期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：22
页码：3330-3343
出版社：Journal of Theoretical and Applied
摘要：Modern sentiment analysis models rely on a sentiment lexicon, which is the most essential feature that drives their performance. This resource is indispensable for, and greatly contributes to, sentiment analysis tasks. This is evident in the emergence of a large volume of research devoted to the development of automated sentiment lexicon generation models. The task of tagging sentiment-bearing words with a positive or negative connotation, and sometimes with a strength, comprises of two core approaches: the dictionary-based approach and the corpus-based approach. The former involves making use of a digital dictionary to tag words, while the latter relies on co-occurrence statistics or syntactic patterns embedded in text corpora. The end result is a linguistic resource comprising a priori information about words, across the semantic dimension of sentiment. This paper contributes to the existing literature by providing a survey on the most prominent research works that have employed lexical resources, dictionaries and thesauri for sentiment lexicon generation. We also conduct a comparative analysis on the performance of state-of-the-art models proposed for this task, and shed light on the current progress and challenges in this area.
