Use Case of NLP – Legal Contract Clause Validation Chat Bot

Use Case of NLP – Legal Contract Clause Validation Chat Bot

February 2, 2019 DATAcated Challenge 0

Every day people are dealing with legal contracts. Legal contracts are at times hard to read for a common man. I really wish to propose an idea for a chat bot to which I can upload a legal contract and I can ask questions to the chatbot so that it will answer based on the contract clauses. This would actually help a lot of people who are stuck with court proceedings just because they are ignorant or they are misinformed about the rights / clauses in the contracts.The user should be able to check if at all the contract has any loop holes that the contract offerer might exploit later, Like length of contract, provision of Followup contract, Method of payment, contract breach penalty. Usually only a skilled lawyer is able to pick up these fields from a legal contract.

How to implement this ? Rough Framework.

  • Entity extraction for regular and common fields like Parties involved in the contract, Contract Length, Contract Breach Penalty etc.
  • Ambiguous Statements – At times contracts are filled with confusing and opposing statements. This could be highlighted and shared for the user so that he is better informed about the opposing clauses. Semantic Similarity/dissimilarity matching using Siamese BiLSTM network could be used for this task.
  • Contract Statement Threat level Assesment. Some statement might be vilfully added by the contract offerer and a Threat level indicator could help the user to know how situation might be exploited by the contract offerer.

A combination of these modules could help a user to understand the contract in a much better way and he can be informed about the possible pitfalls he may fall into at a later stage. As always knowledge is the key for any success and every individual has the right to be informed not to be exploited.

By: Jithin J Kumar

 

Leave a Reply

Your email address will not be published. Required fields are marked *