Fuzzy Order-of-Magnitude Based Link Analysis for Qualitative Alias Detection

Numerical link-based similarity techniques have proven effective for identifying similar objects in the Internet and publication domains. However, for cases involving unduly high similarity measures, these methods usually generate inaccurate results. Also, they are often restricted to measuring over single properties only. This paper presents an order-of-magnitude based similarity mechanism that integrates multiple link properties to derive semantic-rich similarity descriptions. The approach extends conventional order-of-magnitude reasoning with the theory of fuzzy sets. The inherent ability of this work in computing-with-words also allows coherent interpretation and communication within a decision-making group. The proposed approach is applied to supporting the analysis of intelligence data. When evaluated over a difficult terrorism-related dataset, experimental results show that the approach helps to partly resolve the problem of false positives.