Conventional dialogue generating methods focus on generating fluent sentences in context, but insufficient consideration of speaker emotions. They often generate sentences with the same sentimental polarity of the speakers. Such that sentences are hard to change the mood or viewpoints of the speakers, i.e., from a negative mood to positive one, etc. In this paper, we propose a method to generate dialogue sentences to provide different viewpoints. To this end, we propose two novel concepts, polarity co-occurrence (P-Cooc) and modification co-occurrence (M-Cooc). P-Cooc is used to find aspects providing different and comfortable viewpoints, and M-Cooc is used to find proper modification terms to such that aspects. The experimental results demonstrate that our method could provide supplementary viewpoints to promote critical thinking.
CITATION STYLE
Yoshida, S., & Ma, Q. (2020). Generating dialogue sentences to promote critical thinking. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 12391 LNCS, pp. 354–368). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-59003-1_23
Mendeley helps you to discover research relevant for your work.