A rule and model fusion-based enterprise credit reporting risk identification system
The enterprise credit report risk identification system, which integrates rules and models, overcomes the limitations of single rule engines and single machine learning models, and achieves accurate identification and interpretable output of enterprise risks, adapting to different data quality scenarios and business needs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LEXIANG DIGITAL CO LTD
- Filing Date
- 2026-02-10
- Publication Date
- 2026-06-02
AI Technical Summary
In existing technologies, single rule engines struggle to identify implicit and non-linear risks. Single machine learning models experience a sharp drop in prediction accuracy when data is scarce in SMEs, and they cannot provide clear risk explanations, leading to high false negative rates and cold start problems. Data and decision-making are disconnected, resulting in insufficient robustness.
The enterprise credit report risk identification system adopts a rule-based and model-based fusion approach, which includes a multi-source data parsing module, an expert rule engine module, a machine learning model module, a fusion decision module, and an application output module. It forms a closed-loop identification system through signal interconnection, dynamically adjusts weights to compensate for data gaps, and provides interpretable risk identification results.
It achieves comprehensive coverage of both explicit and implicit risks, improves the accuracy and robustness of risk identification, avoids the cold start problem, meets the interpretability requirements of financial regulation, and supports multi-format data access and business scenario adaptation.
Smart Images

Figure CN122134365A_ABST