The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

PhD Thesis #9317

Title:Relational Extensions to Feature Logic: Applications to Constraint Based Grammars
Date: 1993
Abstract:This thesis investigates the logical and computational foundations of unification-based or more appropriately constraint based grammars. The thesis explores extensions to feature logics (which provides the basic knowledge representation services to constraint based grammars) with multi-valued or relational features. These extensions are useful for knowledge representation tasks that cannot be expressed within current feature logics. The approach bridges the gap between concept languages (such as KL-ONE), which are the mainstay of knowledge representation languages in AI, and feature logics. Various constraints on relational attributes are considered such as existential membership, universal membership, set descriptions, transitive relations and linear precedence constraints. The specific contributions of this thesis can be summarised as follows: 1. Development of an integrated feature/concept logic 2. Development of a constraint logic for so called partial set descriptions 3. Development of a constraint logic for expressing linear precedence constraints 4. The design of a constraint language CL-ONE that incorporates the central ideas provided by the above study 5. A study of the application of CL-ONE for constraint based grammars The thesis takes into account current insights in the areas of constraint logic programming, object-oriented languages, computational linguistics and knowledge representation.

[Search These Pages] [DAI Home Page] [Comment]