Contract Annotator

Inspiration

Contracts are confusing! They're long, they have obscure jargon, but they need to be carefully read and understood. We created this tool in order to make this process easier by highlighting the most important parts of your contract so you don't miss them.

What it does

Given a file of an unannotated contract, Contract Annotator uses NLP to find the most important parts of a document and return an annotated copy of the contract. It uses the DocuSign API to then share it with other involved parties and track viewings, comments, and signings. Contract Annotator streamlines contract comprehension, and allows entrepreneurs, students, and all people who don't have the time or money to retain and wait for a lawyer to work with contracts effectively.

How we built it

We used a python script to analyze PDFs and interface with DocuSign. First, it extracts the information from the contract, either by stripping text data from a text-format PDF or by using OCR on scanned PDFs.

It uses TextRank to select important parts of the contract using extractive summarization, augmented using domain knowledge regarding contract law and conventions in business documents. It also tokenizes sentences and runs a keyword, then generates a Word document with the appropriate highlight and exports it as a PDF to send back through email. Using DocuSign's API, it sends contracts to collaborators and keeps track of progress towards an agreement.

Challenges we ran into

We had some fun technical challenges, including one broken laptop and strange clang installs! Additionally, integrating across file formats and different categories of PDF proved to be a challenge, which we were eventually able to figure out.

Accomplishments that we're proud of

We did what we set out to do! Contract Annotator is a useful tool which simplifies the contract review process. It might not be the most buzzword heavy project, but it solves a pressing business need and is something that every startup or employee should have access to! "Boilerplate" or "standard" contracts can be really tricky to actually analyze, and this removes one more step.

What we learned

We learned about how hackathons work (this was our first one!), and about using OAuth to access company APIs. Furthermore, this is the first time we've gone into the contract space, and learning about the ins and outs of it from a legal perspective was really interesting.

What's next for Contract Annotator

We want to build up the UI to make the experience and usage more smooth and easy to control. We want to incorporate an email interface, where like the workflow with lawyers, you can email your contract to us, and our script will return the annotated version in DocuSign.