Automatic Identification And Classification of Contract Data Types with NLP
- Mentors
- Niall Roche
- Organization
- Accord Project
Mapping between the natural language text of a contract and attempting to classify data types such as monetary amounts, dates and legal specific terms such as agreement parties, etc. into cicero variables from the existing model library.
Therefore, users can export a smart legal contract by natural language contract text.
How to do it?
A. Data Mapping using NER by RoBERTa.
Build a scaleable NER model by Adapter Transformers based on RoBERTa.
The model also have Active Learning pipelines. User can define their own custom data type label then upload data and train the Adapter. By doing so, the model will recognize their new tag.
B. Suggest about templates by Classification Model
When users first upload their natural language contract, NLP model will tell them which smart legal contract template is suitable. So users can use or fine-tune the contract easily.
C. Identification contract variables by BERT QA Model
Input contract text, NLP model will suggest user which they need to put onto smart legal contract's variables.
D. API backend.
User can call the NLP model by API. Provide a Swagger UI Documents and plenty of examples.