Compare Modeling Contagion Through Facebook News Feed and Cascading Behavior in Large Blog Graphs

From Cohen Courses
Jump to navigationJump to search

This page shows a comparison between two scientific papers: Gesundheit! Modeling Contagion through Facebook News Feed (AAAI 2009) and [Cascading Behavior in Large Blog Graphs (SDM 2007). This page is written by Tuan Anh.

Problem

Both the two papers present empirical studies on how information diffuse in social media and propose models for modeling the diffusion process. However, each paper focus on a different perspective of diffusion. In the paper [Cascading Behavior in Large Blog Graphs, the authors employ the independent cascade model and studies size and shape of the cascades; and propose generative models fitting the observed data. On the other hand, the paper Gesundheit! Modeling Contagion through Facebook News Feed focuses on analyzing effects of the starter nodes (the one starting the diffusion) on scale of diffusion, e.g., the chain length.

Dataset

The paper Cascading Behavior in Large Blog Graphs use a large blog dataset which allow explicit diffusion of information can easily be tracked using citing link among blog post.

The paper Gesundheit! Modeling Contagion through Facebook News Feed use a large "like the page" dataset which contains time stamp when a friend link is created or a "like" action is performed. This requires a heuristic way to determine if a "like the page" is diffused from a user to her friend. In the paper, the authors do this by employing a 24 hours window: a "like the page" is considered to be diffused from a user to another user if the later sees the former doing the "like" (via News Feed) within 24 hours.

Method

Both two papers conduct some descriptive statistics to get insights from datasets. Based on the obtained observations, the authors propose some models for fitting the dataset. Two different methods are used in the two paper. In the paper Cascading Behavior in Large Blog Graphs, the authors use SIS model and its variations, while in paper Gesundheit! Modeling Contagion through Facebook News Feed, Binomial regression is used.

Other

Both the two paper do not study the diffusion at individual level, i.e., the decision to adopt the information at each user, and do not consider the difference in diffusion of different content.

Additional Questions

1. How much time did you spend reading the (new, non-wikified) paper you summarized? I read the paper sometimes before, and spent about 1 hour to read it again before writing the summary.

2. How much time did you spend reading the old wikified paper? 1 hour

3. How much time did you spend reading the summary of the old paper? 10 minutes

4. How much time did you spend reading background materiel? 20 minutes

5. Was there a study plan for the old paper? Yes, I spent 1 minute to read a short explanation about a technique used in the paper.

6. Give us any additional feedback you might have about this assignment. I think the interesting paper by Lada Adamic How You Met Me is more appropriate to make a comparison with with the old paper than the work by Leskovec et. al.