Preprint / Version 1

Metaphor Identification Based on Selectional Restrictions

##article.authors##

DOI:

https://doi.org/10.51094/jxiv.1080

Keywords:

Metaphor, Selectional Restriction, BERT

Abstract

Metaphorical expressions are frequently used in literature and daily conversations. A corpus is needed for handling metaphors in natural language processing, such as machine translation. However, it has been pointed out that large-scale data collection of metaphors is difficult due to their unfixed form. In this paper, we propose a method for automatically identifying metaphors to facilitate building a corpus. The proposed method focuses on the tendency of metaphors to violate selectional restrictions and identifies them using a pre-trained BERT model. In the proposed method, we mask the word to identify and estimate the word by BERT. The estimated word is expected not to violate selectional restrictions. By comparing the cosine similarity between the estimated and original words, we can measure the quantitative degree of violation of selectional restrictions and identify whether the word is a metaphor. We consider three variation methods of calculating cosine similarity from estimated words by BERT. Experimental results show that the proposed method is effective, with an F value of approximately 0.881 in the best case.

Conflicts of Interest Disclosure

There are no conflicts of interest (COI) to disclose regarding this research.

Downloads *Displays the aggregated results up to the previous day.

Download data is not yet available.

References

伊藤薫, "比喩表現コーパスの構築と問題点 -言語学の立場から-", 言語処理学会年次大会, 2014.

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova, "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding", arXiv preprint arXiv:1810.04805, 2018.

真鍋陽俊, 岡照晃, 海川祥毅, 高岡一馬, 内田佳孝, 浅原正幸, "複数粒度の分割結果に基づく日本語単語分散表現", 言語処理学会年次大会, 2019.

高橋水晶, 坂野遼平, "選択制限規則に基づくメタファーの識別", データ工学と情報マネジメントに関するフォーラム, 2023.

最上英明, "メタファーと認知", 香川大学一般教育研究, Vol. 46, pp. 71–80, 1994.

田畑博敏, "隠喩へのイニシエーション", 鳥取大学教養部紀要, Vol. 19, pp. 1–22, 1985.

瀬田幸人, "メタファーについて", 岡山大学大学院教育学研究科研究集録, Vol. 142, pp. 49–59, 2009.

杉本巧, "隠喩は何をなすのか", 広島大学大学院教育学研究科紀要. 第二部, 文化教育開発関連領域, Vol. 52, pp. 159–166, 2004.

徳永健伸, 寺井あすか, "比喩理解のための言語処理", 言語, Vol. 37, No. 8, pp. 46–53, 2008.

大竹孝樹, 横井祥, 井之上直也, 高橋諒, 栗原樹生, 乾健太郎, "言語モデルによる物語中のイベントの顕現性推定", 言語処理学会年次大会, 2020.

浅原正幸, 今田水穂, 保田祥, 小西光, 前川喜久雄, "Web を母集団とした超大規模コーパスの開発 : 収集と組織化", 国立国語研究所論集, Vol. 7, pp. 1–26, 2014.

Tohoku NLP Group, "Pretrained Japanese BERT models", https://github.com/cl-tohoku/bert-japanese

WorksApplications, "chiVe: Sudachi による日本語単語ベクトル", https://github.com/WorksApplications/chiVe

工藤拓, "MeCab: Yet Another Part-of-Speech and Morphological Analyzer", https://taku910.github.io/mecab/

SamuraiT, "mecab-python3", https://github.com/SamuraiT/mecab-python3

Posted


Submitted: 2025-02-07 08:13:11 UTC

Published: 2025-02-12 23:33:47 UTC
Section
Information Sciences