If there’s something strange in your literature… who do you call? Gale Digital Scholar Lab! This week it’s time for text analysis.
After playing around with Gale I feel pretty confident teaching it to someone with the provided vivero documentation, on-site documentation and tool tips. To me, it is a very straightforward tool in its function. I think the main cause of frustration will be that the site can be clunky and slow, especially when needing to index or analyse a large data set. Another piece of pain is how the OCR (Optical Character Recognition) deals with non-standard fonts or god forbid strange formatting. I found many cases where the text is fully human readable but the OCR completely freaks out at borders and decorative elements. Also Gale is not good for using library resources from different languages, especially those that don’t use predominantly characters from the English alphabet. When I searched for it, I found that the library actually has a vast amount of Chinese texts but Gale’s OCR makes the generated plaintext virtually unusable without retyping everything. This makes me very sad knowing the texts are there but the analysis tool isn’t capable enough to properly search through or use them.
I find the different algorithms used for analysis really interesting especially the “Sentiment Analysis” and “Topic Modelling”. Because who gets to determine what is grouped into what and what happens when modern language usage groupings gets mapped onto old archives? I would look further into how the analysis algorithms work but I learned that may be a years long (possibly futile) learning journey.