narratives and transformations

Joris was talking about the multiple narratives that weave through this code and Underwood & Seller's writeups.

matt was thinking about the various transformation on the data, starting at teh Hathitrust (and even farther back possibly) and moving though the code that Underwood & sellers have written to do a series of transofrmations on the data, in preparation for the logistic regression, but also including the logistic regression.

arguement that the data preparation and the logistic regression are not ontologically any different. They are all just transformations that produce new representations of the data. The representations themselves are important because they can reveal new insights on the data. But the key is the representation, not the transformation. in this article we focus on the transformations and what is happening in them (describing them empricially), especially the logistic regression. But maybe the focus could or should be the representation of the data on either end of a transofrmation. Case in point, the predictions or the coeffecients, because those are the represetnations of the dataset on the other side of the logistic regression transformation.

Keywords:

• transformation
• representation

of the data. analytical focus, is on the data and on the representstions of the data.

Joris: why does a set of different representations of the same data allow us to infer firm statements about that data. Epistemologically, this is very weird. We haven't added anything NEW to the data, yet we are drawing NEW conclusions. However, we are REMOVING information from data. when we do the logistic regression we are compressing the data. looking for correlation, looking for one pattern (the hypothesis representation). The implication is the insight, the information already exists in the data, but we (humans) can't see it.

the statue is already there, it just needs to be liberated from teh stone.

begs the question, there are probably multiple insights living in the data (just as there are multiple posisble sculptures in the block of stone). This then leads matt to wonder, what are the limits? Does all knowledge in the universe potentially live inside the data?

Computational Narrative - A computational narrative is composed of a series of representations and transformations. They are composed by human and non-human actors (see latour). Computational narratives have, by their very nature, discrete components due to the design of computational technology (computer architecture, programming language design, etc.). Computational narratives document the provenance of representations and transformations. Computational narratives are linear (this one is bold and I'm not convinced is true just yet...)