Improving learning by choosing examples intelligently in two natural language tasks

摘要

In this chapter, we present relational learning algorithms for two natural language processing tasks, semantic parsing and information extraction. We describe the algorithms and present experimental results showing their effectiveness. We also describe our application of active learning techniques to these learning systems. We applied certainty-based selective sampling to each system, using fairly simple notions of certainty. We show that these selective sampling techniques greatly reduce the number of annotated examples required for the systems to achieve good generalization performance.