Presentation 2020-11-13
[Poster Presentation] Automation of Ontology Generation by Pre-trained Language Model
Atusi Oba, Ayato Kuwana, Paik Incheon,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) As an initial attempt of ontology generation with neural network, Recurrent Neural Network (RNN) based method is proposed recently. This method efficiently achieves the ontology generation task by combining sequential neural network and pretrained vector representation of words. However, updating the architecture is possible due to the rapid development of Natural Language Processing (NLP). In other hand, transfer learning of language models trained by huge unlabeled corpus such as BERT has brought a breakthrough in NLP. These methods allowed the language model to have a wide language knowledge and its processing ability by two-stage learning which divide training procedure into two stages, Pre-training and Fine-tuning. Inspired by these achievements, to apply transfer learning of language model, we propose novel workflow for ontology generation consists of two-stage learning. This paper also provides a quantitative comparison between the proposed method and existing methods. Empirical evidence showed that our best configuration update previous accuracy over 12.5% without sophisticated structure.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) ontology / neural network / pre-trained model / natural language processing
Paper # KBSE2020-22,SC2020-26
Date of Issue 2020-11-06 (KBSE, SC)

Conference Information
Committee KBSE / SC
Conference Date 2020/11/13(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Virtual / Kikai-Shinko-Kaikan Bldg.
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Hiroyuki Nakagawa(Osaka Univ.) / Shinji Kikuchi(NIMS)
Vice Chair Takuya Saruwatari(NTT Data) / Yoji Yamato(NTT) / Kosaku Kimura(Fujitsu Lab.)
Secretary Takuya Saruwatari(OKI) / Yoji Yamato(NII) / Kosaku Kimura(Kobe Univ.)
Assistant Shinpei Ogata(Shinshu Univ.) / Erina Nakihara(Doshisha Univ,)

Paper Information
Registration To Technical Committee on Knowledge-Based Software Engineering / Technical Committee on Service Computing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) [Poster Presentation] Automation of Ontology Generation by Pre-trained Language Model
Sub Title (in English)
Keyword(1) ontology
Keyword(2) neural network
Keyword(3) pre-trained model
Keyword(4) natural language processing
1st Author's Name Atusi Oba
1st Author's Affiliation the University of Aizu Graduate School(UoA)
2nd Author's Name Ayato Kuwana
2nd Author's Affiliation the University of Aizu Graduate School(UoA)
3rd Author's Name Paik Incheon
3rd Author's Affiliation the University of Aizu(UoA)
Date 2020-11-13
Paper # KBSE2020-22,SC2020-26
Volume (vol) vol.120
Number (no) KBSE-231,SC-232
Page pp.pp.40-40(KBSE), pp.40-40(SC),
#Pages 1
Date of Issue 2020-11-06 (KBSE, SC)