Muhammad Muneeb Raza

Academic

Career

Verbatim text

Summix - Automated Privacy Policy Analyzation and Summarization 
 
According to Global Digital 2009 reports, there are 5.11 billion unique smartphone and computer users in the world today and 97% of people consent to legal terms and services conditions without reading them. Therefore we focused on developing a tool to tackle this problem. Summix automates the process of Privacy Policy Analyzation and Summarization by extracting key features of Privacy Policies and generating an extractive and abstractive description of the key information. For this, we have used Natural Language Processing Techniques and Deep Learning Models. Since there is no large dataset available, the Privacy Policies were scraped from thousands of applications and websites and then after sufficient pre-processing the corpus was made. To cop up with this Unsupervised problem, Firstly, we have trained ConceptNet Word Embeddings on the corpus and then the trained embeddings were passed to the deep learning models. We’ve also used Attention mechanism to only pay attention to some important points for key features extraction and maintaining the semantics of the generated output. For Topic Generation, we have used another self-scraped dataset and fine-tuned our previously used models. We have also developed an interactive Android-application to users for using the system. Features include: 
 Search for any privacy policy, the system will scrap the text automatically or enter by own.  
 Extractive Summary of the input. 
 Abstractive (human like and more intuitive) Summary of the input. 
  
Topics/Headlines of the input. 
 
 
 
 
 
 
 
 
 
Technology Used: 
Python, TensorFlow, NLTK, Android  
Supervisor Name: 
Dr. Omer Beg 
Group Members:   
Muhammad Shahzeb Ali (i16 - 0246) 
Muhammad Muneeb Raza (i16 - 0128)