JVN-TDT Entity Linking Systems at TAC-KBP2012
Hien Phuong Nhat Nguyen, Huy H. Minh, Tru Hoang Cao, Trong T. Nguyen, Viet Nam · Theory and applications of categories · 2012
We present two methods for entity linking in two of our systems submitted to TAC/KBP 2012. The first one, namely Method 1, learns coherence among co/occurrence entities re/ ferred to within a text by exploiting Wikipe/ dia’s link structure and the second one, namely Method 2, combines some heuristics with a statistical model for entity linking. Method 1 exploits two features to train a classifier and exploits coreference relations among co/ occurring mentions. Method 2 is a hybrid me/ thod containing two phases. The first phase is a rule/based phase that filters candidates and, if possible, it disambiguates mentions with high reliability. The second phase employs a statis/ tical model to rank the candidates of each re/ maining mention and choose the one with the highest ranking as the right referent of that mention. Experiments are conducted to eva/ luate two methods on two datasets – TAC/ KBP2011 and TAC/KBP2012 datasets.