Cost Effective Meta-Data – From Stefano Mazzocchi [via Seb] [Paraphrasing] The more, higher quality meta-data the better, because the quality of meta-data heavily influences the effectiveness of retrieval systems. Meta-data mark-up-based “semantic” web solutions (RDF, Ontologies) are economically infeasible (because the human assignment of the right tags is in addition to defining the content itself and is either expensive to do properly or compromised by cutting corners.) The best semantic meta-data solutions are based on transparent inference, without heuristics (in publishing the content). User feeling is important (in creating meta-data, and in how it is perceived as relevant later). [End Paraphrasing – my italic emphasis]
IMHO – The quest is for some kind of ontology that captures (or infers) the creator’s actions and intent at the point of creation and/or publication, not some misleading post-rationalisation according to some fixed prescribed ontology. ANKOO – A New Kind of Ontology ? Maybe not, “AnOint” perhaps – An Ontology of Intent ? Has your blog / output been annointed yet ?
If the human assigns meta-data to their own output, the “intent” problem just shifts along one – from why did Mr X say that, to the question of why did Mr X assign that meta-data to it. When the Mr X says his latest offering is the best thing since sliced bread, do you believe him because it is, because he says it is, or because you believe him (or not as the case may be) ?