Self-Compositional Data Augmentation for Scientific Keyphrase Generation
This addresses the challenge of costly keyphrase labeling for researchers and practitioners in scientific domains, though it is incremental as it builds on existing data augmentation techniques.
The paper tackles the problem of limited training data for keyphrase generation by introducing a self-compositional data augmentation method that combines similar documents based on shared keyphrases to create synthetic samples, resulting in consistent improvements across multiple datasets in three domains.
State-of-the-art models for keyphrase generation require large amounts of training data to achieve good performance. However, obtaining keyphrase-labeled documents can be challenging and costly. To address this issue, we present a self-compositional data augmentation method. More specifically, we measure the relatedness of training documents based on their shared keyphrases, and combine similar documents to generate synthetic samples. The advantage of our method lies in its ability to create additional training samples that keep domain coherence, without relying on external data or resources. Our results on multiple datasets spanning three different domains, demonstrate that our method consistently improves keyphrase generation. A qualitative analysis of the generated keyphrases for the Computer Science domain confirms this improvement towards their representativity property.