Incorporating ontological background knowledge into Information Extraction
Doctoral Consortium, co-located at International Semantic Web Conference 2009, ISWC, 10/2009.
In my PhD work I apply formal domain knowledge to support domain specific Information Extraction tasks. My main research goal is revealing strategies incorporating domain ontologies for: (i) Interchanging domain ontologies by letting systems adapt on new domains without any additional engineering effort. (ii) Allowing extraction templates to be specified in the ontology’s vocabulary with the canonical query language SPARQL. (iii) Improving Ontology-based Information Extraction approaches by making extraction pipelines access existing knowledge in the earliest possible stages. (iv) Returning potential query results in RDF (graphs of facts and instances) as scenarios weighted with textual and ontological evidences. (v) Improving methods using instance knowledge to train statistical models for extraction. In summary, my PhD thesis’ contribution is a system letting users load up ontologies about their domains of interest, query domain relevant text with SPARQL and get results as weighted RDF graphs.