2017-02-08 - Questions 6 (Dylan)
DigiPal seems like a cool project. I’ve been thinking about automated transcription for a while (using computer recognition of paleography) and just figured it wasn’t yet possible. Stokes gives some reasons for why it isn’t–for instance, medieval manuscripts are really quite irregular. Even on formatted and more modern printed texts, OCR still requires human verification most of the time, so for medieval manuscripts it’d make all kinds of mistakes(I imagine)–hence the need for DigiPal and drawing boxes around a lot of letters. That said, will we ever reach the point where automated transcription is possible/viable?
Fairly straightforward; IIIF arose out of a need for interopability(it’s right there on the tin), because image repositories were “walled gardens of technology, with institutions implementing similar solutions in vastly different ways with few mechanisms for easy exchange of data and little sharing of code or methodology” (1). I wondered about how progress is made. There’s a popular idea that progress requires some type of destruction–growing pains, revolution, etc.– but these types of projects seem to necessitate non-intrusive progress, where older systems can remain compatible. What different things does one have to consider when approaching progress/improvements this way?
Basically, this reading highlighted the considerations of being able to document even the oddest objects (use cases). It seems so simple, yet very important, that we have to be able to record information even about the lack of a page. Also, how does one even go about transcribing a palimpsest?