Retrieval Augmented Zero-Shot Text Classification
Keywords: Query reformulation, Zero shot classification, Text Embeddings, Retrieval Augmented learning
TL;DR: We present QZero, a novel training-free approach that reformulates queries by retrieving supporting categories from Wikipedia to improve zero-shot text classification performance.
Abstract: Zero-shot text learning enables classifiers to handle unseen classes efficiently, alleviating the need for task-specific training data. A
simple approach often relies on comparing embeddings of queries to those of potential classes. However, the embeddings of a simple
query may lack rich contextual information, which hinders performance. Traditionally, this has been addressed by improving the embedding models’ quality with expensive training. We introduce QZero, a novel training-free approach that reformulates queries by retrieving supporting categories from Wikipedia to improve zero-shot classification performance. Our experiments across six diverse datasets demonstrate that QZero enhances performance for state-of-the-art static and contextual embedding models without the need for retraining. Notably, in News and medical topic classification tasks, QZero improves the performance of even the largest OpenAI embedding model by at least 5% and 3%, respectively. Acting as a knowledge amplifier, QZero enables small word embedding models to achieve performance levels comparable to those of larger contextual models, leading to significant computational savings. Additionally, QZero offers meaningful insights that illuminate query context and verify topic relevance, aiding in understanding model predictions. Overall, QZero improves embedding-based zero-shot classifiers while maintaining their simplicity. This makes it particularly valuable for resource-constrained environments and domains with constantly evolving information.
Submission Number: 18
Loading