An improved Automatic Text Summarization System for Punjabi Text using Modified Features Selection

Abstract

There is a lot of online Punjabi information available in today’s world. Internet provides more information in response to a query than is required; and it is time consuming to manually extract the relevant information from large documents, reports, papers and articles which are stored in digital form, such as digital newspaper systems. In order to solve this problem, an Automatic Text Summarization (ATS) is necessary that provides the most important information from the source document in the form of summary by preserving its overall meaning and content. The system proposed in this thesis is used to summarize the text document written in Punjabi language by retaining its relevant sentences using extractive summarization technique. The importance of proper score for features used in summary generation for the text document written in Punjabi (Gurumukhi script) is also discussed. Features selection plays a fundamental role in automatic text summarization. The effectiveness of the proposed system is presented in form of significant gains attained in terms of recall, precision and F-score in comparison with the existing system which uses intrinsic measures.