Interactive Semantic Featuring for Text Classification
This work addresses the need for interpretable features in text classification, but it is incremental as it builds on existing dictionary methods.
The paper tackled the problem of improving dictionary features for text classification by introducing smoothed dictionary features that consider document contexts, and demonstrated that models using these human-comprehensible features are competitive with Bag of Words models.
In text classification, dictionaries can be used to define human-comprehensible features. We propose an improvement to dictionary features called smoothed dictionary features. These features recognize document contexts instead of n-grams. We describe a principled methodology to solicit dictionary features from a teacher, and present results showing that models built using these human-comprehensible features are competitive with models trained with Bag of Words features.