Date(s) - 11/28/2018
2:30 pm - 3:30 pm
3910 15th Ave NE
Please join us for our next Data Science Studies meeting on Wednesday, November 28th from 2:30-3:30 pm in the WRF Data Science Studio on the 6th floor of the Physics/Astronomy Tower, where guests Andrew S. Hoffman and Emily M. Bender will help us explore the theme of “Domains and the Data Sciences.”
Domains and the data sciences: The conception and consequentiality of boundary making and breaking
Within the discourse and practice of data science, “domains” are a frequent topic of discussion. Whether juxtaposing domain science with data science, positing the constitutive role of domain knowledge in data science, or assessing the state of data science methods across various domains of inquiry, the term is invoked as an organizing concept for making sense of the relationship between data science and other distinct academic traditions. In this session, we will hear from two scholars who interrogate such relationships from different perspectives. Andrew S. Hoffman, a scholar of Science and Technology Studies in the Department of Human Centered Design and Engineering, will address the historical and contemporary mobilization of the “logic of domains,” surfacing its implications for the distribution of resources and the organization of data science within the academy. Emily M. Bender, a computational linguist with appointments in Linguistics and Computer Science and Engineering, will discuss the sometimes fraught relationship between linguistics and natural language processing, suggesting ways that relationship might be rendered more fruitful.
Abstract for Andrew S. Hoffman’s Talk:
In contemporary data science parlance, the notion of a ‘domain’ presupposes the existence of general or ‘domain-independent’ techniques and processes. Domains, in other words, are the worldly material upon which these more ‘universal’ data scientific approaches operate as they work to inform computational advancements and/or to intervene upon existing domain-specific practices. While the contours of ‘domains’ were once a matter of debate, since the 1990s we notice an increasing vernacularization of the term, culminating in a style of organizing we call the logic of domains. This talk briefly presents an historical and conceptual overview of the logic of domains before moving on to discuss some of its implications for the funding and doing of data science, as well as how it relates to other modes of engaging within and across communities involved in data-intensive work.
About Andrew S. Hoffman:
Andrew S. Hoffman is a Research Scientist in the Data Ecologies Lab, Human Centered Design & Engineering, University of Washington. Working at the nexus of STS, pragmatic sociology, and infrastructure studies, he uses ethnographic and documentary methods to understand how humans and technologies coordinate to get things done together. Currently, Andrew’s main project is a socio-technical analysis of the institutionalization of the data sciences in the US vis-a-vis the development of large-scale knowledge and information infrastructures. Prior to joining HCDE, Andrew carried out his doctoral work in the departments of Social Studies of Medicine and Sociology at McGill University, where he conducted an ethnography of a multi-center initiative aimed at developing novel methods for prioritizing and designing clinical trials of diagnostic technologies in cancer care.
Abstract for Emily M. Bender’s Talk:
Natural language processing (NLP) produces many useful tools for people working with text-as-data. The field is nominally interdisciplinary, but in many cases falls short of that ideal: much work at the top NLP conferences is machine learning algorithms applied to linguistic data written by people without much or any training in how language works. This leads to less effective technology, less effective evaluation of technology, less inclusive technology, and less understanding of the biases in the technology. So what can be done? These are complex problems which demand multifaceted solutions, but starting points include: more truly interdisciplinary collaboration, making linguistics accessible to specialists in computer science, and creating and promoting practices that bring properties of data into focus.
About Emily M. Bender:
About Emily M. Bender:
Emily M. Bender is a Professor of Linguistics and Adjunct Professor of Computer Science & Engineering at the University of Washington. She earned her PhD in Linguistics from Stanford University in 2000 and has been a member of the faculty at UW since 2003. Her primary research interests are in multilingual grammar engineering, the study of variation, both within and across languages, computational semantics, and the relationship between linguistics and computational linguistics. She is the founding faculty director of UW’s professional master’s in computational linguistics (CLMS), past-chair of the North American Association for Computational Linguistics, and a member of UW’s Tech Policy Lab.