Resolution-based Query Answering for Semantic Access to Relational Databases: A Research Note

TL;DR

This paper presents a novel approach for semantic querying of relational databases using first-order logic reasoning to generate schematic answers, enabling more expressive and scalable semantic access, including for legacy databases.

Contribution

It introduces a method combining FOL reasoning with database instantiation to improve semantic query answering over relational databases, including web-scale applications.

Findings

Applicable to regular and legacy RDBs

Uses abstractions and constrained clauses for schematic answers

Potential for efficient web-scale semantic search

Abstract

We address the problem of semantic querying of relational databases (RDB) modulo knowledge bases using very expressive knowledge representation formalisms, such as full first-order logic or its various fragments. We propose to use a first-order logic (FOL) reasoner for computing schematic answers to deductive queries, with the subsequent instantiation of these schematic answers using a conventional relational DBMS. In this research note, we outline the main idea of this technique -- using abstractions of databases and constrained clauses for deriving schematic answers. The proposed method can be directly used with regular RDB, including legacy databases. Moreover, we propose it as a potential basis for an efficient Web-scale semantic search technology.