The purpose of this research is to summarize customer reviews of cosmetic products that were written in Thai by using sentiment analysis technics. We expect that text summarization system will help refine the results ...
详细信息
ISBN:
(纸本)9781509044207
The purpose of this research is to summarize customer reviews of cosmetic products that were written in Thai by using sentiment analysis technics. We expect that text summarization system will help refine the results from customer review. The consumer will then be able to select the products that meet their needs while the cosmetic company will also have a chance to improve the quality of the products to meet the customers' need. We used the data from popular web sites, which has cosmetic products reviews. In this paper, the Naive Bayes algorithm is used to classify the products while we classify the data as positive or negative polarity through opinion words used. Then Cosine similarity was used to find similar phrases. They were then ranked by PageRank algorithm and hopfield Network algorithm to find the most important phrase. The performance of the program was measured with ROUGE and can be concluded that the data which were tested in the hopfieldalgorithm has the highest accuracy at 0.82040, from ROUGE-1 performance measurement.
暂无评论