| Autor: |
A.B. Muhammad, A.A. Dahiru |
| Jazyk: |
angličtina |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
Journal of Computer Science and Its Application; Vol 26, No 2 (2019); 1-11 |
| ISSN: |
2006-5523 |
| Popis: |
Sentiment Analysis is the computational study of opinions that are expressed in text (e.g. blogs, comments and discussion forums). Web discussion forums provide a means to express and share user views on different topics. Such content is usually rich in opinion which if mined can provide useful insight into the interaction. Although several contextual sentiment analysis techniques have been proposed and tested in the domain of product reviews, little of such techniques have been tested in a web discussion setting which fundamentally differs from reviews. This research presents the result of accounting for lexical and non-lexical contextual valence shifters to the problem of sentiment classification of web discussion posts using SentiWordNet lexicon. The approach comprises the extraction of prior polarity scores from the lexicon and modification of such scores based on context. An evaluation on three discussion forums data sets shows that SentiWordNet together with the proposed approach could be used as important resource for sentiment classification of web discussion posts.Keywords: sentiment analysis, contextual polarity, genre-adaptation |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
