.png)
38 Views
Society for Cultural Anthropology
Oral Presentation Session
Luis Felipe Murillo
Data Science Institute, University of Virginia
Natural language processing (NLP) is one of the long-established areas of artificial intelligence (AI) research. As such, it is a constitutive part of the grand promises of AI with the application of large-scale data analytics for informational retrieval, machine translation, and natural language understanding for personal assistants and question answering systems. “Ambiguity,” however, persists as one of the hardest problems in this area. In computer science and computational linguistics, the “natural language problem” has been addressed by several research programs with a myriad of formal and statistical approaches in the past 60 years. In the humanities and social sciences, ambiguity has been primarily connected with questions of interpretation, thus having a radically different statute: it has not figured as a problem to be solved with probabilistic and data-driven models, but as a symptom for the functioning of language in context. In this paper, I describe an early application of NLP for the study of political discourse to examine an intersection between anthropology and computing. The goal is to describe the limits of NLP to recast questions of meaning and interpretation, charting new terrain for ethnographic studies of computational analytics. For the conclusion, I suggest how “ambiguity” can serve as a privileged vantage for anthropological studies of autonomous systems. Beyond human-computer interface studies—where microsociological frames have been applied to examine affordance, intentionality, phenomenological modification, and affect—ethnographically-informed language constructs could substitute formal, machine-actionable implementations of language as a transparent medium by integrating lived and historical experiences.