Data Mining for Navigation Generating System with Unorganized Web Resources

Users prefer to navigate subjects from organized topics in an abundance resources than to list pages retrieved from search engines. We propose a framework to cluster frequent itemsets (sets of common words) into topics, produce a hierarchical list, and then generate topics sequence from a collection of documents. The framework will regenerate a next sequence when users click a topic. Consider browsing to any topic as a kind of searching for that topic, the framework makes an inquiry using feature terms within the document representation of selected topic as query keywords. Our ranking method in searching process considers content analysis that still retaining spatial information of search keywords and link analysis of documents. Utilizing implementation of navigation generating system the experiments show that a navigation list from clustering results can be settled with regard to variance ratio of between and within distances. Agglomerative clustering is used in restructuring the extracted topics in order to produce a hierarchical navigation list.