Department of Computer Engineering
S E M I N A R
Multi-Document News Summarization Using Sentence Containment
Computer Engineering Department
Due to high volume of news released in the Internet, getting an accurate and short representation of news topics is a challenge. In order to address this we present a multi-document summarization method to summarize several news articles about a topic into a short and clear summary to make it undestandable to users. Our method makes use of a document containment detection algorithm at a sentence level and tries to find most important sentences to extract from original documents and to generate the summary. It aims to increase coverage of important information from documents to be summarized and to decrease the redundancy within the generated summary as in most summarization methods in the literature. We present evaluation results in both English and Turkish data sets and show that method we develop can be a competitive summarization method.
DATE: 14 April, 2014, Monday @ 16:10