Effect of Normalization Techniques in VIKOR Approach for Mining Product Aspects in Customer Reviews

Product aspect ranking becomes an important field of research as the tremendous number of aspects discussed on the retail Websites disallows the probable customers to focus on specific product aspects to compare among the presented products. Frequency-based, opinion-based, and aspect relevancy are common criteria for ranking the extracted product aspects from customer reviews. The multi criteria nature of the aforementioned problem makes Multi-Criteria Decision Making (MCDM) approach provide promising solution of product aspect ranking. However, one of the most important problems of MCDM is the ranking abnormality. Thus, the focus of this research is to analyze the performance of VIKOR approach with various normalization techniques in addressing the product aspect ranking problem. The experimental results on different product reviews demonstrate that Vector normalization approach is more efficient in prioritizing important product aspects using VIKOR approach.