Descriptive Types for Linked Data Resources

• Main Authors: Horne, Ross, Sassone, Vladimiro, Ciobanu, Gabriel
• Other Authors: Voronkov, Andrei
• Format: Conference or Workshop Item
• Language: English
• Published: 2015
• Online Access: https://hdl.handle.net/10356/80953; http://hdl.handle.net/10220/39005
• Institution: Nanyang Technological University

This work introduces the notion of descriptive typing. Type systems are typically prescriptive in the sense that they prescribe a space of permitted programs. In contrast, descriptive types assigned to resources in Linked Data provide useful annotations that describe how a resource may be used. Resources are represented by URIs that have no internal structure, hence there is no a priori type for a resource. Instead of raising compile time errors, a descriptive type system raises runtime warnings with a menu of options that make suggestions to the programmer. We introduce a subtype system, algorithmic type system and operational semantics that work together to characterise how descriptive types are used. The type system enables RDF Schema inference and several other modes of inference that are new to Linked Data.