Compressing semantic metadata for efficient multimedia retrieval

Abstract

The growth in multimedia production has increased the size of audiovisual repositories, and has also led to the formation of increasingly large metadata collections about these contents. Deciding how these collections are effectively represented is challenging due to their variety and volume. Besides, large volumes also affect the performance of metadata retrieval tasks, compromising the success of multimedia search engines. This paper focuses on this scenario and describes a case study in which semantic technologies are used for addressing metadata variety, and advanced compression techniques for dealing with the volume dimension. As a result, we obtain a multimedia search prototype that consumes compressed RDF metadata. This approach efficiently resolves a subset of SPARQL queries by implementing representative multimedia searches, and also provides full-text search in compressed space. © 2013 Springer-Verlag.