作者: Mark Lee
DOI:
关键词:
摘要: The automatic classification of sentiment in text is becoming an im- portant area research. In this work, we present a linguistic system for sen- tence-level valence annotation. Our uses the formalism Combinatory Categorial Grammar to represent words as functions acting on their syntactic arguments, which provides unified way implementing various classes shifters. We propose two simple semi-automatic methods estimat- ing individual terms based lexical relations WordNet. evaluate data generated Affective Text task SemEval 2007 and show that it compares favourably with systems partici- pating task.