In the late 1980's and early 90's, the field of temporal databases was in ferment, split into two camps, the first-normal-form (1NF) types, who felt that the simplicity and ease of implementation of period-stamped conventional tuples was the way to go, and the Non-1NF (or N1NF, or NF2) types, who felt that the collection of multiple values, say of salary for a particular person, combined together into a single, grouped tuple brought semantic fidelity to the model and expressiveness to the query language. The arguments on both sides were rather ad hoc and unconvincing (at least to the other side), and so the debate raged on ad infinitum.
This TODS paper proposed two formal standards of historical completeness, one for ungrouped (i.e., 1NF) query languages and data models, and one for grouped query languages and data models. In doing so, the paper brought amazing clarity to this long-standing controversy. The paper is persuasive in arguing that it is this notion of "grouped" which is central. This is a very nicely done paper that has brought considerable insight to bear on historical data models and query languages. In particular, it handles in a precise way the simulation of one data model in another data model, such as simulating groupness via a group id column in a non-grouped model.
Jim Clifford's earlier TODS paper was equally influential in the temporal database field. Tragically, Jim passed away in August, 1995, the year after this second paper appeared, thereby depriving the research community of subsequent insights that Jim would undoubtably have contributed, and the world of a unique and wonderful person.
Copyright © 2000 by the author(s). Review published with permission.