Chatting Activity Recognition in Social Occasions Using Factorial Conditional Random Fields with Iterative Classification

Abstract

Recognizing activities in social occasions plays an important role of building human social networks. For example, the recognition of social interactions could be of great help to determine whether any two attendees have the same interests in am academic conference or a cocktail party. Among the various types of social interactions, chatting with others is a significant indicator. Furthermore, the duration of a chatting activity may imply the strength of the interaction in reality. It is therefore important to recognize the patterns of chat-ting activities in social occasions. During a real-world con-versation, a person often begins talking following the other person’s utterance is completed. Linguistic experts have ob-served that chatting interaction is usually performed as an interlaced dialogic process. As a result, it is intuitive to ap-ply dynamic probabilistic models to learning and detecting