I’ve mentioned in the past that I have devised a methodology whereby I utilise Deep Neural Networks1 to analyse a dataset of manuscript images. The dataset presumes that the same scribe produced the handwriting in the fragments, but it does not necessarily assume this. The algorithmic process is quite helpful to rule out whether a fragment belonged to the same scribal hand as another group of fragments. As part of my dissertation at the University of Toronto, my method has helped discover some very interesting things about the Community Rule manuscripts at Qumran.
I have spoke about and presented on the use of Schriftenmetric in several informal and formal settings.2 I have really enjoyed talking about my method in the formal settings, especially my talk at Yale, a guest lecture in a Alex Jassen’s Dead Sea Scrolls course at NYU, and at Birmingham.
So what is Schriftenmetric? Here is my characterisation of Schriftenmetric:
Here is an visualisation I made for an SBL presentation, where the Hebrew dalet is being processed:
I’ll write more on Schriftenmetric in the coming days. It turns out that reconstructing fragmentary manuscripts is near impossible to do without a Schriftenmetric.
- I have found that Fastai is producing the best results for my purposes. I will write a series about using Fastai in Ancient Jewish Studies in a future series.
- I’ve spoke about it twice in Göttingen, once on February 16th, when Jonathan Ben-Dov was invited to a Göttingen Scripta Qumranica Electronica team meeting. The second time was also in Göttingen, when Sacha Stern presented in the Doktorandenkolloquium on January 14, 2020. The third was at 10th International Meeting for Qumran studies, in Aberdeen.