Pluralizing Logics: The Binaries and Discontents of Knowledge Representation

Friday, January 5, 2018: 10:30 AM
Roosevelt Room 4 (Marriott Wardman Park)
Lindsay Poirier, Rensselaer Polytechnic Institute
The knowledge representation community in the field of artificial intelligence has had a

dramatic impact on the aims and logics that have oriented the design of the semantic web – an

information infrastructure that aims to structure data on the World Wide Web so that it can be

understood by machines. In this paper, I show how the design of the semantic web and the

politics that it engenders has been influenced by debates that emerged in the history of artificial

intelligence.

To begin, I trace several divides that emerged in the knowledge representation community in

the 1970s and 1980s – in particular those between the ‘neats’ vs. the ‘scruffies’ and the

‘declarativists’ vs. the ‘proceduralists’. Neats and declarativists tended to apply "clean"

approaches to knowledge representation; aiming to design knowledge systems characterized by

completeness and soundness, these researchers tended to model systems with formal

logic. Scruffies and proceduralists, on the other hand, used more pragmatic means to model

knowledge; they aimed to get systems to work, even if this meant modeling knowledge in messy

ways.

While these divides are well-documented in historical accounts of AI, little attention has been

paid to the attempts made by some researchers in the knowledge representation community to cut

across these binaries. Throughout the 1970s and 1980s, several researchers working in the neat

and the scruffy tradition published papers that attempted to rethink the divides, showing how

logics could be pluralized and neat and scruffy approaches could be brought together to advance

knowledge representation. In this paper, I show how these attempts to rethink the binaries that

characterize knowledge representation work have shaped approaches to designing the semantic

web. I argue that efforts to rethink the binaries that construct knowledge representation work has

opened possibilities for codifying alternative and more democratic logics into the World Wide

Web.

Previous Presentation | Next Presentation >>