Automatic Utterance Segmentation in Instant Messaging Dialogue

Instant Messaging (IM) chat sessions are real-time, text-based conversations which can be analyzed using dialogue-act models. Dialogue acts represent the semantic information of an utterance, however, messages must be segmented into utterances before classification can take place. We describe and compare two statistical methods for automatic utterance segmentation and dialogue-act classification in task-based IM dialogue. It is shown that IM messages can be automatically segmented and classified to a very high accuracy using statistical machine learning.