In December the ADAPT team travelled to India to take part in the Forum for Information Retrieval Evaluation (FIRE ) 2016 conference where they too part in the FIRE 2016 “Information Extraction from Microblogs Posted during Disasters” task. Competing against a number of international entrants, the ADAPT team of Wei Li, Debasis Ganguly and Gareth Jones took first place in the competition.
The task required participants to mine an archive of tweets captured during a natural disaster to identify those containing information describing details such as resources needed by the population in the disaster, the affected area, what resources are available, what resources are required / available in individual geographical regions, etc., based on search topics describing the information needed in general terms. The key challenge of the task is to identify the relevant tweets since they are very short and use highly variable informal language describing specific situations, using the information needs that are stated in more general terms.
Gareth Jones was also an invited speaker at the FIRE 2016 workshop. He gave a presentation on the evolution of the automated Video Hyperlinking task that he has co-coordinated at the MediaEval and TRECVid evaluation benchmarks since 2012.
Share this article: