Digitizing Historic Broad Pen Writing

[Cast]

The Atlantic has an interesting article on the problems of digitizing material in the Vatican Library.

Specifically the problems of dealing with broad pen texts and the similarity of letters.

 

 

https://www.theatlantic.com/technology/archive/2018/04/vatican-secret-archives-artificial-intelligence/559205/

[Randal6393]

I have, over the years, worked with the Vatican Library off and on. So I pity the computer, as it is difficult to say the least. Scholars everywhere should salute this advance and feel it holds out hope for the future.

 

Enjoy,

[inkstainedruth]

Very interesting article. I may forward the link to my husband, who has worked on AI projects, as well as natural language processing, and -- at one point -- for a company that did software for high end printing (think art books).

Ruth Morrisson aka inkstainedruth

[dcwaites]

It seems to me that the techniques being taught to the AI are pretty similar to those used by people, i.e. -

• read word by word
• use context to predict what the next word is likely to be
• keep a dictionary of real words in mind
• if all else fails, use stroke identification within the above limits

/begin pot stirring

next we'll be teaching Phonics to the AI

/end pot stirring

[amberleadavis]

This was cool. Thank you for sharing.