Automatic Comparison of Products based on Opinion Features using Synonym and Jaccard Similarity

Abstract

The opinion features can be used to improve the quality of product recommendations. There are some previous researches that try to compare products using opinion features. They compare products using the weight of the most important features. However, that kind of approach is less effective because does not consider other features that has similar meaning in the weight counting. In this paper, we proposed new strategy that interpolates important features using synonym and Jaccard similarity for automatic product comparison. There are two main processes: get the synonym of the extracted features and count the similarity of product. The result from the experiment shows that this new strategy can get the accuracy up to 90%.