Frequent Term Based Clustering of Stories with Semantic Analysis for Searching and Retrieval

Effective document organizations are often those which provide a concise representation of text content in a large collection ofdocuments. We have considered the task of clustering of stories (documents) as a facilitation of effectual document arrangement for searchingand retrieval. We propose a novel representation for a story, based on the essential parts of speech - the nouns, verbs and adjectives. We thenperform a clustering of these story representations, resulting in a graph structure where the story representations are conjoined at nodes havingthe same or synonymous noun. Such a structure can be queried for stories by giving a search string. We employ the use of a knowledge bankthroughout the system as a step to realize semantic analysis of the text. For testing the goodness of cluster, we carry out the classification test, ontwo data-sets. We are able to achieve significantly high quality of clustering, with promising results in regard to memory compaction