From a single "like" to social influence: How can we make sense from social network dynamics?

Theoretical (Analytical):

Practical (Implementation):

Literature Work:


Overview

Understanding the dynamics of information propagation, influential identification and trend analysis spark great interest within the field of social network analysis. The typical approach is to view a social media user, representing a registered person or organization, as a single node within the network. The edges between two nodes are usually drawn by user-to-user interactions as, for instance, befriending, writing, liking, or forwarding a message to another user. These abstractions enable researchers to apply selected methods of graph theory to solve practical problems or to use visualization analytics to gain further insights.
 

Problem Statement

Recent developments within the fields of information diffusion and text mining enable us to analyze social networks in close detail. However, to eventually gain insight into developed models and detected network dynamics, Visual Analytics is needed to answer various questions:

 

Influence Research
How can we display influence and influencials within a given network? How can we understand an influence model's decision from a given social network? Which network artifacts are relevant to be visualized and which interaction methods support us to gain further insight into given influence dynamics?

Trend Analysis & Entity Progression
How can we display trends in social media and which interaction methods are favorable to analyze them? How can the progression of entities be visualized? How can we visualize and identify filter bubbles and similar groups within a given network? 

Information Diffusion Prediction
How can we use current information diffusion models and network attributes to visually predict social media cascades? How can we understand the effect of employing new social media users and artifacts? 

 

 

Tasks

This project description is held intentionally broad and aims to encourage to find own, exciting research questions to answer. The common procedure can look like the following:

  • Find a relevant research question within the field of information diffusion which can be answered using Visual Analytics.
  • Identify related projects and prepare relevant data source(s). 
  • Implement a feature engineering pipeline.
  • Develop a VA system to answer your research question.

Requirements

Good programming skills in Java or Python and Javascript. Knowledge in D3.js preferable.

 

Scope/Duration/Start

  • Scope: Bachelor/Master
  • 6 Month Project, 3 Month Thesis (Bachelor) / 6 Month Thesis (Master)
  • Start: immediately

Contact

References

Deng, Zikun, et al. "Visual cascade analytics of large-scale spatiotemporal data." IEEE Transactions on Visualization and Computer Graphics 28.6 (2021)