Abstract |
The sentiment analysis approach has become an essential customer analysis as a result of the fast growth of internet technology and social media. Several studies have detailed the efficacy of several sentiment classifications, ranging from lexicon-based to machine learning approaches. While lexicon-based approaches have limitations due to the limited number of terms in dictionaries and labeled data, machine learning approaches are frequently flawed because it requires massive datasets to train each domain dataset. This paper presents a framework and algorithms that bridge the gap between the lexicon and linked open data methods (DBpedia) for resolving semantic conflicts and improving SentiWordNet’s sentiment score to obtain higher performance. Furthermore, we also evaluate our sentiment classification using precision, recall, and F-measure metrics, which have values of 0.76, 0.91, and 0.82, respectively. |