Evaluation of Dialogue Systems with Different End of Sentences Using Seq2seq Model

Abstract

The purpose of this study is to construct a more natural relationship between dialogue system and human by using the politeness theory, which is known as a consideration for maintaining smooth human relations by mutual speech and behavior. Concretely, based on the politeness theory, from the chatting data between human beings, the words which shorten the psychological distance by casual talk and communication and the words which maintain the psychological distance by honorific language were extracted. The output was changed so that the Seq2seq model would preferentially use the extracted two kinds of words, after the learning of the appropriate reply for the utterance of the partner was carried out using the dialogue data by the Sequence to Sequence model suitable for the learning of the paired data. On the dialogue system which adjusted learning, the evaluation experiment was carried out whether the appropriate response was possible.