This paper uses a Transformer-based model to categorize documents more accurately by figuring out the specific meaning of a word based on the domain it's used in (e.g., "bank" in finance vs. "bank" in geography). Source: ACL Anthology P19-1105
2. Sentences with Gapping: Parsing and Reconstructing Elided Material (2018) Computational Linguistics
3. Text Categorization by Learning Predominant Sense of Words (2019) Machine Learning / NLP