||ContextTag - Context-aware media annotation utilizing Social Networks
||Larsen, Jakob Eg (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Petersen, Michael Kai (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||In this thesis the design and implementation of a novel photo annotation system is presented. ContextTag is an application specifically designed to take advantage over user and social context available in social networks. The aim of the application is to support the user in the photo annotation process by providing real-time tag suggestions with the use of an algorithm. To archive this, ContextTag automatically extracts user and social context from various sources and utilizes it for its tag predictions. ContextTag also incorporates collaborative tagging, face detection and tag classification using WordNet1.
Photos are tagged at the instance they are captured and can be shared immediately from the camera phone to the user’s social network. To easily locate already shared photos a custom Facebook application has been implemented which includes enhanced browsing features.
The approach was evaluated with a user test including N=5 participants, that carried a camera phone using ContextTag for a duration of four weeks. All phone activities were recorded and an individual user interview completed the user test. Further, simple Decision Tree modeling was applied in order to determine which contexts define the breakpoints. The tag suggestion algorithm has then been optimized accordingly.
Results have indicated that generation of real-time tag suggestion from user and social context is possible by following the proposed methodology.
||Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Creation date: 2010-06-08
Update date: 2010-06-08