Paper: Imbalanced Classification Using Dictionary-based Prototypes and Hierarchical Decision Rules for Entity Sense Disambiguation

ACL ID C10-2098
Title Imbalanced Classification Using Dictionary-based Prototypes and Hierarchical Decision Rules for Entity Sense Disambiguation
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2010
Authors

Entity sense disambiguation becomes dif- ficult with few or even zero training in- stances available, which is known as im- balanced learning problem in machine learning. To overcome the problem, we create a new set of reliable training in- stances from dictionary, called dictionary- based prototypes. A hierarchical classifi- cation system with a tree-like structure is designed to learn from both the prototypes and training instances, and three different types of classifiers are employed. In addi- tion, supervised dimensionality reduction is conducted in a similarity-based space. Experimental results show our system out- performs three baseline systems by at least 8.3% as measured by macro F1 score.