IBM Research - Australia 2013 Colloquium: Living in an Instrumented World - Demos
Australian Crisis Tracker
Australian Crisis Tracker (ACT) is a solution that communities and agencies can use to better understand what is being said in social media streams. During large-scale complex crises, individual disaster responders, government officials and/or citizens who wish to access this vast source of information are met with a torrent of text that can quickly cause information overload. Project ACT uses IBM technologies for text analytics, machine learning and intuitive visualizations to help improve communication and provide information .
Australian Disaster Management Platform
The purpose of the demo is to showcase the capabilities of the Australian Disaster Management Platform through a hypothetical scenario involving a predicted fire breakout in the Dandenong region. The demo will be driven from the perspective of a decision maker who triggers an assessment into the region of interest. The assessment will spawn a process involving a sequence of modeling and simulation systems, the orchestration and coordination of which, highlights the core capabilities of the platform.
This real-world solution, which has been deployed in Sao Paulo, uses 'humans as sensors' to map and report problems in a city. Problems include broken sidewalks or streetlamps. It can easily be tailored to run in different cities.
Provide aspect-oriented deep sentiment mining, multi-dimension product comparison and competitive analysis by leveraging large-scale heterogeneous social content, which helps enterprises to enhance their capability for business and product management. The key techniques include:
- Unsupervised domain aspect categorization with latent semantic association
- Aspect-oriented Sentiment Association Model
- Multidimensional sentiment summarization
Evacuation Planning in the Dandenong Ranges
The Dandenong Ranges have been identified as region with significant vulnerability to bushfire. This is due to the bushland surroundings, the high population, its popularity with tourists, and the low number of roads in and out. In this study we investigate evacuation traffic, and the impact of establishing designated refuges sites within the at-risk area.
Expediting expertise (EE) is a solution that provides individuals with objective expertise scores in specific topics and a roadmap of personalized recommendations to increase expertise. It uses a data-driven approach, analysing online content and categorizing it into relevant topics. EE then compares automatically created user models to models for an expert and visualizes the recommendations using an intuitive and interactive panel of tasks. Each recommended learning activity is personalized for the individual by showing the estimated difficulty and learning gain as part of the visualization.
Healthcare and Health Systems Hub
The Healthcare and Health Systems Hub provides a platform through which data are collated, integrated and analysed. Advanced analytics are employed to address key questions, coupled with an interactive user-interface that provides informative visualisation of results. Here we demonstrate how the hub could be used to help relevant authorities plan intervention programs targeting malaria or diarrhoeal disease.
This demo showcases a new generation of cognitive systems with advanced textual and visual reasoning abilities that helps clinicians in their clinical decision making. It is envisaged as a new layer above PACS and EHR combining imaging and clinical data to provide a variety of new services such as holistic case summaries, coincidental diagnosis, automatic triage, second opinion, and, statistical comparisons to similar patients. We will show several applications of the Medical Sieve demonstrating how differential diagnosis evolves with progressive accumulation of clinical data and image analytics. The demo is built on advanced multimodal analysis methods of Medical Sieve and iMARS systems.
Predictive Traffic Management for Smarter Cities
This demo shows a traffic management solution that supports decision making of city operators. It consists of 4 technology elements that are weather simulation, traffic simulation, social media analytics, and what-if simulation. By using these technologies, the city operators optimize extra bus scheduling and traffic control to help people go home smoothly from the big event.
Smarter Financial Life
Smarter Financial Life aims to provide users with a way to consider the likely impact of their financial choices and how these affect their life choices by examining financial planning from an individual’s point of view and visualising results in an understandable and glanceable way.
TwitterViz is a series of visualizations showing text analytics applied to the Twitter stream. The goal of the solution is to support sense-making of what is happening in real-time in the TwitterSphere. The solution supports key word search, and applies sentiment analytics and text analysis techniques to help people understand what is being said and felt about this topic in Twitter.
uReport: UNICEF Uganda
UNICEF Uganda has developed an SMS text message system called U-Report that allows over 200,000 Ugandan youth to participate in surveys touching on all aspects of their daily lives. In addition to poll responses, the system now receives over 50,000 unsolicited text messages a month – about 6% of these messages require immediate action. IBM Research has developed a supervised text-message classification system trained against 54,000 messages labeled by UNICEF volunteers. This model has been deployed on the UNICEF server in Uganda, and text messages are being routed to specific UNICEF agencies as well as to 388 members of the Uganda parliament based on the geographical location of the originating message.