A Lexicon-Based Sentiment Analysis for Amazon Web Review

摘要

The development of internet more quickly over time as well as the development of e-commerce one of them is amazon.com. The amazon.com is one of the largest e-commerce in the world by providing various needs that can be accessed by internet. Review feature provided to make amazon.com party can know the various responses from consumers. However amazon.com difficulties in summarizing the various kinds of reviews that are positive or negative. By using one natural language processing the main purpose is to help the amazon.com in knowing the most responses from consumers to improve the quality service. In this research we will be using the dataset that was obtained from UCI Machine Learning that contain 1000 set of data which has 478 of negative data and 522 positive data and this will be combine a variety of classification methods for comparison and in preprocessing process will be added with lexicon technique to improve the quality of preprocessing. The result of this research is K-Nearest Neighbor with lexicon technique get highest accuracy with value 92.67% followed by SVM with lexicon get 91.33% accuracy and last Decision tree with 82% accuracy.