Simplifying Privacy Agreements using Machine Reading Comprehension and Open Domain

Privacy Agreements are critically important from the company as well as the user's point of view. Upon surveying, we realise there are three most common issues faced by the user while trying to read their privacy agreements. First is excessive use of complex language and jargon that make it extremely difficult for the user to comprehend the agreement. The second issue is that if the user has a few questions with respect to the agreement, they are not answered directly. Lastly, there lies an absence of an agreement evaluator, that can help compare the document with the standards of the industry. We aim to alleviate this situation by proposing a three-fold technique involving Text Summarization, automated question answering and a Document Evaluator. Reading comprehension is one of the most critical abilities in human intelligence, and machine reading comprehension is one of the most popular research directions in the field of natural language processing. Machine reading is one of the key aspects to achieving artificial intelligence. The introduction of deep learning and large datasets in the recent few years has encouraged the success of MRC. Additionally, the open domain automatic question answering models have been widely studied in recent years. However, the problem of lack of sufficient data to read information from exists that makes questions out of the context of the text seem difficult to answer. In this study, we propose a novel framework to integrate MRC along with Open Domain to answer questions with increasing accuracy. This is achieved by classifying the question and retrieving the answer from multiple sources. Our model should be able to extract answers from discontinuous spans across the passage text. The answer that has the highest confidence associated with it is returned as the final answer. Moreover, the existing passage has been updated based on the questions answered.