Сентимент-анализ читательского комментария: автоматизированная vs ручная обработка текста
Sentiment analysis of reader comments: Automated vs manual text processing

Summary/Abstract: The verbal and structural features of the reader comment, a genre of Internet communication, were studied. The method of sentiment analysis (ParallelDots API) was used to reveal and measure the emotive component of the reader comments (N = 3000) in the English and Russian languages. The results obtained were verified by the manual linguistic text analysis. The experts were specialists in the field of philology of the English and Russian languages (N = 6), students of philology, as well as native speakers of the Russian language for whom English is a foreign language, i.e., their level of proficiency is C1 (N = 4). As a result of the comparison of the data collected through the automated and manual text processing, a number of factors that reduce the reliability of the results of automated sentiment analysis of the reader comments were singled out. Difficulties hindering the objective determination of the sentiment by the program were found in the reader comments in both analyzed languages. This is indicative of the structural similarities between the English and Russian reader comments at the lexical and syntactic levels. The feasibility of the mixed automated and manual text processing in order to obtain more detailed and objective data was demonstrated. The results of this work can be used for comparative studies of two or more languages performed by the method of sentiment analysis, as well as for drawing parallels between the lexical, grammatical, and cultural components of languages.

• Details
• Contents