A new approach for information retrieval in semantic web mining involving weighted relationship

To extract the relevant information from the web is the primary focus of semantic web mining. With the increasing volume of data in web the real challenge is to extract the required information. The basic input for semantic web mining is the user input but the accuracy of the data extraction is based on the domain classification. To achieve the same we have proposed a new approach where a “T” based Semantic structure is maintained for each training sentence where the relationship of each word in the training sentence is established in the form of cosine similarity weight and also link towards the possible terms of the same words are established with weights. Cosine similarity involved here not only calculates the similarity weight between the words of training sentence but also to establish the semantic relationship between sentences of the same group. This paper explains in detail regarding how the training sentences are grouped and the relationship are established between them using a new weight relationship algorithm.