Semi-supervised Algorithm for Human-Computer Dialogue Mining
Calkin Suero Montero, Kenji Araki · 2007
This paper describes the analysis of weak local coherence utterances during human-computer conversation through the appli-cation of an emergent data mining tech-nique, data crystallization. Results reveal that by adding utterances with weak local relevance the performance of a baseline conversational partner, in terms of user satisfaction, showed betterment. 1