Prediction Focused Topic Models via Feature Selection
arXiv:1910.05495
Abstract
Supervised topic models are often sought to balance prediction quality and interpretability. However, when models are (inevitably) misspecified, standard approaches rarely deliver on both. We introduce a novel approach, the prediction-focused topic model, that uses the supervisory signal to retain only vocabulary terms that improve, or at least do not hinder, prediction performance. By removing terms with irrelevant signal, the topic model is able to learn task-relevant, coherent topics. We demonstrate on several data sets that compared to existing approaches, prediction-focused topic models learn much more coherent topics while maintaining competitive predictions.
AISTATS 2020. arXiv admin note: substantial text overlap with arXiv:1911.08551
References in corpus (5)
- Supervised Topic Models
- Prior matters: simple and general methods for evaluating and improving topic quality in topic modeling
- Prediction-Constrained Topic Models for Antidepressant Recommendation
- Prediction-Constrained Training for Semi-Supervised Mixture and Topic Models
- Variable Selection for Latent Dirichlet Allocation