Is Sentiment Analysis Domain-Dependent

The purpose of this research carried out within applied linguistics is to consider the dependency of the sentiment lexicon and other sentiment analysis tools on the domain under study. For the experiment, we used the REGEX algorithm including the sentiment lexicon and formal grammar rules applied with the certain priorities. These rules and the corresponding syntactic models are similar to regular expressions which detect certain text elements, simplify each sentence and present the text as a formal model. The reviews in Russian from three domains (Bank Service Quality, Hotel Service Quality and Sightseeing) are analyzed; F1 measure is used as the efficiency criterion. The experiment does not reveal the domain-dependency of the algorithm applied. It is determined that the system generally detects positive reviews better than negative ones. When negative opinions are expressed, there is a tendency to use non-standard vocabulary and syntax.