@techreport{oai:ipsj.ixsq.nii.ac.jp:00080106, author = {服部, 圭悟 and 梅基, 宏 and 増市, 博 and Keigo, Hattori and Hiroshi, Umemoto and Hiroshi, Masuichi}, issue = {8}, month = {Jan}, note = {本稿では,LFG解析を利用した複合語の構造解析手法を提案する.Wikipediaの記事のLFG解析結果を用いて複合語の語基間の述語項構造を明らかにし,複合語の言い換え規則を生成する.また,生成した規則を用いて複合語と複合語の言い換え表現を含む自然文でRTEを行い,複合語の言い換え規則を自然文に適用する上での課題を明らかにする.実験の結果,最大でF値77.97%,適合率87.34%,再現率70.41%を達成した., We propose a method of semantic analysis for compound words based on LFG (Lexical Functional Grammar). We analyze all Wikipedia articles using LFG and determine the grammatical and semantic structures. Then we extract predicate argument structures of compound words. Finally semantic relations of compound words are determined. We conduct experiments to recognize the entailment of Japanese text including compound words using semantic relations and clarify the problems when applying the proposed method. Experimental results showed that we achieved up to 77.97% of f-score, 87.34% of precision, and 70.41% of recall.}, title = {LFG解析を利用した日本語RTEにおける複合語の対応}, year = {2012} }