Advancing Automatic Subject Indexing: Combining Weak Supervision with Extreme Multi-label Classification

The multi-label automatic classification of scientific publications based on a pre-defined taxonomy, also called automatic subject indexing is a continuing research endeavor with significant cross-domain applicability. In this paper, we assess the performance of X-transformer and its variants with other extreme multi-label classification models for the above task. Our model Weak X-transformer achieves a micro F1-score of 0.65 and 64% accuracy on the task outperforming all other methods. We also investigate the impact of incorporating additional unlabelled data and hierarchical structure into the models. Our findings demonstrate that the transformer-based model with weak supervision outperforms other approaches, providing insights into effective strategies for extreme multi-label classification in scholarly publications.