Natural language processing (NLP) has made considerable progress in the last decade, especially since the advancements of deep neural networks. Several of these technologies have been developed in academic research labs and were applied in real-world applications, powering several tasks like search, recommendation, autosuggestion, etc. Another promising subfield of NLP is document processing (e.g. scholarly articles, Wikipedia articles...). In this seminar, we will explore literature in this domain and look into the following questions:
- What are the drawbacks of current NLP technologies when applied to long documents?
- What are the document-level problems that can be solved with these approaches?
- How can we read academic literature critically and provide useful feedback on unpublished work?
- Dozent*in: Thomas Arnold