Re: Benchmark transcription file

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Bruce,

> That's why I think it makes sense to have a "variorum" version like the EVA
> interlinear file and a "benchmark" file for analysis. (Of course, if you have
> serious disagreements in interpretation, you could have more than one 
> benchmark
> file, but something less than a different one for each analysis/analyzer 
> would be
> nice).

For a benchmark file, the EVA transcription style is well enough thought 
out that you can remap to other transcriptions from it without huge 
difficulty: so that's pretty much OK.

The interlinear file is a set more of interleaved interpretations than of 
interleaved transcriptions.

The only question, then, is to what degree (and at what stage) do we 
resolve ambiguous characters between those interpretations in order to 
create our benchmark file?

We could introduce a mark into the benchmark (like "~") to indicate "next 
character is ambiguous" - some scripts/filters/rules could then include it 
in, others could remap it to "*".

Or - if we still have access to the OCR scans separated by character - we 
could simply vote on the most contentious ones to form a unified consensus?

Or we could simply agree which single transcription to lock to (and just 
get on with it)?

Cheers, .....Nick Pelling.....