作者: Nigel Collier , Mika Shigematsu , Kiyosu Taniguchi , Roberto Barrero , Ai Kawazoe
DOI:
关键词:
摘要: Amid growing public concern about the spread of infectious diseases such as avian influenza and SARS, there is an increasing need for collecting timely reliable information disease outbreaks from natural language data online news articles. In this paper we introduce BioCaster, a text mining-based system detection tracking currently being developed, discuss development domain ontology schema annotation terms. particular focus on comparison between two approaches, 1) traditional task-oriented approach with simple that does not strictly follow ontological principles, 2) formal which ontologically well-founded but adds extra requirements to schema. We report several critical problems were highlighted by entity experiment, attributable purely design. A second experiment based formally constructed produced improved results despite apparent complexity