Sentiment Lexicon Disambiguation: A Systematic Literature Review

Abstract

In our society, different channels are used to express one’s opinion or feelings about a situation, event, product, or service. The languages used also differ from one person or region to another. Sentiment analysis is carried out to gain a good understanding of the writer’s opinion or feelings. Still, ambiguity in languages affects the result when conducting sentiment analysis. The available sentiment lexicon is also negatively impacted due to ambiguity problems. In this research, a systematic review of sentiment lexicon disambiguation was carried out. The databases used are ScienceDirect, Scopus, Semantic Scholar, and SpringerLink. The first string search that was carried out resulted in 89,605 journals, conference papers, and book chapters. An inclusive and exclusive criterion was applied, and 61 papers were finally selected for the research. Findings show that ambiguity occurs in most available languages and negatively affects the reader’s or audience’s understanding when conducting sentiment analysis. We also look at the techniques used to carry out disambiguation and consider their strengths and weaknesses. Future research can focus on developing more sentiment lexicons free of ambiguity in low-resource languages, which will further help researchers conducting sentiment analysis.