Presentation 2022-09-13
A Proposal of an Ensemble Learning Method by Using Pre-Trained BERT Models for Authorship Attribution
Taisei Kanda, Liu Yejia, Jin Mingzhe,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Although bidirectional encoder representations from transformers (BERT) is a highly versatile model, there have been reports indicating that pretrained data affect its performance. Developing a BERT model pretrained with data from a wide variety of domains is ideal, but it is cost-prohibitive for individuals. Therefore, in this study, we propose an ensemble learning method using multiple BERTs that are constructed from different pretraining data, and we report the experimental results.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Natural language processing / BERT / Authorship attribution / Ensemble learning
Paper # NLC2022-6
Date of Issue 2022-09-06 (NLC)

Conference Information
Committee NLC
Conference Date 2022/9/13(1days)
Place (in Japanese) (See Japanese page)
Place (in English) Keio Univ. Yagami Campus.
Topics (in Japanese) (See Japanese page)
Topics (in English) The 19th Text Analytics Symposium
Chair Mitsuo Yoshida(Univ. of Tsukuba)
Vice Chair Hiroki Sakaji(Univ. of Tokyo) / Takeshi Kobayakawa(NHK)
Secretary Hiroki Sakaji(NTT) / Takeshi Kobayakawa(Hiroshima Univ. of Economics)
Assistant Kanjin Takahashi(Sansan) / Yasuhiro Ogawa(Nagoya Univ.)

Paper Information
Registration To Technical Committee on Natural Language Understanding and Models of Communication
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Proposal of an Ensemble Learning Method by Using Pre-Trained BERT Models for Authorship Attribution
Sub Title (in English)
Keyword(1) Natural language processing
Keyword(2) BERT
Keyword(3) Authorship attribution
Keyword(4) Ensemble learning
1st Author's Name Taisei Kanda
1st Author's Affiliation Doshisha University(Doshisha Univ.)
2nd Author's Name Liu Yejia
2nd Author's Affiliation Doshisha University(Doshisha Univ.)
3rd Author's Name Jin Mingzhe
3rd Author's Affiliation Doshisha University(Doshisha Univ.)
Date 2022-09-13
Paper # NLC2022-6
Volume (vol) vol.122
Number (no) NLC-180
Page pp.pp.9-12(NLC),
#Pages 4
Date of Issue 2022-09-06 (NLC)