Descriptive Semantic Annotations for Science Media Repositories


Inspired by the grammatical structure of common, everyday English, we develop a structured tagging language around nouns, verbs, adjectives and adverbs, for describing the content of multimedia resources in terms of the objects and interactions depicted. It has better expressive power than prevailing free-text-based methods such as captions or keyword tagging, enabling more detailed descriptions and more specific queries, and thereby improving the quality of search results. The language structure is defined in RDF Schema, providing extensibility as well as interoperability with existing semantic web technologies, while achieving greater ease of use and understanding compared to complex ontology-based approaches, thanks to its inspiration from natural (human) languages. We believe that large repositories of science data and multimedia can benefit from this combination of structured representation, expressive power, and user friendliness, making data more readily accessible, discoverable and usable.