AUTOMATIC CONSTRUCTING ONTOLOGY BASED PETROCHEMICAL PLANT INSTALLATIONS FROM ACCIDENT REPORTS
This study proposed a method for automatic construction of petrochemical plant installation ontology from accident reports of PECSAFER. This method generated the rule based on Japanese grammar and description, enable automated construction of ontology. The ontology is systematical summary of concept and vocabulary. It is useful in mitigating of knowledge gap and management of knowledge system. This ontology includes three relations, Hyponymy-relation of plant equipment, Meronymy -relation, and Abnormal-event relation. This method, the semantic information in sentences is extracted with the rule base by using the Natural Language Processing (NLP). The first rule is generated from Adnominal-expression in Japanese. Second is generated from description rule in PECSAFER accident reports. Third is generated from frequent appearance pattern of Accident-term. We extracted relations by using these rules and constructed ontology. Comparisons between ontology constructed in this study and artificial ones. Complete-recall-ratio reach 52.54％, Partial-recall-ratio reach 57.62％, Moreover, Precision-ratio reach 57.85％.
natural language processing ontology petrochemical plant
Yuichiro Furuta Hirotsugu Minowa Yoshiomi Munesawa Kazuhiko Suzuki
Division of Mechanical and System Engineering Department of Intelligent Mechanical Systems, Okayama University, Okayama 700-0082, Japan
国际会议
The Tneth International Conference on Industrial Management(第十届工业管理国际会议 ICIM 2010)
北京
英文
578-582
2010-09-16（万方平台首次上网日期，不代表论文的发表时间）