Protein Name Tagging for Biomedical Annotation in Text

摘要

We explore the use of morphological analysisrnas preprocessing for protein namerntagging. Our method finds protein namesrnby chunking based on a morpheme, thernsmallest unit determined by the morphologicalrnanalysis. This helps to recognizernthe exact boundaries of protein names.rnMoreover, our morphological analyzerrncan deal with compounds. This offersrna simple way to adapt name descriptionsrnfrom biomedical resources for languagernprocessing. Using GENIA corpus 3.01,rnour method attains f-score of 70 points forrnprotein molecule names, and 75 points forrnprotein names including molecules, familiesrnand domains.