Efficient remote image-based situational queries through mobile devices

This paper presents ThereNow, a LBS[Junglas and Watson 2008] mobile application designed to get close-to-real-time answers for situational queries about real-world locations. Two key issues in this scenario are: extracting information from existing data to answer user queries; and easily acquiring more data or information, if it doesn't exist yet in the system. This can be problematic due to the format and semantics of the data or to the cost (time or resources) of collecting it. ThereNow takes a unique design approach where it relies on images/photos as data and in the actual users looking at those images to 'see' if they provide enough information to answer their queries. This approach can both bypass the difficulties in information extraction from data available on the Internet and make collecting more data as easy as just taking a picture. Thus, by leveraging mobile phones being everywhere and the "an image is worth a thousand words" effect, users can easily request and quickly receive information about what is happening now at a certain location. Moreover, ThereNow makes use of an image crawler to bootstrap the system with location-tagged images and utilizes computer vision techniques to extract additional potentially useful information from images.