Shintaro Sakai. Evaluating the impact of lexical gender bias on gender bias measurement with word embeddings. The New Directions in Analyzing Text as Data (TADA 2023). 2023. (Poster)
Shintaro Sakai, Yasuhiro Suzuki. Evaluating the Quality of Word Embedding Trained on Wikipedia Articles. The 9th International Conference on Computational Social Science (IC2S2 2023). 2023. (Poster)
Shintaro Sakai. Evaluating Semantic Changes in the Concept of Happiness with Diachronic Word Embeddings. The 9th International Conference on Computational Social Science (IC2S2 2023). 2023. (Poster)
Shintaro Sakai, Yasuhiro Suzuki. 日本語単語埋め込みモデルにおけるジェンダーバイアスの評価 (Evaluation of Gender Bias in Japanese Word Embedding Model). The 37th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2023). 2023.
Shintaro Sakai. Evaluation of Occupational Gender Bias in Japanese Word Embedding. The 2nd Annual Meeting of the Japanese Society for Computational Social Science. 2023.
Shintaro Sakai, Yasuhiro Suzuki. ヤフーニュースにおけるアクセスランキングとコメントランキングの時系列データの分析 (Analysis on Time Series Data of Access Rankings and Comment Rankings in Yahoo News). Joint Workshop of the Japanese Society for Artificial Intelligence. 2021.
This project aims to quantify 100 years of gender stereotypes in Japan. By utilizing the large amount of data offered by the Japanese National Diet Library, we train word embeddings for each year to measure people’s stereotypes. While word embeddings are increasingly used to measure historical stereotypes, most studies use English data. We examine whether we can replicate similar results in Japanese or uncover any limitations.