Feature based semi-supervised product clustering using weighted word

Product grouping and classification is a challenging field in modern age of computing. Products are grouped from different perspective. But grouping by features is considered very important problem in computing as features describe a product. In reality, different types of products' features can be represented with same kind of words. Again same features can be described using different words or phrases. Clustering can be a useful method to group such type of products in accurate manner. Although there are various methods used in product clustering or grouping by considering features, there is considerable opportunity to improve their performance. In this paper we illustrate our proposed clustering algorithm which enhances the performance and accuracy over existing one. Typical methods for solving the feature based clustering problem are depended on semi-supervised learning using distributional similarity. As there is scope for better clustering results, we have optimized a semi-supervised learning method with predefined weights of the feature words to improve its precision.