Development of Writing Support Software Using Integrated Learning of Semantic and Syntactic Structures

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K00806
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 02100:Foreign language education-related
• Research Institution: Kurume National College of Technology
• Principal Investigator: Oda Mikio 久留米工業高等専門学校, 制御情報工学科, 教授 (80300648)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 言語モデル / 文法誤り訂正 / 第二言語習得

Outline of Final Research Achievements

We examined a language model that integrates syntactic and semantic structure learning models, aiming to support sentence composition by second language learners. We elucidated the model's grammatical error correction capabilities, which are closely related to syntactic and semantic learning, which are closely related to grammar. Additionally, we proposed a learning method to reconstruct synthetic training data for grammatical error correction models trained on linguistic corpora, successfully improving their grammatical error correction capabilities.
Furthermore, we developed application software that, upon inputting an English sentence, identifies grammatical errors using the proposed grammatical error correction model and simultaneously displays the concepts and synonyms of words from the Wordnet database. These achievements contribute to supporting the learning of second language learners.

Free Research Field

自然言語処理

Academic Significance and Societal Importance of the Research Achievements

これまで統語構造学習モデルと意味構造学習モデルを統合した言語モデルを検討した研究はなく，統合言語モデルを構築し，文法誤り訂正能力にどれくらい寄与するかを明らかにした．また，第二言語学習者を支援する文法誤り訂正モデルのための言語コーパスの改良法に関する本研究の手法は，文法誤り訂正の分野に限らず，教師あり学習モデルの構築に用いられる言語コーパスの改良に関する知見を与え，学習モデルの能力向上のために本手法が広く応用できると考えられる．さらに，統語構造と意味構造を表示するアプリケーションソフトウエアの試作は，今後，第二言語学習者の学習支援に寄与していくものと考えられる．