Domain Relevance based Aspect Ranking and Its Applications
This paper presents an improved approach of product aspect ranking from user generated content available online. Basically, the aspects are extracted using dependence analysis and sentiment analysis of user generated content. But, there are some precincts of the approach stated above. Often, they fetches too many aspects. As a result, it increases the complexity of ranking algorithm with respect to the time and space. At times, they also termed non-aspects as aspects of the product. The suggested methodology reduces the aforementioned problems and comprises of three steps. Firstly, using dependency analysis it mines the aspect which seems to be potentially strong. Then, it identifies the genuine aspects using domain relevance score and then, ranking of aspects is performed. The significance of the proposed work is that it lessens the running time and storage complexity of classification system. Also, the proposed system results only significant and required number of features as the concept is based on domain relevance. By varying the threshold value of domain relevancy, one can achieve only relevant aspects of particular domain. The comparison of the result of the proposed work shows that it performance is quite well. In addition, paper, also discusses the analysis and application of the proposed work.