Design of an Emotion Profiler Using SNePS

SNePS is being used to implement an emotion profiler which is based on a new emotion theory. The system is designed to read a transcript of an actual doctor/patient conversation and produce a profile of the patient's emotional state. Because conversations are not restricted in topic and patients use words idiosyncratically, the emotion profiler's use of knowledge is limited. This paper describes a design and SNePS implementation which uses domain-independent knowledge to effectively parse and process transcripts.