Relations between features for automatic recognition of abnormal cases in emergency telephone call systems

Abstract

This paper presents the use of relations between features, which describe a caller for automatic recognition of abnormal cases in emergency telephone call systems. The proposed correlation based procedure, as an integral part of the authors' advanced database search mechanism dedicated to emergency notification centers, extends the record matching procedures and takes into account dependences between features. Illustrative examples and experiments show that the presented procedure significantly improves recognition of impossible cases, e.g. those reported by cheaters.