Building a scalable framework for the collaborative annotation of real time data streams

With the fast development of Internet technology, competent on-line collaboration tools are currently being used daily to improve group productivity. There are no longer such limitations for people to work on digital contents using designated platforms on specific networks. They can choose different collaboration services with diverse connectivity and user interface options. As an important feature of on-line collaboration systems, collaborative annotation on real time streams is being thoughtfully investigated and studied by different research entities all over the world. Systems such as Vannotea [Schroeter et al., 2003] and SIDGrid [Bertenthal et al., 2007] have been invented for common collaboration requirements such as audiovisual communication and digital annotation, and they have accomplished the targets very well. However the majority of such systems are designed and implemented to process merely specific data such as multimedia streams, which can be hardly extended to support generic contents such as real time data from earthquake sensors, traffic monitors and medical instruments. It is challenging to design and develop such a framework that supports creating, sharing and replaying annotations on generic data streams regardless of end user's multiplicity of connectivity and supporting platforms. In this dissertation, we investigated major characteristics of popular collaboration and annotation systems on both desktop and mobile platforms, summarized key requirements and research difficulties of building a distributed collaborative annotation framework, and then presented our prototype of such a system that supports annotating generic data streams in heterogeneous environments. The analysis of experiment results demonstrates that our decisions on the system architecture and design have provided various advantages over previous systems on both performance and scalability.