Page 184 - AWSAR 2.0

160 || AWSAR Awarded Popular Science Stories - 2019
in a forest and these factors were effectively mapped into the context of social network. In social networks, there are certain network- based and user-based features that play a prominent role in the spread of rumor, and by using these features; we can predict the probability of a user to get affected by a rumor and the extent of rumor spread in the network. The main factors influencing the forest-fire spread are the type of vegetation in the area, density of the forest, impact of velocity and direction of the wind, and topography of the area, and based on these factors, we can predict the chances of a tree catching fire. Similarly, the features we considered for modeling rumor propagation through social networks are the account analysis of the user, follower following ratio, presence of hashtags and keywords and the general account activity of the user, respectively. In addition to the above-mentioned factors, there is one more feature that
plays a significant role in the fast spread of forest-fire, which is the number of neighboring trees burnt. This factor is also taken into consideration in our approach where we analyze the number of neighboring nodes of a particular user who have already shared the rumor, which increases the tendency of that user to share the rumor.
Based on these factors,
we proposed an algorithm
which finds out the probability
of users to get affected by the
rumor and the extent of spread
in the network. The proposed
algorithm aims to find out the different paths through which rumor diffuses across the network and identify the users who played the key role in its spread. We evaluated our proposed algorithm in two rumor datasets
which we compiled based on the rumor tweets that circulated on Twitter. Rumors propagated in Twitter regarding our Indian National Anthem and 2000 rupee notes during demonetization were collected for a period of 6 months to conduct our study. One of the major conclusions from our study is that most of the rumor tweets propagating in social networks are in the form of retweets since the users involved in rumor diffusion share the rumor news from their neighbors without verifying the truth. The increased number of neighboring users who shared the rumor and the number of retweets increases the chances of rumor sharing by a user. This clearly indicates the fact that normally rumors are initiated by a few people, while the majority of the people involved in rumor diffusion share the rumor tweets posted by others. Another important finding from our research is that majority of the rumor spreaders have retweeted the
rumor tweet of a specific user and most of the people who initially shared the rumor have a huge number of followers in their network which enhanced the speed of rumor diffusion. This work entitled “A Nature- inspired Approach Based on Forest Fire Model for Modeling Rumor Propagation in Social Networks “was published in the Journal of Network and Computer Applications – Elsevier (IF-5.273) (doi= “https://doi.org/10.1016/j. jnca.2018.10.003”), which is one of the most prestigious journals in our research area.
Our proposed approach helps to identify the key factors intensifying the quick spread of rumors in social network and to detect the influential spreaders involved in misinformation propagation.
The identification and labeling
of fake content from the huge volume of data flowing through these networks with human intervention is a laborious task. Hence, there is a need to develop novel and effective mechanisms to limit the spread of fake news through these networks and allow users to enjoy the benefits of these networks to the fullest.