decision result evaluation module calculates decision effectiveness evaluation parameters, including mean square error, absolute deviation, or hit rate, by comparing the predicted results with the actual business results.

7. The data asset-driven intelligent decision support system for SMEs according to claim 6, characterized in that, The parameter self-updating module updates the weight parameters according to the following process: When the decision effect evaluation parameter is lower than the preset performance threshold, the business association weight field corresponding to the data asset object participating in the decision is increased; when the decision effect evaluation parameter is higher than the preset performance threshold, the current weight parameter is kept unchanged or the weight field corresponding to the data asset object with low participation is decreased.

8. A data asset-driven intelligent decision support method for SMEs based on the system described in any one of claims 1 to 7, characterized in that, Includes the following steps: S1. Obtain business data from multiple business systems and perform field standardization, missing value imputation, and outlier removal; S2. Encapsulate the processed business data into a data asset object containing a quality score field, a timeliness weight field, and a business association weight field; S3. Calculate the data asset value score based on the fields and filter them to form a decision input set; S4. Input the decision input set into the prediction model unit and the rule reasoning unit respectively to generate decision candidate schemes; S5. Collect the actual business results after the decision candidate solutions are implemented and update the weight parameters.

9. The method according to claim 8, characterized in that, When selecting and forming the decision input set, only data asset objects with a data asset value score higher than the threshold corresponding to the current decision scenario are selected to participate in the model calculation.

10. The method according to claim 8, characterized in that, By accumulating the statistical values ​​of prediction errors from multiple decision executions, the weight parameters in the data asset value scoring model are periodically recalibrated.

Images