Feature Selection and Extraction for Dogri Text Summarization

Abstract

Text summarization is defined as the process of condensing information from the source text into a shorter form without affecting the context of the information. Based on the summary generated by the summarization system, it is classified into abstractive and extractive summarization. Extractive summarization is the technique of extracting important sentences from the document that delivers the logical summary of the document. The candidate sentences for summary generation are decided by using statistical and linguistic features of the given source text. The proposed approach for Extractive Dogri Text summarization is presented in this paper. Various statistical and linguistic features that can contribute to the selection of appropriate sentences for Dogri text summarization are also illustrated in the paper. Statistical features like term frequency, length of a sentence, position of a sentence and term frequency-inverse sentence frequency (TF-ISF) are taken into consideration. And the linguistic features like presence of proper noun, numerical information, English-Dogri words are also considered for determining the candidature of the sentences for inclusion in the final summary generation.