Machine Learning of Ambiguous Sentences and Analysis of Relation between Ambiguous Sentences and Diagrams in Technical Manual for Tacit Knowledge Acquisition

  • Naoto Kai Kyushu university
  • Kota Sakasegawa Kyushu university
  • Tsunenori Mine Kyushu University
  • Sachio Hirokawa Kyushu University
Keywords: ambiguous sentences, machine learning, technical manual, correlation, diagrams

Abstract

Many companies create manuals to inherit the technology. Manuals contain a lot of information, but few are focused on tacit knowledge. Against this background, this research has two purposes, targeting manuals. First, it is to judge ambiguous sentences for mining tacit knowledge. Second, it is to clarify relation between ambiguous sentences and diagrams for mining tacit knowledge. This study was conducted with maintenance of railway rolling stock as the subject. At first, to confirm the hypothesis that ambiguous sentences include tacit knowledge, we compared result of mining by human and by machine learning. We obtained the results that we can recognize the ambiguous sentences by machine learning with high accuracy. The most striking observation to emerge from this data analysis was that ambiguous sentences can identified not only by adjectives and adverbs but also nouns, post positional particle, or conjunctions. Secondly, since many diagrams are used in the manual, it is thought that there were clues to mining tacit knowledge in the diagrams. As a result, it was found that there is a correlation between ambiguous sentences and diagrams.

References

Jacob Devlin, Ming-Wei Chang, Kenton Lee, Kristina Toutanova, BERT: Pre-training of deep bidirectional transformers for language understanding, arXiv preprint arXiv:1810.04805, 2018

Etsuko Harada, Yuki Endo, Are video manuals better than paper manuals: Testing video superiority with constitution tasks(in Japanese), Proc. Ann. Conf. Japan Academy of Nursing Administration and Policies, p.15, 2017

JRMA, Bogie/Wheelset - Maintenance Points (in Japanese), Japan Railway Rolling Stock & Machinery Association, 2019

Yohei Kikuta, BERT Pretrained model Trained On Japanese Wikipedia Articles, GitHub repository, https://github.com/yoheikikuta/bertjapanese, 2019

Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, Jeff Dean, Distributed representations of words and phrases and their compositionality, In Advances in Neural Information Processing Systems, pp. 3111-3119, 2013

Masaki Murata, Masato Tokuhisa, Qing Ma, Manual Analysis of Redundant Sentences (in Japanese), Proc. Ann. Meeting of Association for Natural Language Processing, pp.984-987, 2015

Takuya Murayama, Toru Sadakata, Masayuki Ihara, A Tacit Knowledge Extraction Evaluation Method based on the Availability of Manuals (in Japanese), Japanese Society for Artificial Intelligence Technical Report SIG-KST 33(4), pp.1-6, 2018

Tsukasa Negishi, Akifumi Tsuruta, Jie Li, Min Xian, Toshiyuki Murao, Hiroshi Hashimoto, Consideration of Difference of Technical Transfer between Japan and China (in Japanese), Bulletin of Advanced Institute of Industrial Technology, No.7, pp.53 - 58, 2013

Ikujiro Nonaka, Hirotaka Takeuch, The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation, Oxford Univ Press, 1995

Kijun Oda, A Psychological Study on the Japanese Qualitative and Quantitative Words (in Japanese), Japanese Journal of Educational Psychology, 18(3), pp.38-48, 1970

Michael Polanyi, The Tacit Dimension, University of Chicago Press, 1966

Chihomi Sannami, Shouhei Iba, Shin-ichi Nakayaa, Current status and issues of the manuals of mobile phones for elderly viewed from components of the manual components of the manual components of the manual components of the manual –with respect to the correspondence to the issues that are expected from the physical and mental characteristics of the elderl characteristics of the elderl characteristics of the elderl characteristics of the elderl (in Japanese), Journal of Japanese Society for Information and Media Studies, 12(1), pp.14-27, 2013

Masatoshi Suzuki, w2v pretrained model, http://www.cl.ecei.tohoku.ac.jp/ m-suzuki/jawikivector/

Hitomi Terajima, Tacit Knowledge of Nurse Manager Working to Improve Nursing Services (in Japanese), The Journal of the Japan Academy of Nursing Administration and Policies, 13(1), pp.67-75, 2009

Shunsuke Totou, Masaki Murata, Masato Tokuhisa, Qing Ma, Mechanical Analysis and Detection of Redundant Sentences (in Japanese) Proc. Ann. Meeting of Association for Natural Language Processing, pp.11141117, 2013

Hiroaki Yamada, Yasuaki Fukuta, Shu Hayakawa, Aiming at Externalization of Tacit Knowledge in Knowledge Construction (in Japanese), Journal of Japan Society for Production Management, 13(2), pp.9-19, 2007

K. Iwatsuki, The role of graphic aids in comprehension of an expository prose (in Japanese), Journal of educational psychology, 1998, 46, pp.142- 152.

Published
2021-06-30
Section
Technical Papers