View Distillation with Unlabeled Data for Extracting Adverse Drug Effects from User-Generated Data
This work addresses the challenge of extracting adverse drug effects from user-generated data, which is important for pharmacovigilance and public health monitoring, but it appears incremental as it builds on existing transformer methods with a novel distillation approach.
The paper tackles the problem of identifying Adverse Drug Reactions (ADR) in social media data by proposing a multi-layer transformer algorithm that uses view distillation with unlabeled data, resulting in significant outperformance over transformer-based models pretrained on domain-specific data on the largest publicly available ADR dataset.
We present an algorithm based on multi-layer transformers for identifying Adverse Drug Reactions (ADR) in social media data. Our model relies on the properties of the problem and the characteristics of contextual word embeddings to extract two views from documents. Then a classifier is trained on each view to label a set of unlabeled documents to be used as an initializer for a new classifier in the other view. Finally, the initialized classifier in each view is further trained using the initial training examples. We evaluated our model in the largest publicly available ADR dataset. The experiments testify that our model significantly outperforms the transformer-based models pretrained on domain-specific data.