An Efficient Approach for Automatic Text Summarization of Hindi Text

Abstract

Due to the rise of the Internet, the number of digital documents has increased newlinesignificantly. This has led to the need for a text summarization system that can newlineautomatically produce a summary of the documents. The text summarization system newlinealso eliminates the need for manual work and time spent compiling the information. newlineBasically, a summary contains key phrases and other relevant text material without newlinealtering the source document s general context or key information. The process of newlinetext summarization started in 1958 and is still being studied by researchers. Text newlinesummarization systems are classified as extractive or abstractive. The extractive text newlinesummarization process aims to extract the most appropriate sentences and phrases newlinefrom documents. It then compiles these into a summary. In contrast, abstractive newlinesystems produce a summary of the documents to describe words other than the ones newlinecontained in an input document. There has been a significant amount of research on newlinedocuments written in English but relatively little on documents written in Hindi. newlineThis research aims for the Hindi language documents as Hindi is India s most newlinespoken language. From the literature, we have not found any efficient summarizing newlinesystem for Hindi language documents. newlineIn general, to produce a summary, first, extract the features from the newlinedocument s text, then compute the score of each document s sentence based on newlinefeature value, and then select the sentences with the highest score. The number of newlinesentences that can be included in a summary depends on the user-defined newlinecompression ratio. This research investigated several methods used in the literature newlineto rank each sentence of Hindi or English text documents. These are graph-based, newlinedeep learning-based, sematic-based, statistical-based, fuzzy-based, and machine newlinelearning-based methods. newline