Comparison of Naive Bayes and SVM Algorithm based on Sentiment Analysis Using Review Dataset

Now a day's sentiment analysis is the most used research topic. The sentiment analysis result is based on different investigation for example politics, terrorism, economy, international affairs, movies, fashion, justice, humanity. Social media are the main resource for collecting people's opinion and their sentiment about a different trending topic. People use many abusing words in social media to express their emotion. Using sentiment analysis, we will build a platform where one can easily identify the opinions are either positive or negative or neutral. This research paper will contain supervised learning which is under the machine learning approach. We run an experiment on different queries from humanity to terrorism and find out an interesting result. First of all, we have preprocessed the dataset to convert unstructured airline review into structured review form. After that, we convert structured review into a numerical value. We have to preprocess the data before using it. Stop word removal, @ removal, Hashtag removal, POS tagging, calculating sentiment score have done in preprocessing part. Then an algorithm has been applied to classify the opinion as either it is positive or negative. In this research paper, we will briefly discuss supervised machine learning. Support vector machine as well as Naïve Bayes algorithm and compares their overall accuracy, precession, recall value. The result shows that in the case of airline reviews Support vector machine gave way better result than Naïve Bayes algorithm.