Multi-dimension multimedia high-level semantic structure and locating algorithm

Derived from the thought of data cube, this paper presents a new structure model which organizes multimedia data in a multi-dimensional data cube. For the problem of semantic gap in retrieving multimedia information, the paper gives the definition and operation laws of matrix division method, which describes the high-level semantic structure for a lot of kinds of multimedia information. We retrieve multimedia information by using the high-level semantic structure to solve the problem of semantic gap. Based on the structure, the paper presents a partial priority algorithm by the results users locate multimedia information. The locating efficiency is improved by the algorithm based on the structure through modifying the high-level semantic branch weight by every result users locate multimedia information automatically.