Factorized Databases: Past and Future Past

                               Dan Olteanu

           Computer Science Department University of Oxford
                      dan.olteanu@cs.ox.ac.uk



 Abstract. In this talk I will overview the FDB project at Oxford on
 succinct, lossless representations of relational data that I call factor-
 ized databases. I will first present a characterization of the succinctness
 of results to conjunctive queries and how factorizations can speed up
 query processing.I will then comment on how this succinctness charac-
 terization relates to seemingly disparate results on: readability of prove-
 nance polynomials, representation systems for incomplete information,
 one-pass query evaluation using finite cursor machines, tractability in
 probabilistic databases, and parallel query evaluation with one synchro-
 nization step. I will conclude with two near-future projects that brought
 me back to factorized data representations: scalable machine learning
 over relational data and distributed database systems with low commu-
 nication cost.