Summarization using Wikipedia

Shu Gong, Youli Qu, Shengfeng Tian
2010 Text Analysis Conference  
Summarization utilizes the processing speed of computer to provide a concise and comprehensive glance at the data mass. This paper presents our work in the Text Analysis Conference 2010 guided summarization task. We make use of Wikipedia, currently the world's largest online encyclopedia, to catch the semantic meaning under words. Three features are extracted for sentence selection, based on disambiguated Wikipedia concepts and their first paragraphs in the corresponding Wikipedia articles. Our
more » ... system ranks in the middle tier of the 43 peer systems (including 2 baselines). From the analysis of evaluation results, we get the following findings: Firstly, we can find a sufficient number of Wikipedia concepts in each topic. Secondly, it is the context representation but not the number of concepts that affects system performance for a topic. Finally, highly related first paragraphs in Wikipedia article significantly improve system performance.
dblp:conf/tac/GongQT10 fatcat:cfjspao7jzde7jxnvujs72u24u