作者: Alexander Yeh , Alexander Morgan , Marc Colosimo , Lynette Hirschman
DOI: 10.1186/1471-2105-6-S1-S2
关键词:
摘要: The biological research literature is a major repository of knowledge. As the amount increases, it will get harder to find information interest on particular topic. There has been an increasing work text mining this literature, but comparing hard because lack standards for making comparisons. To address this, we worked with colleagues at Protein Design Group, CNB-CSIC, Madrid develop BioCreAtIvE (Critical Assessment Information Extraction in Biology), open common evaluation systems number tasks. We report here task 1A, which deals finding mentions genes and related entities text. "Finding mentions" basic task, can be used as building block other makes use data software provided by (US) National Center Biotechnology (NCBI). 15 teams took part 1A. A achieved scores over 80% F-measure (balanced precision recall). that tried their 1A help tasks reported mixed results. plus results are good, still somewhat lag best some domains such newswire, due complexity length gene names, compared person or organization names newswire.