Practicum 13; Working with algorithms

This first image is the result given when I typed “Simulacra are” in the Talk to Transformer algorithm website.


These are the topic model results from putting Ulysses into Voyant tools.


For this week’s practicum, we explored the use and function of algorithms in the Digital Humanities through Voyant Tool’s document reader and the Talk to Transformer website. I can honestly say I had a lot of fun deciphering the methodology and logic behind the two tools introduced in class. For Voyant, there always seems to be something new to discover about its functionality and how it can understand a text. Likewise, the Talk to Transformer website presents a similar sense of intrigue despite seeming innocent and simple at first. However, as I continued to explore the results it spat out, my curiosity urged me to see what other possible combinations of words could be rummaged onto my screen. In both cases, I was presented with a more subjective glimpse of data regarding a text or topic. I then had to bring myself to explore the meaning behind the algorithm’s presentation, which brought a lot of insight regarding the benefits and limitations of such tools.

One benefit involves what I mentioned above regarding viewing a body of information. By presenting a unique grouping of information as a consequence of an algorithm, the act of interpretation and deciphering importance of that information are placed into my hands. I am granted a larger degree of agency to understand that information as compared to other more objectively defined presentations of information like a pie chart or a graph. For example, the Voyant tools analysis of Ulysses itself gave me more power to understand how and why each word is grouped. Voyant prepared an altered distillation of the text for me to interpret with the same sense of curiosity found when reading the original text itself. Similar to Voyant, the Talk to Transformer website regurgitates a chain of words that may or may not derive from a traditional sense of logic unless I wish to organize its wording with my own sense of logic.

But when exploring the two tools and how pieces of information are presented, I feel the lack of inherent logic acts as a major flaw as well. This flaw can be seen in my Voyant analysis of Ulysses, which is a piece of text that very much strays away from more traditional expectations of writing and general logic. For myself, despite being somewhat acquainted with the novel, the grouping of words often seem like a grouping of words that don’t really have any overlap or commonalities. Perhaps a more coherent, plot-driven novel could allow Voyant to create a steady and understandable cluster of words for me to interpret. Likewise, Talk to Transformer also fails to allow some sense of structure or understanding for me as a viewer. While certain words force the website to regurgitate a silly chain of sentences that make the viewer chuckle, some commands are just an unreadable mess. It is true that the results for “Simulacra are” brought some interesting and varied responses that mostly corresponded with the actual definition of simulacra or its iteration in various fantasy/sci-fi digital media. However, it took me a few tries to find something that actually allowed me to decipher its subjective meaning. There was no rhyme or rhythm regarding its process at times, just certain moments where I could take a glimpse under the hood to guess how and why the algorithms worked in that unique way. I believe having these algorithms set with some parameters or structuring options would grant users like myself a more stable starting point for that subjective viewing, even if that subjective viewing would be tainted by a little ounce of objectivity.