The Open Automation and Control Systems Journal
2015, 7 : 1347-1351Published online 2015 September 14. DOI: 10.2174/1874444301507011347
Publisher ID: TOAUTOCJ-7-1347
FN-Rank: Domain Keywords Extraction Algorithm
College of Information Engineering,
Minzu University of China, Beijing, 100081, China.
ABSTRACT
Domain keywords extraction is very important for information extraction, information retrieval, classification, clustering, topic detection and tracking, and so on. TextRank is a common graph-based algorithm for keywords extraction. For TextRank, only edge weights are taken into account. We proposed a new text ranking formula that takes into account both edge and node weights, named F2N-Rank. Experiments show that F2N-Rank clearly outperformed both TextRank and ATF*DF. F2N-Rank has the highest average precision (78.6%), about 16% over TextRank and 29% over ATF*DF in keywords extraction of Tibetan religion.