Efficient Query Refinement in Multimedia Databases

Increasing application demands are pushing database management systems (DBMSs) towards providing adequate and efficient support for content-based retrieval over multimedia objects (e.g., images, video, audio, time-series, spatial and spatio-temporal data). Recently, several powerful models for multimedia similarity retrieval have been proposed. An important aspect of these models is the notion of query refinement: a technique that allows the users to interactively specify their information need to the system by providing relevance ranking on example objects. Query refinement has several motivations. First, the `starting' query may only partially capture the user's information need. The user may find better examples among the answers returned to the starting query which then become the basis of the `refined' query. Second, multimedia objects are represented as a collection of features. The relative importance of these features in computing the similarity between objects ...

• Type: Conference
• Year: 2000
• Where: ICDE
• Authors: Kaushik Chakrabarti, Kriengkrai Porkaew, Sharad Mehrotra
• Keywords: Database, ICDE 2000, Multimedia Objects, Query Refinement, Similarity Retrieval