The TIGER Corpus Navigator
Sebastian Hellmann, Jörg Unbehauen, Christian Chiarcos, Axel-Cyrille Ngonga Ngomo · DSpace repository (University of Tartu) · 2010
Linguistically annotated corpora are a cen-tral resource in NLP. The extraction of for-mal knowledge from these corpora, how-ever, is a tedious process. We introduce the Tiger Corpus Navigator, a Semantic Web system which aids users to classify and re-trieve sentences from linguistic corpora – here, the TIGER corpus – on the basis of abstract linguistic concepts. These linguistic concepts are specified ex-tensionally, thus, independent from the un-derlying annotation: The user provides a small set of pre-classified sentences that represent positive or negative examples for the corresponding concept, and the system automatically acquires a formal OWL-DL specification of the underlying concept us-ing an Active Machine Learning approach. 1