Karen Coyle> I've been wondering if we shouldn't reverse the markup, and indicate which elements are precisely transcribed. In effect, everything else is supplied.

Very good idea!

If the W3 Provenance standards were used, would not precise transcription have the same semantics as "was quoted from" ( http://www.w3.org/ns/prov#wasQuotedFrom ), with "supplied" values being roughly described by "has primary source" ( http://www.w3.org/ns/prov#hadPrimarySource )?

[I use the term "supplied" to avoid confusion since both relationships described above are specializations of the "was derived from" relationship - ( http://www.w3.org/ns/prov#wasDerivedFrom ).

A further comment on the literal transcription of text: one of the main reasons for creating cataloging rules requiring literal transcription is to improve the accuracy of record matching*. However, simply transcribing structured textual information does little to unlock the potential benefits of using linked data.

For example, If dates are only transcribed textual strings, one cannot aggregate such information from all the items in a collection to compute the range of dates covered, bulk dates.

A very readable analysis of some of the issues involved in item to collection metadata aggregation can be found in the work of Karen Wickett -

http://cirssweb.lis.illinois.edu/News/newsDetails.php?id=74

Simon