The DatalogLab is an experimental 'open' deductive database system with the following main features:

  specification of deductive rules as well as active and normative rules (integrity constraints)
  query optimization (especially tailored for the language Datalog)
  tight integration of deductive and active rules
  update propagation and view updates
  integrity checking and integrity repair
  set-oriented transactions

The DatalogLab supports the definition and generation of extensional relations which are explicitly given sets of facts as well as derived relations which do not physically exist, but are implicitly given in form of rules. Derived relations are either partially computed on demand, i.e. when a user query for a derived relation is evaluated or fully materialized, i.e.automatically adapted whenever base data change. Deductive rules may be non-recursive (corresponding to view definitions in relational databases) or recursive (stratifiable as well as unstratifiable).

Datalog is the favourite relational languages for investigating deduction over databases. Its syntax and semantics is well-agreed upon by most researchers as far as facts and rules are concerned. However, schema information, queries, updates and transactions, constraints and triggers do not have such a 'quasi-standard' syntax in the Datalog community. The DatalogLab therefore offers its own DDL and DML conventions for Datalog turning it into a full DB sublanguages comparable with SQL.

The major goal of the DatalogLab is to provide several alternative methods for accessing implicitly given data and to be open for additional access procedures. The DatalogLab should more be seen as a didactic tool than as a database system for 'real applications' despite of the fact that we try to cover features of many different research areas. It is result of a kind of unease we felt about how deductive databases are taught at universities and in textbooks. Quite often a number of different query evaluation methods (like naive, semi-naive and alternating evaluation) and rewriting techniques (like magic-sets) are formally presented and theoretically described. Unfortunately, students hardly get any 'feeling' for these different methods and cannot study their behaviour 'at work'.  We thus decided to build our own system which implements these methods in Prolog very close to their theoretical definition in order to enable students to watch and to compare their impacts.

Links

  Download the DatalogLabV 0.1
  Online DatalogLab Installation Guide
  DatalogLab: Physical and Logical Structure

Contact

Prof. Dr. Rainer Manthey
Römerstraße 164
D-53117 Bonn, Germany
Tel.: +49 (0)228 73-4528
Fax: +49 (0)228 73-4212
Email: idb@informatik.uni-bonn.de

Last update: A. Behrend, 03-Nov-2000