jsLDA, Bookworm, and Storymaps

[advanced_iframe src=”<iframe src=”https://storymaps.arcgis.com/stories/24a914688f0f44fb97d5da9e3fd32f96″ width=”100%” height=”500px” frameborder=”0″ allowfullscreen allow=”geolocation”></iframe>” width=”100%” height=”600″]

I LOVED Storymaps, and I knew I would from the start. The super basic interface reminded me of working with the blogging tools I was comfortable with, like old school Tumblr and Blogspot (the “point and add” blank canvas model.) I spent way too much time playing around with it, so I was pretty glad I saved it for last. I figured the basic stuff out pretty quickly, and learned a few of the quirks of the website (hard to resize photos, no font size adjusters besides paragraph/quote/heading styles). With more time, I’d like to watch/read more tutorials on how to make more complicated storymaps, possibly using it for my project.

The first thing I tried, however, was jsLDA. After clicking around with the software using the sample texts, I went looking for plain text on Archive.org to plug into the software. I decided to use this 1891 issue of “The Police Gazette.” I took the plain text from this and pasted it into a document.

Once I had done that, I uploaded it to the jsLDA site (under the “Use a different collection” sidebar).

I checked my source document again, and, while the columns in the magazine made for some wacky formatting, it was still legible enough to read. I decided to try something with a more simple original layout, which would hopefully be closer to the example text that the software defaults to state of the union speeches.

So, I uploaded a document with the plain text I copy and pasted from San Diego’s original 1889 city charter, which I also found on Archive.org. This worked better, especially when I put it through multiple iterations and got rid of some of the random unrelated phrases by adding them to the list of “stopwords” under the “Vocabulary” tab. Still, it didn’t seem to work the way I wanted it to, so I reviewed the homepage of the jsLDA website and kept trying for a while with little change. I know what the software was trying to do (find connections between words without the computer having an understanding of the meaning) but I still felt that everything came out too messy thanks to the OCR plain text recognition being scrambled by margins and lines (or, more likely, I don’t know what I’m doing.)

Next, I tried out Bookworm, which sounded slightly similar to the jsLDA site. Or, rather, I tried to try Bookworm– as of writing this (2/26), I had no success in getting the page to load. I will update when I try it in a different browser, but on Firefox it just timed out. (EDIT: tried next day, 2/27, the website says it can’t connect at all now. Tried on Chrome, also nothing.)