Information extraction by an abstractive text summarization for an Indian regional language

The Internet provides many sources of different opinions, expressed through user reviews of products, blogs, and forum discussions. Systems which could automatically summarize these opinions would be immensely useful for those who wish to use this information to make decisions. The previous work in automatic summarization has completely focused on extractive summarization, in which key sentences are identified from the source text and extracted to form the output. An alternative solution is abstractive summarization in which the information from the source text is first extracted into the form of abstract data which is then post processed to infer the most important message from the original text. This work is built upon past work of extractive summarization methods to create abstractive summaries by creating new sentences in it. This paper conveys the methodology for the abstractive summarization process and its evaluation considering Telugu, a south Indian regional language, as the language of study.