This repository contains codes related to the experiments in the following research paper.
Paper Title: An Experimental Evaluation of Japanese Tokenizers for Sentiment-Based Text Classification (PDF)
Authors: Andre Rusli and Makoto Shishido (Tokyo Denki University)
which was presented at the 27th Annual Meeting of the Association of Natural Language Processing (言語処理学会第27回年次大会(NLP2021))
Conference website: https://www.anlp.jp/nlp2021/index.html.
Original Japanese Rakuten sentiment dataset can be downloaded through here, then save to data/rakuten-sentiment-dataset/binary/.