Bacteria Biotope Detection, Ontology-based Normalization, and Relation Extraction using Syntactic Rules

The absence of a comprehensive database of locations where bacteria live is an important obstacle for biologists to understand and study the interactions between bacteria and their habitats. This paper reports the results to a challenge, set forth by the Bacteria Biotopes Task of the BioNLP Shared Task 2013. Two systems are explained: Sub-task 1 system for identifying habitat mentions in unstructured biomedical text and normalizing them through the OntoBiotope ontology and Sub-task 2 system for extracting localization and partof relations between bacteria and habitats. Both approaches rely on syntactic rules designed by considering the shallow linguistic analysis of the text. Sub-task 2 system also makes use of discourse-based rules. The two systems achieve promising results on the shared task test data set.