This afternoon I attended Laura Crossley’s presentation “Text MiningDigital Humanities with API’s, OpenRefine, and R” 1. When Laura started her presentation I was worried that I wouldn’t understand much as I am pretty new to Digital Humanities. As a historian I am much more comfortable with centuries old paper than I am with any of the software she described. Her project was using DHNow, a digital publication originally intended as a case study for “PressForward”, and trying to describe its usefulness and its relevance to future digital humanities projects. She used DHNow’s Editor’s Choice blog as a corpus and was determining how much they are used currently. She noted that it’s number of blog posts has dropped in recent years, but argues that it is still useful and worthwhile project. One of the larger questions Crossley was investigating was whether or not blogs in general could be considered a useful scholarly source, the issue being that over time links break and are no longer a valid pathway to the information in the blog, making citations increasingly difficult.
One really interesting aspect of her presentation was seeing the way that she used tools to analyze the DH Now corpus. During our lecture yesterday the Voyant tools were really intimidating. I was unsure of how this type of text analysis tool would be used in real world projects. I could understand it’s usefulness in the scope of our assignment but not in a larger real world sense. Seeing this presentation changed that for me. Seeing the application of these tools to formulate questions was great. One thing that Crossley mentioned when she was discussing how she used topic models was that they are much better tool for forming questions- than for answering them. This was something that I missed when Professor Jess 2 was walking us through the different text analysis tools, so I was really glad that I attended the presentation. It was informative to see that these tools do more than make visually impressive word art. I was impressed with how she was able to use text mining tools to sift through mountains of data and find usable information for her investigation.
