Media production will become increasingly a computation-intensive practice. But at the level of interface much current practice still clings to shallow representations of mechanical devices for media capture, storage, retrieval, and reproduction. The predominant use of legacy interface representations for computational media belies the potential for radical transformation of the entire production and distribution process. Computational models and intelligent signal processing are in the process of being adopted in mainstream methodologies in the media production chain. Media Authoring is a versatile practice that utilizes computational representations in a zone between interpreted languages and scripting. Authoring enables creative control of dynamic and interactive computation and display of media on a diverse range of platforms. Robust models of computational architectures for media production are becoming more feasible as consumer platforms converge on standard processing capabilities that permit extensible applications in high-level programming languages with open source libraries.

In this context, this colloquium will present a working method and prototype system to demonstrate an example of computational media authoring to produce interactive media presentations supported by an ontological inference engine. The system supports real-time query across heterogeneous media resources, parallel media signal processing, and multiple display formats. Authoring is implemented as path-planning in ontological space; path members are concept nodes that generate queries and return media resources coupled to real-time displays. Ontology supports heterogeneous cross-referential capacity; its structure is reflected in a graphical interface designed to facilitate concept-based navigation across 2D graphics and images, 3D models, video, and audio resources. A dual-root-node data design links ontological reasoning with media metadata, to provide a method for defining hybrid semantic-quantitative relationships. Ontological organization enables users to author and explore media resources by concept-based navigation displaying relationships across media of diverse types, rather than isolating resource types. This approach entails two merits: (1) concept formation takes a significant role in authoring processes and media resources orientation; (2) semantic reasoning applied to intelligent authoring in multimedia production is compatible with information requirements for context recognition applied to feature-based classification and retrieval.