期刊名称：International Journal of Computer Science Issues
印刷版ISSN：1694-0784
电子版ISSN：1694-0814
出版年度：2014
卷号：11
期号：6
出版社：IJCSI Press
摘要：Abstractive summarization is the process of creating a condensed version of the given text document by collating only the important information in it. It also involves structuring the information into sentences that are simple and easy to understand. This paper presents the process that generates an abstractive summary by focusing on a unified model with attribute based Information Extraction (IE) rules and class based templates. Term Frequency/Inverse Document Frequency (TF/IDF) rules are used for classification of the document into several categories. Lexical analysis reduces prolixity, resulting in robust IE rules. Usage of templates for sentence generation makes the summaries information intensive. The IE rules are designed to accommodate the complexities of some Indian languages. This paper analyzes the adaptation of the system methodology over multiple Indian languages and several document categories. It also draws comparison between abstracts generated and summaries obtained by extractive methods.
关键词：Template based Generation; Language Parsing and Understanding; Information Extraction; Template Selection; Abstractive and Extractive Text Summarizations
