Semi-automatic extraction of policy network using web search engine

World Wide Web (WWW) has different type of large collection of data, with available information for every single user query. Policy group or network is helpful to the person those are political scientist. They are used to understand the financial and social topics/Theory. Evaluation of policy groups requires hard and taking so much time. This process is manual including interviews and questionnaires. In this paper we are extracting the feature from www i.e. world wide web and then calculating the strength of relation between the pair of groups or actors after that using evaluation metrics evaluate correlation and then compare with the machine learning algorithm(KNNP). Mainly focus on the improving the correlation using the proposed algorithm. We are saying features it include webpage, web documents, web snippets, extraction of outlinks and also extracting lexical information about those actor.