Inspiration
Many times we see that even after all the privacy policies are right in front of our eye. This is because no one will make an effort to read a document which is 4-5 pages large. So, this inspired us to build something meaningful as well as useful at the same time!
What it does
What it does is when a user will visit a website they have to activate our extension which will then show a list of the important point from all those long pages. These points will contain a score based on how secure the website is.
How we built it
We built it using technology stack like Natural Language Processing, Google's BERT model for training our dataset and using their state of the art model.
Challenges we ran into
We had much less time to train the model and so we were continuously facing problems for that. Secondly, we had to understand the data thoroughly and gain all the information listed above.
Accomplishments that we're proud of
We are proud to say that using our QnA feature, one can ask questions related to the privacy policies and they will get a response like if any human is replying! This feature is totally independent of any automated technology available like DialogFlow. We have trained the model in a TPU and gained a very high accuracy.
What we learned
We learned a lot of things whether it was related to technology or social. Things like healthy competition proved to be the main factor in our efforts.
What's next for dotslash-piedpiper
We are thinking of taking this a step further in which the user can ask the questions from any legal document and can ask questions like to whom is the document is referring to. This thing is under the experimental phase and we have to be very precise as it is a matter of someones property/money. So, we are very curious to bring this thing in the deployment phase as it will be very useful if successful.
Log in or sign up for Devpost to join the conversation.