Top N MM query optimization - The best of both IR and DB worlds
Henk Ernst Blok, Arjen P. de Vries, Henk Blanken, Paul M. De Bra, L. Hardman · 1999
It is commonly recognized that top N queries belong to one of the most important query classes in IR and MM retrieval, or more general, content based retrieval. A lot of work has been done on query optimization in database research but that research has mainly focussed on the area of optimization of databases in a business application environment. In IR research also work has been done on query optimization but this is not directly applicable in a database environment. Exploiting the integrated content based retrieval technology in the mi R Ror database system, we intend to look into optimization of top N queries in MM DBMSs. This paper provides a brief overview of current IR and database technology relevant to top N MM query optimization. It also describes our DB environment and ideas on three major sub topics in the light of top N MM query optimization: incorporate known fragmentation techniques to ensure scalability, introduce a new intermediate optimizer layer that suppo...