Support vector machine-based multi-source multi-attribute information integration for situation assessment

Abstract

Understanding any given situation requires integrating many pieces of information. Such information has in most cases multiple attributes and is obtained from multiple data sources within multiple time slots. Situation assessors’ experience and preference will naturally influence the result of information integration, and hence influence the awareness generated for a situation. This study focuses on how multi-source multi-attribute information about a situation is integrated and how the awareness information for the situation is derived. A learning-based information integration approach, which embeds the fuzzy least squares support vector machine (FLS-SVM) technique, is developed in this study. This approach can assess a situation through integrating and inference obtained information and analyzing related data sources. A series of experiments show that the proposed approach has an accuracy learning ability from assessors’ experience in the information integration for generating awareness for a situation.