English temporal expression recognition based on Conditional Random Fields

Temporal expressions play a very important role in a lot of natural language processing areas, such as topic detection and tracking. This paper carries out extensive studies on English temporal expression recognition based on the Conditional Random Fields model. We mainly explore the usages of the features in the recognition, which are consisted of four kinds of features, i.e. Word Feature, Dictionary Feature, Number Feature and Character Feature. By the analysis of feature selection, we find that the choice and the implementation of features are the key components to the recognition system and their results are very important to the performance of the recognition system. We also analyze the reasons of the recognition errors, and provide the possible solutions to these errors.