Legal Clauses Prediction Model

There is lack of accessible labelled data that has hindered the application of deep learning in various mastered domains. However, a new solution has been introduced to solve this obstacle within the legal domain of this industry. The creators of this solution are The Atticus Project and they have developed the Contract Understanding Atticus Dataset (CUAD), which is a new dataset for legal contract review. CUAD consists of over 13,000 annotations and was created with the help of dozens of legal experts. The aim of this dataset is to highlight the important portions of a contract that require human verification. Whilst initial results of Transformer models have shown promising performance, there is still a significant room for improvement. This dataset is one of the few large, specialized NLP benchmarks annotated by experts, making it a challenging research benchmark for the wider Natural Language Processing community. Thus, CUAD dataset has the potential to address the bottleneck in deep learning within the legal domain.