Typicality and Natural Language Learning

About

People

Publications

Related Publications

About:

There are various ways of measuring how "typical" or representative an instance is with respect to its class. Previous work has shown that when machine learning is applied to many natural language processing tasks, non-typical training examples play an important role in improving generalization accuracy. We are exploring whether such results generalize to several classification tasks in the area of spoken dialogue. In addition, we are exploring how different formalizatons of "typicality" impact the performance of memory-based and rule-based learning algorithms.


People:


Publications:


Related Publications:

  • Previous publications on the spoken dialogue corpus currently being examined