Attribute grammars for the heuristic translation of query languages

Abstract

This work is motivated by the necessity of providing a high-level user-interface for database query languages to improve the accessibility of information to the nonexpert users. The goal is to achieve portable, adaptable query languages which are supported by mechanisms that allow one to change the query language without changing the implementation. An attribute grammar is used for semantic description of the language. Imprecise terms which are hedges (e.g. very, almost) or nonstandard quantifiers (e.g. many, most, few) are permitted. The structure of the database is transparent to the language because a rule-based approach is taken to determine which relations a query implicitly references. The attributes in the database are treated as linguistic variables for which there exists an abstract description in the metadatabase. These heuristic measures encompass a large class of high-level query languages.