wlpotter
commented
1 year ago Hmm I might have uncovered a second problem:
We have several cases of the same work being referred to separately. And since we don't have stable URIs for all authors and works, I have had to rely on string matching to create the author-work groups for sorting. Here are the offenders as of 2023-02-24:
Eusebius of Caesarea. Dictionary of Place Names Eusebius of Caesarea. Ecclesiastical History Eusebius of Caesarea. Historia ecclesiastica Eusebius of Caesarea. Life of Constantine Eusebius of Caesarea. Martyrs of Palestine Eusebius. Dictionary of Place Names Eusebius. Life of Constantine Photius. Library Photius. Library Library Pliny the Elder. Natural History Pliny. Natural History Procopius of Gaza. Letter Procopius of Gaza. Letters Socrates of Constantinople. Church History Socrates Scholasticus. Ecclesiastical History Sozomen. Church History Sozomenus. Church History
We will need some data normalization before we can reliably run an updated string sort. @davidamichelson we should discuss.
The string sort is not essential, and ideally we'd just run it once more before official publication, so this is not a priority.
davidamichelson
josephrife
Now that we have new data we need to update so that the records ahve the string sort attribute letting records within the same work appear properly sorted w.r.t. the section ordering.
Add a test to this script so that it only runs on work-groups that have at least one record that lacks the n attribute. This will cut down on time re-processing unchanged work groups and, more importantly the time spent manually checking the ones that couldn't be sorted by the matching algorithm.