Social Networks 2018

Ongoing Projects

Modeling Fake News Diffusion

fakemodelAs time passes, social networks get more and more important in modern life and they play a vital role in what’s going on in the societies. Furthermore, fake news and misinformation diffusion have become a very common phenomenon in social networks. Thus, controlling the diffusion of information (especially fake news) on social networks can be very beneficial. Based on previous researches, fake news and real news have different propagation patterns and features, thus with some effort, they can be distinguished from each other.

In this project, we propose a novel model for spreading the fake news. Our work is based on fake news and real news propagation pattern and the behavior of network’s users. The work is done over twitter data but it can be generalized to other social networks.

People involved:  Soroush Omranpour, Maryam Ramezani, Negar Mokhberian, Hamid R. Rabiee

Fake News Mitigation via tracking information spreading

fakeToday, the communication of people affected by social networks has undergone some changes. People use these networks to share information and news with others. The speed and effectiveness of the development of information in this context provides the basis for the spreading of false information. Members of a network sometimes unintentional participate in the spreading of false information, sometimes deliberately and for degrading purposes, such as reducing brand customers or creating unrest in a community. Hence, tracing rumors to reduce the extent of its distribution in the network is important.

In this project, we intend to:

  • Early detecting the type of news propagating over the network
  • Identify the behaviour of individuals and propose a method to reduce the range of fake news spreading by tracking the types of information disseminated among the users

People involved:  Mina Rafiee, Maryam Ramezani, Hamid R. Rabiee

Network Completion with Partially Observed Data


Access to real data sets including interactions and diffusion is the main requirement for analysis of social networks. However in most cases, restriction of data access and privacy laws lead to incomplete data collection.

In this project, we proposed a method to jointly complete the missing data of diffusion and interaction of a network, while accessing to a limited portion of these information. The model is based on the extracted information from partially observed data and their dependencies.

People involved:  Maryam Ramezani, Hamid R. Rabiee

Analysis, Clustering and Detecting the Bursty Information Cascades

viral-presidentBurst‎, ‎defined as “a brief period of intensive activity followed‎ by long period of nothingness” is a common phenomenon in human activities‎. ‎In social networks big information cascades of post resharing can form as users share posts with their friends and followers‎. ‎Predicting the ultimate popularity of a post and virality timing is important for content ranking‎, ‎marketing purposes and controlling rumors and fake news‎.

In this project‎, ‎we wish to cluster viral cascades based on their temporal and structural dynamics‎. ‎Apart from studying dynamics of bursty cascades and evaluating the correlation between the behaviors of cascades and different their features‎, ‎our work also includes introducing a framework for predicting whether a cascade will get viral or not‎, ‎the path of bursty cascades , ‎propagation behavior and timing‎.

People involved:  Pouya Kanyanian, Maryam Ramezani, Hamid R. Rabiee

Macro-Level Cascade Prediction

1_2ps4KlEODqLbgYt9u5Li7wHow to promote items, ideas, beliefs, new technologies and products in society is one of the important issues in social networking sciences known as information diffusion. The information exchanging between people is related to their communicational structure.

In this project, relying on how the posts spread among users, we proposed a model which could predict the future behavior of a new cascade in the macro level view of a network by observing a short period partially cascade. The prediction of this method is about the regions which become polluted by the diffusion over the time.

People involved:  Mahta Shafieesabet, Maryam Ramezani, Hamid R. Rabiee



Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Complex Networks | Digital Media Lab