Efficient Search of Multimedia Metadata on XML Database Systems sing Information Retrieval Approach

S. Qiu and S. Li (USA)

Keywords

Metadata, XML, Database, Search,Information Retrieval.

Abstract

It is becoming popular to use XML documents to model multimedia metadata. Performance of multimedia metadata management, especially search performance depends on the storage, indexing, and propoer schema design of the data repository. Search performance can be improved when information retrieval approaches are integrated into the XML database system. In this paper, we identify performance issues in a multimedia metadata system and improve search performance in three ways toward the design of a prototype multimedia metadata system (MMS). 1) We model and design the MMS using native XML database. XML databse stores XML documents directly in tree structures without going through normalization processes required in RDBMS. 2) We build index on the data repository and then introduce as assistant document, the rotated document, for each descriptor that needs be searched. We design proper inter-document schema for efficient search. 3) On the rotated documents, we build indices and integrate lexical processing to further enhance performance. We use TV program information as data sample for analysis and tests. The proposed design strategy achieves a speedup factor of 3.78 compared with the direct DOM method. It can speed up even more than systems based on RDBMS. It also has a better scalability.

Important Links:



Go Back