ToS Summarizer
Term of service summarizer to know what you're signing up for.
Created on 15th December 2019
•
ToS Summarizer
Term of service summarizer to know what you're signing up for.
The problem ToS Summarizer solves
We are working on a Machine Learning-based model in which our dataset will be the data of Terms and Condition and our model will be extracting the important terms that reader or user should know about and will form a kind of summary from the data(contract) provided. The reason of working on this model is that many users don't always read the terms and condition before using a particular application and just check the box of "agreed" and go ahead, so our model will sort the important and red flags, advantages and disadvantages of terms from the dataset and will show it to the user. A lot of more valuable things will also be added in the model to make it more useful for example the length of the summary to be shown to the user.
Challenges we ran into
Unavailability of appropriate dataset for extraction of relevant terms for legal documents. We used a twitter dataset to extract the terms and tried to modify it so as to meet our needs.
Selection of an appropriate machine learning model for relevant term classification. We tried different models and mainly focused on Support Vector Machines for optimization. We finally used Linear Support Vector Classifier because of its faster model training and better minimization of squared hinge loss.