Enabling Technology for the Analysis of Electronic Cultural Archives

Much multimedia content is locked up in legacy formats and content management systems making it difficult to search, access, and distribute. Our aim is to enable technology to support the analysis of higher-order cultural features of multimedia content to effectively liberate this content for new and innovative use.

The main outcome of this project will be to increase the quality of access, usage and distribution of multimedia content currently found within large, tersely annotated databases (such as library catalogues, film archives, and web repositories). We plan to achieve this by integrating automated retrieval techniques used in computer science with descriptive-based metadata approach used in information science.

This will be done in two ways. Firstly, we focus on annotating the artworks using high level semantic contexts or cues found by our techniques and software prototype for object detection and semantic interpretation. Secondly, we will develop standardisations for representation schemes of digital multimedia contents to facilitate their indexing, retrieving, archiving and analysis.

To evaluate the techniques and software prototypes, we use a traditional Vietnamese Woodcuts Gallery as a case study. This Gallery, which had been developed using Macromedia Director authoring tools before the commencement of this project, currently can be searched only by keywords obtained from the narratives of these woodcuts. So far, we have developed a software system for the basic annotation and classification of traditional Vietnamese folk paintings by detecting objects in the artworks. The prototype software system can detect ‘object categories’ (such as people, cows, ducks, musical instruments) in the artworks to augment existing meta-data. Next, we will develop suitable domain ontology to produce further semantic meaning from these annotations. Symbolism, association and cross-reference rules will be included to enrich the access via semantics and context.

An investigation into expert taxonomies and the contrasting user-driven ‘tagging’ schemes has been made to determine the most effective classification systems in the context of art and culture. The focus for this part of the research will now turn to developing standardised metadata schema and domain ontology for the artworks.