Machine Learning, Text Mining and Archives

Argamon and Olsen

This article from 2009 introduces the challenges faced by humanities scholars of dealing with voluminous digital/digitized sources, or the “global digital library.”  These authors believe text mining is complementary to traditional textual analysis.   They used predictive text mining, comparative text mining, and clustering/similarity analysis on three different projects.  In two cases, the digital tools supported the original, traditional scholarly conclusions, and one case found new connections previously unnoticed by traditional scholarly methods.

I thought the warnings were interesting.  You could create the results you want to find based on the construction of the task itself, and there is anxiety inherent in doing criticism by algorithm.   I think these warnings are born of normal professional concerns for historians, whether digital tools are used or not.  I mean, is the evidence there, or are we forcing the evidence to fit our argument?

Kramer

I did not know there were any issues between historians and archivists.  Kramer’s article makes it sound like historians take archivists for granted, and that historians think the archive is for their use alone.  I don’t like the sound of that, but the article is from summer 2014, so apparently it’s a thing.  I agree with his point that digitization will bring the archival and history professions closer together.  I like the idea of digital historiography and all the possibilities that digitization opens up for historical transparency, accessibility and openness.  I think historians and archivists all need to reconsider their relevance in the digital age, so this kind of discussion makes sense.

The history of historical inquiry could become much more interesting than reading traditional historiography.  To follow intellectual journeys in the digital world sounds more fun than reading dry bibliographies to me.  I think this is what Kramer meant when he said we could dynamically link primary sources and their subsequent interpretations.