Cultural factors impact how we interpret data. These factors can vary over time and place. Existing methods for cultural analytics, including AI, generally overlook these factors.
The mission of CANAL is to co-develop and co-research culturally-aware AI methods from a perspective that is both critical and ethical.
Our core analytical challenges include:
- investigating multimodal cultural records that require context analysis to assess their meaning
- understanding dynamic developments over time and across space
- developing a more inclusive and global approach to data
Our core research themes include:
Global Cultural Analytics
Cultural Analytics includes methods to explore and analyze cultural data, agents, and processes. Global Cultural Analytics aims for a global scope in terms of geography and type of data.
Our research within this theme is focused on (1) analyzing large cultural data sets, (2) visualizing cultural collections and patterns, and (3) applying AI to examine trends and patterns in large-scale collections. Through our work in this area we aim to contribute a transnational perspective on data collections and develop new methods for Global Cultural Analytics.
Modeling Complex, Dynamic Systems
The focus on computational methods surfaces the tension between quantitative and qualitative approaches to scholarship. From a humanities perspective, the use of quantification is often criticized for being positivistic and reductionist, and thus leaving out the complexity of culture. Faced with an increasing amount of cultural data and our reliance on AI systems to make sense of this data however, questions surrounding the role of quantification in the humanities resurface.
Current AI systems rely on categorization and considerably reduce the complexity of the world. In the sciences, theories and methods have been developed that address this complexity and set out to model this to learn about how dynamic social systems work. These methods have found their way into economics, the social sciences, and ecological sciences. In the humanities, this is still in its infancy.
Our research within this theme explores how, and to what extent, theories and methods from complexity sciences could be translated to the humanities context.
Computational Visual Studies
Over the last few decades, our world has become more visual. Many of these visual sources have been digitized, many more created in a born-digital form. Visual Culture Studies examines these objects; it is an interdisciplinary domain that approaches visual objects and visuality in culture from (art) history, media studies, and the social sciences.
The methodologies employed in Visual Culture Studies range from critical and hermeneutical approaches to more formalized analyses of the content and distribution of images. AI can play an important part in these analyses — ranging from the analysis of the images themselves, their distribution through media networks, to the power dimensions that determined their production, reception and distribution. As such, there is a clear link to existing computational domains like Distant Viewing or Multimedia Analytics. Computational Visual Studies offers a theoretical and methodological framework that can integrate these fields of analysis.
Our research within this theme is aimed at (1) answering questions related to visual culture using AI, (2) developing AI technology and methods to examine sources in their sociocultural (historical) context, and (3) formalizing multimodal material for computational analysis.