LIST develops a collaborative annotation platform

LIST researchers worked for four years with Spanish, French and Turkish partners on innovative methods to help annotate multimedia archives. Selected in line with the topic of the European CHIST-ERA call for 2011, "From Data to Knowledge", and financed in Luxembourg by the National Research Fund (FNR), the CAMOMILE (Collaborative Annotation of multiMOdal, multILingual and multimEdia documents) project led to the realisation of the first collaborative annotation platform for 3M (Multimodal, Multilingual and Multimedia) data. The Camomile platform uses automated assistance tools, making it possible to manage annotation work remotely via a computer network and to locate the validated versions of annotated documents in a centralised way. 

A unique collaborative annotation platform 

The Camomile collaborative annotation platform consists of a semi-automated annotation infrastructure and a visual collaborative platform developed by LIST. It is characterised by an interactive user interface for managing and overseeing the entire process of collaborative annotation work. The platform is based on state-of-the-art web technologies, and its construction follows a client/server model. While the server focuses essentially on data and authentication management tasks, the client provides overviews and details of the history of the contributions made by the individuals annotating the 3M data, as well as ongoing and resolved conflicts, whilst the type of contributions is primarily LIST's responsibility. Thus the platform takes into consideration many different modalities that can be obtained from videos in a non-supervised way. As envisaged in the project definition, the code for the Camomile platform is provided on an open-source basis, and the entire application can be found at https://github.com/camomile-project.. 

Research issues resolved 

The platform developed in this way makes it possible to respond to research problems by displaying and analysing non-structured and continuously transmitted data. In fact, its flexible visual interface has provided an advanced feasibility demonstration to support collaborative work on multimedia data. Therefore, the platform constitutes an ambitious response to the need of multimedia communities to exploit the wealth of 3M archives. In addition, the visual data search approach developed by Camomile can be reused to evaluate various classification algorithms. 

New opportunities for the future 

This project has now come to an end, but constitutes only the first step in the development of a technology-assisted framework for collaborative annotation. Three follow-up projects are already on the drawing board. The first involves an enhanced version of Camomile with more analytic processing to assist collaborative work, whilst the second involves reusing the platform to track narrative arcs in multimedia documents. Finally, the Camomile methodology could be applied to the display of spectral lexical analysis algorithms. 

>> To find out more about the CAMOMILE project, visit the dedicated page.

>> To find out more about the Camomile platform, visit https://github.com/camomile-project. 

Below, a printscreen of Camomile software platform-