基于本体的G3M语义增强策略

An ontology-based strategy to enhance G3M semantics

针对G3M是1个基于多媒体内容描述接口MPEG-7的通用多媒体数据库模型,能充分利用多媒体内容描述接口(MPEG-7)在多媒体元数据描述上的长处,对多媒体数据进行有效的组织管理,但MPEG-7缺乏规范描述基础,导致描述结构含义不明确、描述内容中存在大量隐含语义信息等问题,引入知识本体的方法来增强G3M的多媒体语义表达和检索能力。研究结果表明:将本体与MPEG-7元数据及多媒体实例无缝关联在关系数据库中,能为G3M多媒体描述结构连接正确的语义,增强其高级语义提取、表达和检索能力;将基于本体的查询建立在经典的关系查询引擎上,能有效地提供基于语义的查询重写和查询推理。

Based on the fact that G3M is a generalized multimedia database model based on multimedia content description interface（MPEG-7）,which can manage multimedia data effectively with the full understanding of MPEG-7 metadata,but MPEG-7 has no formal description grounding,which leads to the vague use of its structure and implicit semantics in its content,a knowledge-assisted approach on G3M was presented for better representing and querying multimedia semantics.The results show that ontologies are unified with MPEG-7 metadata and instances seamlessly in relational databases,which can describe associated valid semantics to G3M schemas,and enhance semantic abstraction,representation and retrieval ability.Ontology-based queries can be built on the basis of classic relational query engines to effectively support query rewriting and inference for truly query-by-semantics.