Call For Paper Manager: Data Mapping and First Step to a CFP Ontology

カテゴリ: 研究会(論文単位)

論文No: IS14049

グループ名: 【C】電子・情報・システム部門 情報システム研究会

発行日: 2014/11/13

タイトル(英語): Call For Paper Manager: Data Mapping and First Step to a CFP Ontology

著者名: Issertial Laurent(大阪府立大学),辻 洋(大阪府立大学)

著者名(英語): Issertial Laurent(Osaka Prefecture University),Tsuji Hiroshi(Osaka Prefecture University)

キーワード: text mining|ontology|data management|call for paper

要約(日本語): This paper introduces the addition of external ontology and creation of internal one to our Call For Paper Manager (extracting relevant data from CFP via text mining) in order to improve the quality of extracted data and lead to CFP Ontology. First, we take previously extracted locations and link them to DBPedia and GeoNames to collect additional information and disambiguate unknown type of location. Secondly, we extract topics from CFP and create a topic ontology based on CFP internal hierarchy to finally link it to external ontology namely DBPedia and the ACM Computing Classification System. Running experiements on a 300 CFP dataset, we find a location matching average of 85% on DBPedia with 60% unknown locations matched. For the topics, on the same dataset we obtain for extraction 0.89 for recall and 0.95 for precision with 8169 topics extracted, leading to the creation of a 5000 nodes for 7080 relations ontology.

要約(英語): This paper introduces the addition of external ontology and creation of internal one to our Call For Paper Manager (extracting relevant data from CFP via text mining) in order to improve the quality of extracted data and lead to CFP Ontology. First, we take previously extracted locations and link them to DBPedia and GeoNames to collect additional information and disambiguate unknown type of location. Secondly, we extract topics from CFP and create a topic ontology based on CFP internal hierarchy to finally link it to external ontology namely DBPedia and the ACM Computing Classification System. Running experiements on a 300 CFP dataset, we find a location matching average of 85% on DBPedia with 60% unknown locations matched. For the topics, on the same dataset we obtain for extraction 0.89 for recall and 0.95 for precision with 8169 topics extracted, leading to the creation of a 5000 nodes for 7080 relations ontology.

原稿種別: 日本語

PDFファイルサイズ: 653 Kバイト