Sentiment Analysis Opinion Mining from Facebook Comments Using Data Mining Technique Sentiment analysis opinion mining from Facebook Comments Using Data Mining Technique

Abstract

Currently, customers use the internet to retrieve information, support their goods selecting decisions. The most common method base on customer reviews which come from customer experiences. However, customers usually read several reviews before making a decision to order goods. This paper focuses on the design and knowledge extraction of an automatic meaning analysis system of customer opinion by word extraction and knowledge collecting technique. Researchers use the natural language processing to find opinion meaning. In the same way, we develop the summarized method to find positive and negative reviews. This research collects the dataset from Facebook by extract positive and negative reviews, and compares the efficiency of the accuracy rate by Naive Bayes, K-Nearest Neighbors and decision tree algorithm. In following the experience, the result shows the Naive Bayes algorithm accuracy rate is 82.97%, K-Nearest Neighbors algorithm accuracy rate is 78.80% and the decision tree algorithm accuracy rate is 74.89%. From the result, we use the Naive Bayes algorithm to develop the online analysis system prototype. Then, we select Lampang ceramic fair online reviews to test the prototype. finally, the prototype system can extract the positive and negative review words in real-time functionally.