University of Glasgow computer scientists are developing a new search engine which can answer personalised questions, such as where your friends are hanging out.
The SMART search engine, which stands for search engine for multimedia environment generated content, will be able to search and analyse data from sensors from cameras and microphone arrays.
Making search results more personalised, SMART will also be able to answer questions such as ‘Where are my friends hanging in the city today?’ by cross referencing data from social media.
“Users will be able to find interesting events happening in various locations of the city and to also know what people are saying about these events on the social networks,” Dr Iadh Ounis, University of Glasgow’s School of Computing Science, told Computerworld Australia.
“This could be particularly useful for a city hosting large events [like] the Olympics or Commonwealth Games where there are many satellite street entertainment activities.”
SMART is a three-year project which began in November 2011 and is being built on open source technology – Terrier – which the university began developing in 2004.
Ounis said the main aim of the project was to alert users about what was occurring in the ‘real’ world by integrating real-time data.
“In contrast, current search engines and social networks can only see what is happening in the real world from the interaction of Internet users, e.g. increased queries or tweets from that area,” Ounis said.
SMART will use social media information, such as the Facebook ‘check-in’ feature, to answer questions about friends’ activities. In the event this information is not available, queries will deliver ‘unpersonalised’ results.
“The main point, however, is that even without the availability of social information, the engine will be informed by sensor evidence gathered from the real world,” Ounis said.
“Our current research findings show that it is possible to identify various events from sensors, such as crowds or traffic jams from video sensors, live music from audio sensors, etc.”
However, Ounis concedes the challenge of the project is to find locations where events are occurring in cities.
“The only [other] barrier to adoption by smart cities are the sensor infrastructure deployment costs, together with the balance between public benefit versus privacy,” he said.
“For the test deployments within our project we are carefully working with the government data protection and privacy experts to ensure that we do not overstep the boundaries of individual privacy.”
The search engine is expected to be available for trial in Europe by 2014.
Follow Stephanie McDonald on Twitter: @stephmcdonald0
Follow Computerworld Australia on Twitter: @ComputerworldAU