An NYU and Columbia Research Collaboration to advance the state of the art in open source legal NLP work.

We believe the open source machine learning systems analyzing long form documents in the legal domain can be improved via a multimodal approach. We aim to improve such systems by incorporating hierarchical document information--such as the structure and layout of documents--in addition to the document's text. This approach aims to improve various downstream tasks that are difficult for current text-only systems due to the length and complexity of legal contracts.