Feature Generation for Robust Semantic Role Labeling
This work addresses the challenge of feature engineering for NLP practitioners by automating feature generation, making it more accessible and efficient, though it is incremental as it builds on existing methods for robust models.
The paper tackled the problem of reducing the expertise and effort needed for hand-engineered feature sets in NLP by automatically generating rich features from simple units called featlets, using information gain to guide the process, resulting in models that rival state-of-the-art performance on two standard Semantic Role Labeling datasets with minimal task-specific insight.
Hand-engineered feature sets are a well understood method for creating robust NLP models, but they require a lot of expertise and effort to create. In this work we describe how to automatically generate rich feature sets from simple units called featlets, requiring less engineering. Using information gain to guide the generation process, we train models which rival the state of the art on two standard Semantic Role Labeling datasets with almost no task or linguistic insight.