How semantics can improve media production

Karel Braeckman
VRT-medialab
karel.braeckman(at)vrtmedialab.be

Metadata creation at pre-production

All too often, the creation of metadata is postponed until the last step of a media production project, which is archiving. However, as lots of valuable descriptive metadata is already created even before the media is shot – e.g., the script, the set list, the characters and actors, camera instructions, etc. – MediaMap takes a radically different approach and uses prescriptive tools to capture highly structured, semantic metadata from the start of a media project.

Figure 1: CameraMate

Guiding the camera operator

The pre-production step resulted in highly structured metadata, describing among others the camera missions, i.e. what has to be shot. By combining this metadata with a media ontology based on the AXIS framework, instructions on how to shoot every shot can be automatically generated. This is done with the Semantic Camcorder and the CameraMate created in the MediaMap project. The user can download a camera mission onto the camera, and depending on whether the user is an amateur or a professional cameraman, the camera will display tailored instructions. The user can finally upload the created media material to subject it to further semantic treatment.

Transparent production archive

Material shot by using the Semantic Camcorder and CameraMate can automatically be linked to its prescriptive metadata, which effectively becomes descriptive metadata now. This metadata is enriched with metadata originating from automated media analysis tools such as speech recognition and shot or speaker segmentation.

To be able to collect and reconcile all metadata and their producers during the production process of a specific mission, taking into account the mission state evolution, it is necessary to manage a knowledge base which gathers prescriptive and carried out elements. It relies on an ontology which plays the part of the roots and the trunk of a tree, structuring metadata along its branches. It is a semantic network linking all rich media information provided by the sweat of its contributors.

Figure 2: Semantic modelling

Conclusion

The MediaMap project clearly demonstrates the impact of semantic metadata on media production. By employing highly structured, semantic metadata from the start of a media project, it is possible to improve several stages of the media production chain.

The semantic modelling improves the pertinent search of media material and the navigation within a media item by several orders of magnitude. As time-coded metadata is employed, the user can actually browse within the video search results, which can save a tremendous amount of time jogging within the video.

Further information is available at
www.mediamapproject.org/project.html