Beat the News

 

About

The production of intelligence from a broad range of data sources is seen by many agencies as an increasingly important capability that can complement the traditional intelligence sources used by defence, national security and law enforcement agencies.

Analysts within these agencies are often interested in monitoring and predicting population-level events such as social disruption, political crises, and disease outbreak. Many such events are preceded by population-level changes in behaviour and communication, which can be directly or indirectly observed through data.

The Beat the News™ program, inspired by the IARPA Open Source Indicators program and the Virginia Tech EMBERS project, seeks to develop, integrate and evaluate technology that will automatically and accurately predict the occurrence of future population-level events such as social disruption, political crises, election outcomes and disease outbreak. The predictions will be generated through analysis of data from a diverse range of sources.

Streams

Development and Operations

Dr Grant Osborne (Program Lead)

Data to Decisions CRC

Beat the News

The Beat the News™ engineering team supports the research streams by providing access to curated data sets via a state-of-the art data science platform it built. The team also develops a secure user-facing demonstration system for viewing predictions, develops its own predictive models, and integrates the models developed by research streams into the system. The team responsible for maintaining the overall project roadmaps.

Disease Prediction

Dr Lewis Mitchell

The University of Adelaide

Beat the News

This stream will develop new operational systems for the real time prediction of both common, seasonal disease outbreaks and rare, intermittent outbreaks by assimilating open data into dynamical epidemiological forecast models. As a result, public health officials will have a better idea of baseline infection rate for the detection of potential outbreaks which means improved estimates of peak timing and burden, many weeks in advance. 

Proactive Approach to Potential Social Disruption

Professior Jiuyong Li

University of South Australia

Beat the News

This stream will build an accurate, interpretable and easily maintainable classification and prediction system for social disruption events as well as develop an evaluation module which will assess the quality of generated warnings. As a result, this project will generate disruptive societal event warnings for end users, help government agencies to take precautionary actions in advance and reduce potential societal costs and disturbances.

Social Disruption and Election Prediction

Dr Lewis Mitchell

The University of Adelaide

Beat the News

This stream will predict social disruption events and forecast election results ahead of time, by building statistical models which represent the network of causal influences on events of interest, based on multiple data sources. This project will provide end-users with a better predictive analysis of potential hot-spots for social disturbances. Election prediction will provide analysts with an overview of political trends, engagement and sentiment in both Australia and abroad.

Participants

PhD Profiles

Ang Yang

University of South Australia

Topic

An Information Quality Model for Big Data

Beat the News

Dinithi Jayaratne

La Trobe University

Topic

Advanced Predictions in Social Media by Incorporating User Generated Content

Beat the News

Jeff Ansah

University of South Australia

Topic

Discovery and Use of Network Structural Features for Social Disruption Prediction

Beat the News

Madhura Jayaratne

La Trobe University

Topic

Scalable Big Data Analytics Techniques for the Integration of Structured and Unstructured Information

Beat the News

Max Glonek

University of Adelaide

Topic

A Census of Social Media Users: Statistical Techniques for Quantifying and Correcting Biases in Big Open Data Sources

Beat the News

Md Zahidul Islam

University of South Australia

Topic

Civil Unrest Events Detection from Multiple Credible Sources

Beat the News

Peter Mathews

University of Adelaide

Topic

Learning Theory and Algorithms for Large Scale Probalistic Graphical Models

Beat the News

Tharindu Bandaragoda

La Trobe University

Topic

Real-time Cognitive Analysis for Capturing Suspicious Behaviours

Beat the News

Yujie Wang

La Trobe University

Topic

New Multi-Dimensional Knowledge Base to Capture and Store Patterns in Evolving Text Streams

Beat the News

Xuying (Ada) Yao

University of South Australia

Topic

Precursor Pattern Analysis and Interpretable Classification

Beat the News

Caitlin Gray

University of Adelaide

Topic

Modelling Information Cascades: Creating Predictive Models on Temporal Networks

Mark Carman

University of South Australia

Topic

New Authorship Attribution Methods to Detect and Manage Social Manipulation and Fake News in Social Media

Sha Lu

University of South Australia

Topic

Risk Assessment Using Causal Based Methods

Dennis Liu

University of Adelaide

Topic

The Interaction of Social Media with Vaccination and Disease Outbreaks

Nam Truong

University of South Australia

Topic

Using Causal Based Methods, Such As Bayesian Networks, For Fraud Detection

Publications

A Temporal Classification based Predictive Model of Recurring Societal Events, J Chen, W Kang, J Li, J Liu, L Liu, B Cooper, N Lothian, G Osborne and T Moschou

A data-driven model for influenza transmission incorporating media effects, L Mitchell, J Ross 

The Nature and Origin of Heavy Tails in Retweet Activity, P Mathews, L Mitchell, G Nguyen, N Bean

Appo: Real-time Influenza Forecasts for Australia, J Walker, L Mitchell and J V Ross

SEAIR DTMC and CTMC Model with False ILI Observations, J Walker

Beat the News Baseline Predictive Models with MITRE GSR Data, J Li and J Chen

Time-series pattern matching based prediction, W Kang

A Conceptual Framework for Bayesian Model Fusion in the Presence of Uncertainty, L Mitchell, A Hossny, J Tuke and N Bean

Can Keyword Volume Predict Civil Unrest from Twitter? An Exploratory Study, A Hossny, J Tuke, N Bean and L Mitchell

Literature Review on Open Source Indicators (OSI) Program Related Publications, J Chen

Reports on Techniques of EMBERS System, J Chen

Domestic Political Crises Detecting & Forecasting, W Huang

Dengue Forecasting Model, G Osborne, T Moschou, N Lothian, A Lane and G Jiang

Phrase Matching for Planned Protest Detection, Y Huang