Title :
Dependency Tree-Based Anaphoricity Determination for Coreference Resolution
Author :
Kong, Fang ; Zhou, Jianmei ; Zhou, Guodong ; Zhu, Qiaoming
Author_Institution :
JiangSu Provincial Key Lab. for Comput. Inf. Process. Technol., Nanjing, China
Abstract :
This paper proposes a new scheme to determine the tree span structure for tree kernel-based anaphoricity determination in coreference resolution. Given a sentence and current mention, it gets all the dependencies, encodes the order of all the words, constructs the corresponding dependency tree and keeps the nodes and the paths related with current mention based on constituent dependencies, while removing the noisy information, eventually leading to a dynamical dependency tree (DDT). Evaluation on the ACE 2003 corpus shows that the DDT is effective for anaphoricity determination, and that applying our anaphoricity determination module in coreference resolution achieves the improvement of performance.
Keywords :
linguistics; tree data structures; coreference resolution; dependency tree-based anaphoricity determination; dynamical dependency tree; tree span structure; Convolution; Hidden Markov models; Kernel; Machine learning; Noise measurement; Semantics; Syntactics; Anaphoricity determination; Coreference resolution; Dependency tree;
Conference_Titel :
Asian Language Processing (IALP), 2010 International Conference on
Conference_Location :
Harbin
Print_ISBN :
978-1-4244-9063-9
DOI :
10.1109/IALP.2010.25