DVPP project work 2023-05-08 to 2023-05-12
to : Martin Holmes
Minutes: 140
On Monday, worked on the one remaining Greek transcription and fixed some broken pointers found by the diagnostics. Also confronted quite a difficult problem: the staticSearch uses a single value for minimum word length throughout its operations, and passes this on to the feature filter constructor, which means that in a feature filter (such as the dominant rhyme filter), nothing happens until you type three characters into the box; however, there is a value (aa) with only two characters. This is now a bug ticket on the staticSearch repo, but for the moment I’ve worked around it in this project by overriding one of the functions in the SSTypeAhead class.
On Wednesday, met with AC, and we came back to the question of untagged cross-stanza rhymes, which I think fall into two categories:
- Line-ends that don’t rhyme with anything in their own stanza, and therefore should show as a dash in the line-group rhyme scheme, but should carry a label matching them with prior instances of the same ending.
- Line-ends that do rhyme within their own stanza, but at the same time reiterate existing rhymes from earlier in the poem.
I’ve written a module to auto-tag the latter, which can be done, but first we need to specify what the maximum separation is between lines in a poem before we can/should tag them as cross-stanza rhymes. However, I will rewrite it so that it only produces output when poems match the criteria, and it puts the output somewhere else where it can be validated before being copied back, to reduce svn changes.
On Friday, I rewrote the module to create output only for files which meet the criteria in a temp folder, and the approach proved very successful, so I will use it again for this sort of thing rather than transforming in-place. 1212 poems had cross-stanza rhymes tagged as a result, with over 18,000 instances of it being tagged.