Extraction of Semantic Information from an Ordinary English Dictionary and its Evaluation

摘要

The automatic extraction of semantic information, especially semantic relationships between words, from an ordinary English dictionary is described. For the extraction, the magnetic tape version of LDOCE (Longman Dictionary of Contemporary English, 1978 edition) is loaded into a relational database system. Developed extraction programs analyze a definition sentence in LDOCE with a pattern matching based algorithm. Since this algorithm is not perfect, the result of the extraction has been compared with semantic information (semantic markets) which the magnetic tape version of LDOCE contains. The resnlt of comparison is also discussed for evaluating the reliability of such an automatic extraction.