Semantic Annotation in Specific Domains with rich Ontologies (in French)

Yue MaYue Ma, Adelina Nazarenko François LévyAdelina Nazarenko François Lévy

Yue Ma, Adelina Nazarenko François Lévy
Semantic Annotation in Specific Domains with rich Ontologies (in French)
20ème conférence du Traitement Automatique du Langage Naturel, to appear

Details
Bibtex

KurzfassungAbstract
Technical documentations are generally difficult to explore and maintain. Powerful tools could help users, provided the documents have been semantically annotated. However, the annotations must be sufficiently specialized, rich and consistent, relying on some explicit semantic model -- usually an ontology -- that represents the semantics of the target domain. In this paper, we analyze that traditional approaches have limited success for this task. Hence, we propose a novel approach, named phrase-based statistical semantic annotation, for predicting semantic annotations from limited training data. Such a modeling makes the challenging problem, domain specific semantic annotation regarding arbitrarily rich semantic models, easily handled. We used several evaluation metrics on two different business regulatory texts, for which our approach achieved a good performance. In particular, it obtained $91.9%sim97.65%$ F-measure in the label and position predictions with different settings. This suggests that human annotators can be highly supported in domain specific semantic annotation tasks.
Bemerkung: Note: To appear.
Forschungsgruppe:Research Group: AutomatentheorieAutomata Theory

@inproceedings{ MaLeNa-TALN13,
  address = {Dresden, Germany},
  author = {Yue {Ma} and Adelina Nazarenko {Fran\c{c}ois L\'evy}},
  booktitle = {20\`eme conf\'erence du Traitement Automatique du Langage Naturel},
  note = {To appear.},
  title = {Semantic Annotation in Specific Domains with rich Ontologies (in French)},
  year = {2013},
}