Automatic Method to Rate Websites Based on Terms of Services

Abstract

Terms of services agreement of a website, though neglected by most of the users, plays a major role in deciding whether the website policies are designed by considering the developer /owner and users rights and needs. This paper focuses on developing an automatic tool that ranks websites based on their terms of services agreements using concept of natural language processing. This is the first such attempt in the field. The developed tool uses bag of words text classification approach and 2-layered artificial neural network. The method works in two phases: first phase consists of training and machine learning. It classifies terms into good, bad and neutral classes. This cond phase defines rating scale and ranks websites into classes A, B, C, D and E. Around 65.5% accuracy observed during testing phase opens the doors for research in developing such tools.

+ View Full Abstract