Sentence Compressor

• Main Author: Zainal Abidin, Nur Nabilah
• Format: Final Year Project
• Language: English
• Published: Universiti Teknologi Petronas 2014
• Subjects: T Technology (General)
• Online Access: http://utpedia.utp.edu.my/13945/1/Nur_Nabilah_14321_Dissertation.pdf; http://utpedia.utp.edu.my/13945/

Nowadays, internet becomes the main source of information. Most people rely on the internet to find the information for research and assignments. People will try to find the right articles, journals, or web pages that are related to their task. In order to choose the right materials, they have to go through every articles, journals, and web pages to find the important points. However, it is very time-consuming to find go through every long articles. This information explosion as led to a constant state of information overload problem. As the solution, a desktop application named Sentence Compressor is developed to compress the long articles. This project aims to develop a desktop application that shortens the length of the long sentences without changing the original meaning. Integer Linear Programming (ILP) techniques is used to solve the sentence compression problem. Bilingual Evaluation Understudy (BLEU) is used to measure the quality of the produced output. Five articles were randomly selected for the experiment. The BLEU score for the articles compressed by Sentence Compressor and articles compressed by human is compared. The system performance evaluation is also done to measure the usefulness of this application. More than 65% of the respondents agreed that Sentence Compressor is useful in information searching.