Expansion Finding For Given Acronyms Using Conditional Random Fields

Jie Liu,Jimeng Chen,Tianbi Liu,Yalou Huang

DOI: https://doi.org/10.1007/978-3-642-23535-1_18

2011-01-01

Abstract:There are increasingly amount of acronyms in many kinds of documents and web pages, which is a serious obstacle for the readers. This paper addresses the task of finding expansions in texts for given acronym queries. We formulate the expansion finding problem as a sequence labeling task and use Conditional Random Fields to solve it. Since it is a complex task, our method tries to enhance the performance from two aspects. First,we introduce nonlinear hidden layers to learn better representations of the input data, under the framework of Conditional Random Fields. Second, simple and effective features are designed. The experimental results on real data show that our model achieves the best performance against the state-of-the-art baselines including Support Vector Machine and standard Conditional Random Fields.