Vertaisarvioitu artikkeli konferenssijulkaisussa (A4)
TurkuNLP Entry for Interactive Bio-ID Assignment
Julkaisun tekijät: Suwisa Kaewphan, Farrokh Mehryary, Kai Hakala, Tapio Salakoski, Filip Ginter
Toimittaja: Cecilia Arighi, Qinghua Wang, Cathy Wu
Konferenssin vakiintunut nimi: BioCreative
Julkaisuvuosi: 2018
Kirjan nimi *: Proceedings of the BioCreative VI Workshop
Aloitussivu: 32
Lopetussivun numero: 35
ISBN: 978-84-948397-0-2
Verkko-osoite: http://www.biocreative.org/resources/publications/bcvi-proceedings/
Rinnakkaistallenteen osoite: https://research.utu.fi/converis/portal/detail/Publication/28249832
We participate in BioCreative VI: Interactive Bio-ID Assignment (Bio-ID) track by developing systems capable of named entity recognition and normalization of 6 entity types, namely Protein, Cell, Organism, Tissue, Molecule and Cellular. Our named entity recognition system is based on conditional random fields. For named entity normalization, we apply fuzzy matching and rule-based system to disambiguate and assign unique identifiers to the entities. The official evaluation shows that average F1-scores of all entity types for our recognition and normalization systems on strict span offsets are 0.720 and 0.668, respectively.
Ladattava julkaisu This is an electronic reprint of the original article. |