Please note: This PhD seminar will take place in DC 3301.
Enamul Haque, PhD candidate
David R. Cheriton School of Computer Science
Supervisors: Professors Grant Weddell, David Toman
Data integration complicates itself because traditional databases conflate a real-world entity with its description, but a new approach using the Abstract Relational Model (ARM) and its companion language SQL with Paths (SQLP) solves this by treating entity identity as a primitive, abstract token distinct from literal values. Grounded in a formal theory of reference, this framework introduces Referring Expression Types, a Global Preference Order, and Duration Constraints operationalised by the ToQ algorithm to automatically resolve temporal, schema, and identity conflicts. By systematically lifting legacy databases into ARM, the BuildARM procedure and Unfold rewriting algorithm compile high-level SQLP path queries into standard SQL over original sources without requiring data materialisation. Validated across eight heterogeneous databases at Waterfall University, this architecture scales to resolve both local and global data heterogeneities, transforming data integration from a fragile engineering challenge into a semantically grounded, robust discipline.