1996 SALS-SIG Seminars
SALS-SIG Research Seminar
A Principled Framework for Constructing Natural Language Interfaces to Temporal Databases
MRI Language Technology Group
When: Friday, 16th August 1996
Where: Room E6A357, Macquarie University
Most existing natural language interfaces to databases (NLIDBs) were designed to be used with "snapshot" database systems, that provide very limited facilities for manipulating time-dependent data. Consequently, most NLIDBs also provide limited support for the notion of time. In particular, they answer questions that refer mainly to the present, and do not support adequately the mechanisms that natural language uses to express time (verb tenses, temporal adverbials, temporal subordinate clauses, etc.). The databases community is becoming increasingly interested in temporal database systems. These are intended to manipulate in a principled manner information not only about the present, but also about the past and the future. When interfacing to temporal databases, it becomes crucial for the NLIDB to interpret correctly the temporal mechanisms of natural language.
I argue that previous approaches to natural language interfaces for temporal databases (NLITDBs) are limited, mainly because they have ignored important time-related linguistic phenomena, assumed proprietary temporal database systems, or both. I propose a principled framework for constructing English NLITDBs, drawing on research in linguistic theories of time, temporal logics, and temporal databases. This framework consists of a formal language, TOP, that can capture the semantics of several English temporal mechanisms, an HPSG-based mapping from a non-trivial fragment of English temporal questions to TOP formulae, and a provably correct algorithm for translating from TOP to TSQ2 (TSQL2 being a temporal extension of the SQL-92 database language). To demonstrate that this framework is workable, a prototype NLITDB was implemented, using the ALE grammar development system and Prolog.
|Last modified: July 1997|