Improvement of Syntactic Dependency Analysis by Syntax Pattern Acquisition and Coordinate Structure Analysis

2016 Fiscal Year Final Research Report

• Project/Area Number: 26240035
• Research Category: Grant-in-Aid for Scientific Research (A)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Nara Institute of Science and Technology
• Principal Investigator: Matsumoto Yuji 奈良先端科学技術大学院大学, 情報科学研究科, 教授 (10211575)
• Co-Investigator(Kenkyū-buntansha): 新保 仁 奈良先端科学技術大学院大学, 情報科学研究科, 准教授 (90311589) ; Duh Kevin 奈良先端科学技術大学院大学, 情報科学研究科, 助教 (80637322) ; 進藤 裕之 奈良先端科学技術大学院大学, 情報科学研究科, 助教 (20734784)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 自然言語処理 / 統語解析 / 並列構造 / 複文構造 / 依存構造解析 / 機械学習 / アノテーション / コーパス

Outline of Final Research Achievements

Aiming at an improvement of syntactic dependency analysis of English sentences, we compiled an English multi-word expression (MWE) lexicon and an MWE-annotated corpus. We also developed an algorithm for coordination structure analysis on top of dependency structure analysis.
For English MWEs, we collected English functional multi-word expressions and performed MWE annotation on Penn Treebank. We also implemented an MWE-aware English dependency parsing algorithm. We collected English complex sentence patterns and made annotation on made annotation on complex sentence examples.
For coordinate structure analysis, we proposed a joint algorithm of dependency and coordinate structure analysis and implemented a prototype system.

Free Research Field

自然言語処理