Quantcast

An MPEG-7 query language and a user preference model that allow semantic retrieval and filtering of multimedia content

Research paper by Chrisa Tsinaraki, Stavros Christodoulakis

Indexed on: 30 Jun '07Published on: 30 Jun '07Published in: Multimedia Systems



Abstract

We present in this paper the MPEG-7 Query Language (MP7QL), a powerful query language that we have developed for querying MPEG-7 descriptions, as well as its compatible Filtering and Search Preferences (FASP) model. The MP7QL has the MPEG-7 as data model and allows for querying every aspect of an MPEG-7 multimedia content description. It allows the users to express the conditions that should hold for the multimedia content returned to them regarding semantics, low-level visual and audio features and media-related aspects. The MP7QL queries may utilize the users’ FASP and Usage History as context, thus allowing for personalized multimedia content retrieval. The FASP model supported is compatible with the MP7QL and has the model of the standard MPEG-7 FASPs as a special case. The proposed FASPs essentially are MP7QL queries. Both the MP7QL and its compatible FASP model allow for the exploitation of domain knowledge encoded using pure MPEG-7 constructs. In addition, they allow the explicit specification of boolean operators and/or preference values in order to allow both the combination of the query conditions according to the user intentions and the expression of the importance of the individual conditions for the users. The MP7QL query results are represented as MPEG-7 documents, guaranteeing the closure of the results within the MPEG-7 space. The MP7QL and the FASP model have been expressed using both XML Schema and OWL syntax. An implementation of the MP7QL, on top of an XML Native Database is currently in progress. A real world-world evaluation study on the expressive power of the MP7QL shows that it covers both general purpose and domain specific requirements in multimedia content retrieval.