Social media analysis

By studying patterns in social media data we discover more about community structures, their topics and trends. We can use this research to extract socially meaningful indices. In our research into social media, we applied these indices to help recognise early warning signs for suicide.

Understanding social media dynamics and pervasive signals

We explore not only text in blogs, but also connectivity through comments. We seek an understanding of the underlying dynamics and patterns of interaction across specific cohorts of people, particularly those with mental health problems.

So how do we research social media data and what do we look for? In our research we study the following social media patterns:

  • Who is talking to whom?
  • What are they talking about?
  • What's their sentiment, and how does it change over time?

Our project also aims to use signals obtained from unobtrusive data sources in personal devices, such as GPS location information, to extract socially meaningful indices for personal media.

We extract meaningful social context

  • What is the activity?
  • Who is co-located with whom?

Recognising early signs of deterioration of mental health: the Groundtruth project

PRaDA is working with the Black Dog Institute to identify deviations in expression on social media. The aim is to predict deterioration of mental state, and develop an early-warning system.

The collaboration, termed the Groundtruth Project, calibrates data from social media blogs with clinical measures of anxiety and depression.

Associate Professor Dinh Phung is leading this project in collaboration with Black Dog Institute. Dr Thin Nguyen’s expertise in large-scale social media extraction and analysis enables him to efficiently extract and parse diverse social media.

Contact us

Centre for Pattern Recognition and Data Analytics (PRaDA)
General enquiries
+61 3 5227 8797
Email PRaDA

School of Information Technology
Deakin University
Locked Bag 20000
Geelong VIC 3220