Situation Schemata as Representation Format and Query Language for a Non-Standard Database System

Computers will probably never achieve the level of human skills in natural language processing, but they can well aid in such actions as searching in a database. The project NALOG – NAtural language processing and LOGic programming – aims at querying a database in natural language and getting answers in natural language again. As query language and internal representation format, we explore the use of situation schemata which were developed in an attempt to bring together the unification-based frameworks prominent in NL syntax analysis and situation semantics, one of the leading NL semantic theories. We describe an experimental question-answering system based on the Complex-Relational Data Base. Operations of the Complex-Relational Algebra perform directly on situation schemata generated from NL questions. As far as we know, we present the first evaluation strategy for situation schemata directly based on a relational data model.