Solution and Reference Recommendation System Using Knowledge Fusion and Ranking

Abstract

Building various types of recommendation systems has been a long-term goal for information management community in the last decades. Knowledge based recommendation systems have been studied widely in different areas to provide comprehensiveness information to users aiming at making the desired systems accessible and efficient. However, developing recommendation systems based on knowledge (academic papers) of subject-specific research fields has been neglected. Such systems can potentially return higher practical and theoretical values for both industry and academia communities. In this study, the concept of solution-oriented information network (SIN) is introduced which contains the information of research issues and proposed solutions embraced in the academic papers. A Knowledge based Solution and Reference Recommendation System (KSRRS) is then developed based on the knowledge included in the SIN. The constructed KSRRS ranks the solutions and recommends the superior solutions to any identified issue. KSRRS relies on two integrated modules: knowledge fusion module, and knowledge ranking module. The knowledge fusion module automatically extracts and reconstruct required information into a knowledge map, and the map is then augmented with a novel feature of knowledge ranking by using the ranking module. Evaluation experiment is constructed to build the customized knowledge map in a practical scenario in which intrusion detection is selected as the subject-specific field for demonstration.