UT Dialogue System at NTCIR-12 STC

This paper reports a dialogue system developed at the University of Tokyo for participation in NTCIR-12 on the short text conversation (STC) pilot task. We participated in the Japanese STC task on Twitter and built a system that selects plausible responses for an input post (tweet) from a given pool of tweets. Our system first selects a (small) set of tweets as response candidates from the pool of tweets by exploiting a kernel-based classifier. The classifier uses bagof-words in an utterance and a response (candidate) as features. We then perform re-ranking of the chosen candidates in accordance with the perplexity given by Long Short-Term Memory-based Recurrent Neural Network (lstm-rnn) to return a ranked list of plausible responses. In order to capture the diversity of domains (topics, wordings, writing styles, etc.) in chat dialogue, we train multiple lstm-rnns from subsets of utterance-response pairs that are obtained by clustering of distributed representations of the utterances, and use the lstm-rnn that is trained from the utteranceresponse cluster whose centroid is the closest to the input tweet.