These include: their accidental nature, their open availability toresearchers, and the ubiquity of their presence in everyday urbanlife.

It seems to us, based on the datasetsthat we have examined, that the key boundary charac-teristics of Big Data, which together differentiate itfrom small data, are velocity (both frequency of gener-ation, and frequency of handling, recording, and pub-lishing) and exhaustivity

ontology,

Uprichard (2013) notes several other v-words thathave also been used to describe Big Data, including:‘versatility, volatility, virtuosity, vitality, visionary,vigour, viability, vibrancy...virility...valueless, vam-pire-like, venomous, vulgar, violating and very violent.’More recently, Lupton (2015) has suggested droppingv-words to adopt p-words to describe Big Data, detail-ing 13: portentous, perverse, personal, productive, par-tial, practices, predictive, political, provocative,privacy, polyvalent, polymorphous and playful.

The task is to shift from studying the effects of technology

A crucial task is to expose the myth of data as “given” and observer- independent.

Discussing Martin Heidegger and his profound influence, Bruns describes the shift from positing the site of meaning within a text to identifying it with the act of reading as a process of negotiation.2

between what is communicated and how have to be acknowledged.

Every presentation structures arguments— it doesn’t simply “reveal” facts, or forms, in all their purity.

The development of a means of systematically inscribing enunciation in visualizations is essential to advancing the interpretative agenda of the humanities.

A simple example should help make this clear: if I have a diagram on my screen and decide that two of the points in the display are related to each other in a particular way, I draw a bold line of connection between them. Connection is an interpretative concept. Connection is not a thing, not an entity being represented, it is a concept that is being modeled. The two points may have been part of a representational display, a conventional chart or graph.

Visualization software can be divided into the following categories: (1) drawing programs that generate images algorithmically (e.g., Processing) or through rendering (e.g., Rhino) in pixel/raster or vector formats with surface textures and other visual effects; (2) visual displays of quantitative (numerical or statistical) informa-tion (Tableau, Google Fusion, Excel charts, scatter plots, etc.); (3) forced or directed graphs (Cytoscape, Gephi, and other network visualizations) gen-erated through computational analysis of betweenness and other factors; (4) simulations of complex, nonlinear, or dynamic systems (Game of Life, VisSim); (5) visualizations from integrated data analysis (Inscriptifact, imag-ing, forensics, etc.); (6) visual presentations of data mining or analysis (Voy-ant, Google Fusion Tables, Tableau).

As noted here repeatedly, the visualizations adopted by digital human-ists (charts, graphs, diagrams, maps, and timelines) were mainly developed in the natural sciences, social sciences, statistics, business applications, and other fields. These bear the hallmarks of positivist, quantitative and/or sta-tistical approaches to knowledge that limit their application to interpretative practices in the humanities.

Nonrepresenta-tional images are ones that do not serve as surrogates for an already existing object, whether that object is a thing, a place, an experience in the world, or a data set. In this context, the term nonrepresentational emphasizes the constructed character of knowledge production as interpretation

The timeline of these works also aligns with an increased interest in the decolonization of knowledge and all that this implies.

User- dependent knowledge is interpretative, partial, and situated. There-fore, empirical modes of graphic display of information are as unsuited to modeling interpretation as a thermometer is for measuring the warmth of a human emotion or the strength of an embrace.

The idea of modeling implies that a graphical expression serves as a primary mode of knowledge production, not a secondary expression of preexisting data.

But these pro-cesses of transformation— from phenomenon to data and then to display—are rarely documented or noted.

But the idea that the graphic display is a presentation of the data stands unques-tioned. No additional consideration is necessary. The image is considered to be the data expressed in graphic form. The goal is to find the “best” and “clearest” display of data, and the correlation between the data set and the presentation is assumed to be direct, a matter of equivalence.2

modeling interpretation is fundamentally at odds with cur-rent methods of information visualization

borrowed

n direct contrast (not opposition, but distinction) to user- independent knowledge, visual inscrip-tions argue for the specificity of user- dependent conditions of knowledge production as interpretative— where interpretation signals the subjective, located, inflected, and particular character of knowledge located within a subjective experience

The crucial epistemological issue in advocating visual, graphic expressions for interpretative work is based on the distinction between inscriptional specificity and notational generalizability.

Powerful

Even the most basic aspect of my argument— that visual methods can produce knowledge and enact interpretation, not just serve as representations or displays— is considered polemical by many.1

Visual epistemology, therefore, can be both declarative/descriptive and propositional/provocative.

They have the capac-ity to produce and embody information, not just represent it.