When
Ellen Riloff
Department Head, Professor
Computer Science
Interests: natural language processing and artificial intelligence
Zoom: https://arizona.zoom.us/j/86882451263
Identifying Implicit Affective States in Text
Abstract: Recognizing affective states is essential for narrative text understanding and for AI applications such as conversational dialogue, summarization, and sarcasm recognition. Many natural language processing (NLP) tools have been developed to recognize explicit expressions of sentiment in text. But humans frequently infer affective states from events and personal situations, so NLP systems also must be able to recognize implicit indicators of affect. This talk will focus on "affective events", which are experiences that implicitly suggest a positive or negative affective state for the experiencer. For example, buying a home is usually desirable and associated with a positive state, but losing your job is typically undesirable and associated with a negative state. We will outline the many roles that affective events play in natural language understanding and overview computational models designed to improve the recognition of implicit affective states in NLP systems.