[Paper Review] Treating Coordination with Datalog Grammars

Verónica Dahl, Paul Tarau+2 more|ArXiv.org|May 3, 1995

Logic, programming, and type systems9 references3 citations

TL;DR

This paper introduces a Datalog grammar variant that enables a meta-grammatical treatment of coordination in logic grammars, improving over prior approaches by leveraging database-inspired query mechanisms. It demonstrates enhanced efficiency and expressiveness in handling coordination phenomena such as conjunction and disjunction in natural language syntax.

ABSTRACT

In previous work we studied a new type of DCGs, Datalog grammars, which are inspired on database theory. Their efficiency was shown to be better than that of their DCG counterparts under (terminating) OLDT-resolution. In this article we motivate a variant of Datalog grammars which allows us a meta-grammatical treatment of coordination. This treatment improves in some respects over previous work on coordination in logic grammars, although more research is needed for testing it in other respects.

Motivation & Objective

• Address the limitations of traditional Definite Clause Grammars (DCGs) in handling coordination phenomena in natural language syntax.
• Overcome inefficiencies and expressiveness gaps in existing logic grammar approaches to coordination.
• Introduce a Datalog grammar variant that supports a meta-grammatical treatment of coordination, inspired by database query mechanisms.
• Demonstrate improved performance and flexibility in parsing coordinated structures using Datalog-based inference under OLDT-resolution.
• Lay the foundation for future evaluation of the approach on broader linguistic coordination patterns.

Proposed method

• Adapt Datalog grammar formalism to represent syntactic rules with recursive, relational structure, drawing from database theory.
• Represent coordination rules as Datalog predicates that can be queried using standard Datalog evaluation strategies.
• Employ OLDT-resolution as the evaluation mechanism to ensure termination and efficient execution of grammar rules.
• Model coordination constructs (e.g., 'and', 'or' in phrases) as logical combinations of sub-constituents using Datalog’s relational composition.
• Use meta-grammatical rules to generalize coordination patterns across different syntactic categories.
• Integrate the grammar with a logic programming framework to support parsing and recognition of coordinated structures.

Experimental results

Research questions

• RQ1 How can coordination in natural language syntax be modeled more effectively using logic grammars?
• RQ2 Can Datalog grammars provide better efficiency and expressiveness than traditional DCGs for coordination?
• RQ3 What are the advantages of using a database-inspired formalism for handling recursive and compositional coordination?
• RQ4 How does the meta-grammatical treatment of coordination in Datalog grammars compare to existing approaches in terms of coverage and termination?
• RQ5 What performance improvements does the Datalog-based approach offer under OLDT-resolution compared to standard DCGs?

Key findings

• The proposed Datalog grammar variant enables a more systematic and extensible treatment of coordination in logic grammars.
• The use of Datalog's relational model allows for cleaner separation of syntactic rules and coordination logic.
• The approach demonstrates improved efficiency over traditional DCGs under OLDT-resolution, particularly in handling recursive and complex coordination structures.
• The meta-grammatical approach supports generalization across coordination types, such as NP coordination and clause-level coordination.
• The framework shows promise for integration with existing logic programming systems, though further empirical testing is needed.
• The formalism supports termination and predictable execution, which is critical for practical parsing applications.