Modular Content and Segmentation

Modular content is media content (audio, audiovisual, text) that can be easily repurposed, reused, assembled in different ways.

Modular content can be described as bite-sized, semantically coherent and metadata-rich.

Semantically coherent (atomized) content can make content repurposing, access, recommendation and search much easier. By reducing down and richly describing content, it can more precisely be discovered and suggested. This allows for some new routes for consuming content:

• Content headline navigation
• Adaptive content aggregation
• Highly-relevant recommendations,
• Similar content search
• Content indexing

How content can be modularized depends on the production workflows for each media organisation, the existing availability of metadata, on the type of content and other factors. For example, archive content can be more effectively modularized after metadata extraction, analysis and segmentation using manual, assisted or fully-automated techniques.

The group has regular meetings in which member share updates, proposals and new ideas. Part of the meetings are dedicated to workshops on specific themes, where the group identifies possible lines of collaboration or information exchange.

The group has the following guiding principles:

• Share knowledge about process, technology, tools and research needed to create and leverage modular content
• Share modular content use cases and services

With the following technical goals:

• Applying (semi) automated segmentation techniques (prototypes, POC)
• Programme start/end detection
• Music (and commercial content) detection and filtering
• AI applied to segment detection leveraging on video, acoustics and data-fusion features 

Related topics covered by other groups and activities include: ASR, topic extraction, leveraging open data sources