Standards and Metadata

The film industry uses a large amount of information for decision-making and reuse of existing material. However, it lacks techniques and tools to enable the effective organisation, discovery and reuse of such information. In November 2006 we interviewed AFTRS staff to gain insights into current work practices and requirements, and to assess the scope for innovation. Based on these interviews, we developed a number of use cases for the pre-production, production and post-production stages. These will be used to evaluate the methodology and software system.

Overview
Plans for 2008

We plan to develop a number of context-sensitive ontologies. For example, the Agent ontology identifies the types of people involved in the pre-production and production stages (producers, actors, concept artists, director of photography, etc.), while the Abstract Concepts ontology models actions that are required to achieve certain abstract concepts such as style and look. An example is an ontology for cross-cutting.

A number of algorithms will be constructed and implemented for information classification, query translation and information retrieval. They will be evaluated using the use cases we have already developed. Papers will be written on these aspects during the year as the work is completed.

Katie Shortland, an AFTRS staffer, has joined the team to facilitate the communication between AFTRS staff and the research team.

We will also explore collaborative opportunities with other film agencies and/or schools, to obtain access to at least one more film collection for the overall evaluation of the techniques and tools.

Progress in 2007
In 2007, our work focused on developing conceptual models, and implementing the core system and sample plug-in modules for the post-production stage (e.g., film distribution, festival entry).

A new information model was constructed which captures the nuances of the motion picture industry. This was then used to develop a set of data models describing the relationships between various forms of motion picture industry data. A metadata-based framework was devised that enables the ingestion of metadata from an existing system (e.g., AFTRS metadata database of short films), and the management and transformation of such metadata for different re-use purposes. To provide proofs of concept, we designed a lightweight Loculus metadata wrapper schema, and implemented the core module and plug-in modules for the film festival entry process.

To enable effective extraction of relevant information from diverse sources through queries expressed using the industry vocabulary, we have devised a detailed plan for developing context-sensitive ontologies which underpin the inferencing process, allowing new information be deducted from existing information in a meaningful way.

A PhD student, Tinni Choudhury, has successfully completed the requirements for the confirmation of her candidature. Two research papers were presented at the 11th IASTED International Conference on Internet and Multimedia Systems and Applications, Honolulu, Hawaii, in August 2007.