CONFERENCE (INTERNATIONAL) SSNN: Sentiment Shift Neural Network

Tomoki Ito (The University of Tokyo), Kota Tsubouchi, Hiroki Sakaji (The University of Tokyo), Kiyoshi Izumi (The University of Tokyo), and Tatsuo Yamashita

SIAM2020 International Conference on Data Mining (SDM20)

May 07, 2020

Deep neural networks are powerful for text sentiment analysis; however, in the real world, they cannot be used in situations where explanations are required owing to their black-box property. In response, we propose a novel neural network model called sentiment shift neu- ral network (SSNN) that can explain the process of its sentiment analysis prediction in a way that humans find natural and agreeable. The SSNN has the following three interpretable layers: the word-level original sen- timent layer, sentiment shift layer, and word-level con- textual sentiment layer. Using these layers, the SSNN can explain the process of its document-level sentiment analysis results in a human-like way. Realizing the in- terpretability of these layers is a crucial problem. To realize this interpretability, we propose a novel learn- ing strategy called joint sentiment propagation (JSP) learning. Using real textual datasets, we experimentally demonstrate that the proposed JSP learning is effective for improving the interpretability of layers in SSNN and that both the predictability and explanation ability of the SSNN are high.

* Proceedings only. Conference cancelled due to COVID-19 outbreak. To be presented at SDM21.