<![CDATA[The raw form of data is also an issue that creates a lot of problems while attempting to extract useful insight, thus requiring the use of NLP algorithms for text mining. This paper discusses sentiment analysis, with emphasis on user comments regarding cars on the microblog X that was formerly known as Twitter, work which employs Naïve Bayes Algorithm in text categorisation. The steps involved are the formation of the corpus and use of InsetLexicon dictionary for sentiment analysis with the help of weighted keywords and then going through pre-processing of the text data that includes cleaning, normalization and tokenization. The Naive Bayes algorithm estimates the probability of text under positive or negative sentiment class. The work shows that the “Comfortable” component of car reviews obtained the highest score in terms of recall, precision, and F1-score, which equals 0.83, 0.85, and 0.563, and the second set consists of 87 instances overall including an overall data set accuracy of 71%. The result validates the use of lexicon-based sentiment analysis in specific domain and at the same time exposes the weakness of the Naive Bayes, especially with complex word dependencies. Further studies should incorporate more advanced models and suitable dictionaries which facilitate sentiment analysis in ever-shifting online media settings.]]>
Copyrights © 2024
