Hybrid Knowledge Graph and Transformer-Based Framework for Intelligent Decision Support in Enterprise Systems

Abstract

Enterprise decision support systems are a type of intelligent computing system that processes and analyses the enterprise data to make accurate decisions. Typically, machine learning and data analytics techniques are used to analyze enterprise data that are structured and unstructured for decision support. In many cases, conventional approaches to decision support lack deep semantic understanding, proper relationship modelling, explainability, and classification accuracy when dealing with enterprise data. To address these issues, this proposes a Hybrid Knowledge Graph and Transformer-Based Framework for Intelligent Decision Support in Enterprise Systems using the Knowledge Base Efficiency Evaluation Dataset. The framework begins with the implementation of the Data Entity Graph Building (DEGB) algorithm to pre-process data, extract entities, and build the relationship graph from the enterprise records. Furthermore, Multi Source Feature Integration Framework (MSFIF) then extracts semantic transformer features and graph-based embeddings to generate the integrated feature representation. Finally, Explainable Hybrid Decision Classification System (EHDCS) then performs enterprise classification and generates the positive or negative output with reasoning support for explainability. The proposed framework can achieve higher decision accuracy and learn better semantic relationships with more powerful feature representation capability and intelligent enterprise classification.