Semantic Transcoding

Semantic transcoding is the transformation of web content using explicit semantic information about the structure, role, or meaning of page elements — typically supplied through external annotations, ontologies, microformats, or ARIA. Because the transformation uses real knowledge about what each element is for, semantic transcoding produces more accurate, useful output than purely heuristic approaches: it can confidently move the main content ahead of a sidebar, or strip navigation aliases without accidentally removing the actual navigation. The cost is the annotation work itself, which has traditionally been done manually and therefore scales poorly. Systems like SADIe (Lunn, 2007) reduce this cost by annotating CSS-class-level ontologies once per site rather than once per page.