Activating The Archive, Or: Data Dandy Meets Data Mining

The joint availability of massive amounts of digitized cultural heritage as well as all the born-digital content (along with the data about people’s production, sharing, and reception of this content)
allows for the new ways of researching, teaching, and exhibiting culture. What are the theoretical
and methodological issues that rise when we start treating culture as data that can be automatically
analyzed and visualized? What are the consequences of treating a pattern as a new basic epistemological element of knowledge? Is it sufficient to borrow the techniques from the fields of computer
science, information visualization and media art – or do we need to develop new techniques specific to humanities? How do we address the new “data divide”- between the people and cultural
processes which leave rich digital traces (and therefore will be analyzed and written about) and
those which do not?
In my talk I will address these and other conceptual issues around “cultural data mining.” My focus
will be on two emerging areas: analysis of visual media and analysis of born digital and web native content. I will demonstrate the techniques developed at Software Studies Initiative at Calit2/
UCSD for the analysis and visualization of patterns in images and video - feature films, cartoons,
television art, user-generated video, etc. I will discuss challenges and new exiting possibilities
which arise when we start looking at web sites and blogs, social media sites, digital art, games and
other interactive media. I will also show the results emerging from our large scale study of cinema,
video games and social media which we are currently undertaking at NERSC (National Department of Energy Supercomputer Center) with the support from NEH Humanities High Performance
Computing grant.