A multi-source heterogeneous data credible question and answer traceability method, device and medium for investment project management

Abstract

The present application relates to the field of computer technology, specifically a multi-source heterogeneous data credible question and answer traceability method for investment project management, equipment and medium, including investment field double knowledge base construction, multi-modal information fusion and quantitative processing, initial weight module based on knowledge, multi-project game competition dynamic module, multi-attribute comprehensive decision evaluation calculation module, index ablation precise discrimination module, priority ranking and decision output module, knowledge base dynamic iteration update module, main control chip, data optimization module and power supply module. The present application realizes more practical "optimizing the best" or "short board excavation" through the multi-project game competition dynamic module, which makes the evaluation standard adapt to the specific competitive environment and avoids the rigid problem of static evaluation, and in combination with the index ablation precise discrimination module, the real competitiveness and risk points of the project can be identified by penetrating the surface score, greatly improving the accuracy of project ranking and classification.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically, to a method, device, and medium for tracing the source of multi-source heterogeneous data in investment project management using trusted question-and-answer methods. Background Technology

[0002] Investment decisions rely on a wide range of data sources, including both quantitative and qualitative information. These diverse and heterogeneous data sources have inconsistent standards and mixed formats, making unified measurement and effective integration difficult. This leads to insufficient information utilization. Traditional evaluation methods often employ a fixed set of evaluation criteria and weights. This fails to adapt to changes in the relative strengths and weaknesses of different batches of projects, potentially resulting in a "choosing the best among the worst" approach in a dimension where all projects in a batch are weak, or ineffective comparisons in highly homogeneous dimensions. This reduces the discriminatory power and scientific rigor of the selection process. Many complex evaluation models output results but lack interpretability, making it difficult for decision-makers to understand why a particular project received a high or low score, and hindering the ability to trace key scoring and demerit points. This renders the decision-making process unauditable and unreliable, especially failing to meet the traceability requirements of risk control and compliance. The investment environment and evaluation standards are dynamic, but traditional knowledge bases and evaluation rules are often slowly updated, failing to incorporate the latest post-investment performance feedback, industry policy changes, and emerging risk factors. This causes evaluation models to gradually become out of touch with reality. Therefore, a reliable question-and-answer traceability method, equipment, and media for multi-source heterogeneous data in investment project management are needed to address these issues. Summary of the Invention

[0003] The purpose of this invention is to provide a reliable question-and-answer tracing method, device, and medium for multi-source heterogeneous data in investment project management, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a reliable question-and-answer tracing method for multi-source heterogeneous data in investment project management, comprising the construction of dual knowledge bases in the investment field, multimodal information fusion and quantitative processing, a knowledge-based initial weight module, a multi-project game competition dynamic module, a multi-attribute comprehensive decision evaluation calculation module, an indicator ablation precise discrimination module, a priority ranking and decision output module, a knowledge base dynamic iterative update module, a main control chip, a data optimization module, and a power supply module; The process is divided into several steps: S1, construction of a dual knowledge base in the investment field; S2, multimodal information fusion and quantitative processing; S3, knowledge-based initial weight module; S4, multi-project game competition dynamic module; S5, multi-attribute comprehensive decision evaluation and calculation module; S6, indicator ablation and precise discrimination module; S7, priority ranking and decision output module; and S8, knowledge base dynamic iterative update module. The dual knowledge base construction, multimodal information fusion and quantitative processing, knowledge-based initial weight module, multi-project game competition dynamic module, multi-attribute comprehensive decision evaluation and calculation module, indicator ablation and precise discrimination module, priority ranking and decision output module, and knowledge base dynamic iterative update module are all connected to the main control chip. The main control chip is connected to the data optimization module and the power supply module.

[0005] As a preferred embodiment of the present invention, the data optimization module is connected to the operation and maintenance and deployment module, and the operation and maintenance and deployment module is connected to the data storage and backup module.

[0006] As a preferred technical solution of the present invention, the dual knowledge base construction in the investment field includes an investment planning knowledge base, a historical case knowledge base, and a knowledge enhancement embedding unit.

[0007] As a preferred technical solution of the present invention, the multimodal information fusion and quantification processing includes a quantitative index preprocessing unit, a qualitative index grading unit, and a knowledge base numerical correction unit.

[0008] As a preferred embodiment of the present invention, the knowledge-based initial weight module includes an AHP hierarchical analysis unit and a combined weight model unit.

[0009] As a preferred technical solution of the present invention, the multi-attribute comprehensive decision evaluation calculation module includes a unit for constructing an index evaluation matrix, a unit for calculating the comprehensive membership vector of the project, and a unit for judging the level threshold.

[0010] As a preferred technical solution of the present invention, the multi-project game competition dynamic module can perform global statistics on all projects to be evaluated in the current batch, analyze the overall distribution of advantages and disadvantages of each indicator, and automatically increase the weight of a certain dimension if a certain dimension is generally weak, thereby identifying high-quality projects in the weak dimension. If a certain dimension is highly homogeneous, the weight of that dimension is reduced to avoid ineffective differentiation. Through game equilibrium iteration, dynamic game weights adapted to the current batch of projects are output to achieve accurate selection of the best projects in the batch.

[0011] As a preferred technical solution of the present invention, the indicator ablation precision discrimination module adopts the indicator layer-by-layer ablation analysis method to eliminate low contribution and low sensitivity indicators layer by layer, recalculate the project score, observe the score fluctuation range, determine the core advantage dimension and rigid weakness dimension of the project, and lock the true importance level of the project based on the level change before and after ablation, thus solving the problems of difficulty in ranking with the same score and false high scores.

[0012] As a preferred technical solution of the present invention, the priority ranking and decision output module generates a global priority ranking of projects based on the comprehensive score and ablation sensitivity level. The projects are divided into strategic core projects, benefit-oriented key projects, reserve cultivation projects, and restricted optimization projects. The module can automatically archive evaluation link data to achieve full auditability and traceability.

[0013] As a preferred technical solution of the present invention, the knowledge base dynamic iterative update module is divided into a rule knowledge base that is updated quarterly based on the newly added post-investment results and industry policies. By reverse-entering the evaluation results and implementation effects into the case knowledge base, the membership function and weight correction strategy are iteratively optimized to continuously improve the evaluation accuracy.

[0014] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention is a reliable question-and-answer tracing method, device and medium for multi-source heterogeneous data in investment project management. Through the dynamic module of multi-project game competition, this invention enables the evaluation criteria to adapt to the specific competitive environment, avoids the rigidity of static evaluation, and realizes more realistic "selection of the best" or "weakness mining". Combined with the indicator ablation precise judgment module, it can penetrate the apparent score and identify the true competitiveness and risk points of the project, which greatly improves the accuracy of project ranking and classification.

[0015] (2) This invention is a method, device, and medium for tracing the source of multi-source heterogeneous data through trusted question-and-answer methods for investment project management. From data input, quantification, weight calculation, comprehensive evaluation to ablation analysis, the data, parameters, and intermediate results of each stage of the system are automatically archived by the priority sorting and decision output module. Any final decision can be traced back to the specific indicator score, weight assignment, or even the quantification process of a certain qualitative information, meeting the stringent requirements of internal risk control, auditing, and compliance for decision transparency. Supported by the multimodal information fusion and quantification processing module and dual knowledge bases, the system can uniformly incorporate information in different forms, such as financial statements, text reports, and expert opinions, into a computable evaluation framework, breaking down information silos and making decision-making more comprehensive. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is an overall block diagram of a multi-source heterogeneous data trusted question-and-answer tracing method, device and medium for investment project management according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a multi-source heterogeneous data trusted question-and-answer tracing method, equipment, and medium for investment project management, according to an embodiment of the present invention, for constructing a dual knowledge base in the investment field; Figure 3 This is a schematic diagram of the structure of a multi-source heterogeneous data trusted question-and-answer tracing method, device, and medium for multimodal information fusion and quantification processing for investment project management, according to an embodiment of the present invention. Figure 4 This is a schematic diagram of the structure of a knowledge-based initial weighting module for a multi-source heterogeneous data trusted question-answering tracing method, device, and medium for investment project management according to an embodiment of the present invention. Figure 5 This is a schematic diagram of the structure of a multi-source heterogeneous data trusted question-and-answer tracing method, equipment, and media multi-attribute comprehensive decision-making evaluation calculation module for investment project management, according to an embodiment of the present invention. Detailed Implementation

[0018] The invention will now be further described with reference to the accompanying drawings and specific embodiments: Example

[0019] refer to Figures 1-5 Example 1 describes the following: construction of a dual knowledge base in the investment field, multimodal information fusion and quantitative processing, knowledge-based initial weight module, multi-project game competition dynamic module, multi-attribute comprehensive decision evaluation calculation module, indicator ablation and accurate discrimination module, priority ranking and decision output module, knowledge base dynamic iterative update module, main control chip, data optimization module and power supply module. The process is divided into several steps: S1, construction of a dual knowledge base in the investment field; S2, multimodal information fusion and quantitative processing; S3, knowledge-based initial weight module; S4, multi-project game competition dynamic module; S5, multi-attribute comprehensive decision evaluation and calculation module; S6, indicator ablation and precise discrimination module; S7, priority ranking and decision output module; and S8, knowledge base dynamic iterative update module. The dual knowledge base construction, multimodal information fusion and quantitative processing, knowledge-based initial weight module, multi-project game competition dynamic module, multi-attribute comprehensive decision evaluation and calculation module, indicator ablation and precise discrimination module, priority ranking and decision output module, and knowledge base dynamic iterative update module are all connected to the main control chip. The main control chip is connected to the data optimization module and the power supply module. In this implementation, the dual knowledge base construction in the investment field includes an investment planning knowledge base, a historical case knowledge base, and a knowledge enhancement embedding unit. The investment planning knowledge base solidifies industry investment thresholds, policy red lines, access conditions, return benchmarks, risk thresholds, and industry guidance rules, forming structured rule entries. The historical case knowledge base collects the indicator characteristics and review conclusions of invested, uninvested, and reserve projects over the years, forming a case sample library. The knowledge enhancement embedding unit uses the knowledge base rules as prior constraints for the model to guide indicator scoring, threshold determination, and anomaly correction, solving the problems of insufficient data and sparse samples.

[0020] In this embodiment, multimodal information fusion and quantitative processing includes a quantitative indicator preprocessing unit, a qualitative indicator grading unit, and a knowledge base numerical correction unit. The quantitative indicator preprocessing unit standardizes and normalizes numerical indicators such as investment amount, rate of return, payback period, and growth rate. The qualitative indicator grading unit classifies textual evaluations of policy adaptability, team capability, technological advancement, and market prospects into excellent, good, and poor categories. The knowledge base numerical correction unit ensures that the quantitative results conform to common sense in industry investment. Example

[0021] refer to Figure 1Example 2 further illustrates Example 1, including the connection between the data optimization module and the operation and deployment module, and the connection between the operation and deployment module and the data storage and backup module. The multi-project game competition dynamic module can perform global statistics on all projects to be evaluated in the current batch, analyzing the overall distribution of strengths and weaknesses of each indicator. If a certain dimension is generally weak, its weight is automatically increased to identify high-quality projects in the weaker dimension. If a certain dimension is highly homogeneous, its weight is reduced to avoid ineffective differentiation. Through game equilibrium iteration, dynamic game weights adapted to the current batch of projects are output, achieving accurate selection of batch projects. The indicator ablation and accurate discrimination module uses a layer-by-layer indicator ablation analysis method to eliminate low-contribution and low-sensitivity indicators layer by layer, recalculating the project results. The system analyzes and observes score fluctuations to determine the core strengths and weaknesses of projects. Based on changes in the level before and after ablation, it identifies the true importance level of projects, addressing issues such as difficulty in ranking projects with the same score and pseudo-high scores. The priority ranking and decision output module generates a global priority ranking of projects based on the comprehensive score and ablation sensitivity level, categorizing them into strategic core projects, benefit-focused projects, reserve and cultivation projects, and restricted optimization projects. It can automatically archive evaluation link data, achieving full auditability and traceability. The knowledge base dynamic iteration update module updates the rule knowledge base quarterly based on new post-investment results and industry policies. By reverse-engineering the evaluation results and implementation effects into the case knowledge base, it uses iterative optimization of membership functions and weight correction strategies to continuously improve the accuracy of the evaluation. In this embodiment, the knowledge-based initial weight module includes an AHP hierarchical analysis unit and a combined weight model unit. The AHP hierarchical analysis unit can solve for subjective experience weights; the combined weight model unit realizes bidirectional weighting of "knowledge and experience + real data" to obtain the initial basic weights of each indicator.

[0022] In this embodiment, the multi-attribute comprehensive decision evaluation calculation module includes a unit for constructing an indicator evaluation matrix, a unit for calculating the comprehensive membership vector of an item, and a unit for judging the level threshold. The unit for constructing the indicator evaluation matrix combines dynamic game weights to complete the multi-dimensional indicator fusion calculation; the unit for calculating the comprehensive membership vector of an item can obtain the comprehensive excellence score of the item; and the unit for judging the level threshold can complete the preliminary importance classification of the item.

[0023] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A reliable question-and-answer tracing method for multi-source heterogeneous data in investment project management, characterized in that, It includes a dual knowledge base construction for the investment field, multimodal information fusion and quantitative processing, a knowledge-based initial weight module, a multi-project game competition dynamic module, a multi-attribute comprehensive decision evaluation and calculation module, an indicator ablation and precise discrimination module, a priority ranking and decision output module, a knowledge base dynamic iteration update module, a main control chip, a data optimization module, and a power supply module; The process is divided into several steps: S1, construction of a dual knowledge base in the investment field; S2, multimodal information fusion and quantitative processing; S3, knowledge-based initial weight module; S4, multi-project game competition dynamic module; S5, multi-attribute comprehensive decision evaluation and calculation module; S6, indicator ablation and precise discrimination module; S7, priority ranking and decision output module; and S8, knowledge base dynamic iterative update module. The dual knowledge base construction, multimodal information fusion and quantitative processing, knowledge-based initial weight module, multi-project game competition dynamic module, multi-attribute comprehensive decision evaluation and calculation module, indicator ablation and precise discrimination module, priority ranking and decision output module, and knowledge base dynamic iterative update module are all connected to the main control chip. The main control chip is connected to the data optimization module and the power supply module.

2. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The data optimization module is connected to the operation and maintenance and deployment module, and the operation and maintenance and deployment module is connected to the data storage and backup module.

3. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The dual knowledge base construction in the investment field includes an investment planning knowledge base, a historical case knowledge base, and knowledge enhancement embedding units.

4. The reliable question-and-answer tracing method for multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The multimodal information fusion and quantification process includes a quantitative indicator preprocessing unit, a qualitative indicator grading unit, and a knowledge base numerical correction unit.

5. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The knowledge-based initial weight module includes an AHP (Analytic Hierarchy Process) unit and a combined weight model unit.

6. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The multi-attribute comprehensive decision evaluation calculation module includes a unit for constructing an indicator evaluation matrix, a unit for calculating the comprehensive membership vector of the project, and a unit for judging the level threshold.

7. The method for tracing the source of multi-source heterogeneous data for investment project management according to claim 1, characterized in that, The multi-project game competition dynamic module can perform global statistics on all projects to be evaluated in the current batch, analyze the overall strengths and weaknesses of each indicator, and automatically increase the weight of a certain dimension if a certain dimension is generally weak, thus identifying high-quality projects in the weak dimension. If a certain dimension is highly homogeneous, the weight of that dimension is reduced to avoid ineffective differentiation. Through game equilibrium iteration, dynamic game weights that are suitable for the current batch of projects are output, enabling accurate selection of the best projects in the batch.

8. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The indicator ablation and precise discrimination module uses the indicator layer-by-layer ablation analysis method to eliminate low-contribution and low-sensitivity indicators layer by layer, recalculate the project score, observe the score fluctuation range, determine the core strength dimension and rigid weakness dimension of the project, and lock in the true importance level of the project based on the level change before and after ablation, thus solving the problems of difficulty in ranking tied scores and false high scores.

9. The method for tracing the source of multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The priority ranking and decision output module generates a global priority ranking of projects based on a comprehensive score and ablation sensitivity level. The projects are divided into strategic core projects, benefit-oriented key projects, reserve and cultivation projects, and restricted optimization projects. The module can automatically archive evaluation link data to achieve full auditability and traceability.

10. A reliable question-and-answer tracing method for multi-source heterogeneous data in investment project management according to claim 1, characterized in that, The knowledge base dynamic iteration update module is divided into quarterly rule knowledge base updates based on new post-investment results and industry policies. By reverse-engineering the evaluation results and implementation effects into the case knowledge base, the membership function and weight correction strategy are iteratively optimized to continuously improve the evaluation accuracy.

Images