Graph based technique for hindi text summarization

Abstract

Automatic Summarization is the process of generating or extracting the important sentences from the given input document. Since there are many such systems for English language so this proposed system is mainly focused on the Hindi language. The basic idea of this summarization system is to identify the important sentences and also to extract them based on its relevance with other sentences. In case of summarization the sentences in the summarized document should be meaningful and relevant to each other, which are achieved using sentential semantic analysis. For finding the relation between each sentence and also to analyze for the importance, the Graph based approach is found to be more appropriate. Based on the frequency of words occurrence in the input document, the sentences are ranked and the ranks are used to identify the important sentences in the document. The relevance between each sentence in the document with other sentences is found using semantic similarity. There may be same information conveyed by two different sentences whose semantic similarity score is very high. Such kind of sentences has to be kept only once in the output. For which an analysis has been performed over various semantically similar sentences. Finally, the identified relevant sentences are merged using the rank and the semantic analysis of the sentences. These identified sentences are rearranged to provide a proper meaningful summarized text to avoid textual continuity in the output text. The system is found to perform well in terms of precision, recall and F-measure with various input documents.