"Reading the Deeds and Sasines: Irish and Scottish Land Records" was sub-titled "a digital humanities approach" which meant that I was introduced to the term "word bags" and can't wait to see how this technique helps to improve genealogical research.
Dr. Patrick Walsh of Trinity College Dublin and Dr. Andrew Mackillop of the University of Glasgow are leading a project that seeks to apply emerging technology to the problem of both finding and unlocking the valuable data that is contained in historical land records. These records are complex beasties that challenge us with their old-fashioned hand-writing and byzantine legal phrasing, including flourishes of Latin. I was also somewhat heartened to hear that it's not just me who struggles to make sense of what is actually going on in some of these deeds - is the land being sold or leased or mortgaged or what? The archaic legal language used does not always consistently describe the same transactions in exactly the same way.
By "training" an automatic transcription tool (Transkribus) and applying natural language processing techniques to the text, they have been able to start picking out combinations of similar words or phrases ("word bags") that more consistently identify the type of legal and financial instrument that is being described.
This project sounds really exciting and I really hope that the results can be turned into a new finding aid that helps us all access these great records more easily and understand more about what we find in them.
The PRAI promised to make the talk available on their YouTube channel but the link is not there yet. When it is, I'll add it to my Registry of Deeds - History & Future page.