Dialog System for Open-Ended Conversation Using Web Documents

We have developed a new type of open-ended dialog system that generates proper responses to users' utterances using the abundant documents available on the World Wide Web as sources. Existing knowledge-based dialog systems provide meaningful information to users, but they are unsuitable for open-ended input. The system Eliza, while it can handle open-ended input, gives no meaningful information. Our system lies between the above two dialog systems; it can converse on various topics and gives meaningful information related to the user's utterances. The system selects an appropriate sentence as a response from documents gathered through the Web on the basis of surface cohesion and shallow semantic coherence. We developed a trial system to converse about movies and experimentally found that the proposed method generated appropriate responses at a rate of 66%.