Unstructured Data Integration through Automata-Driven Information Extraction

Extracting information from plain text and restructuring them into relational databases raise a challenge as how to locate relevant information and update database records accordingly. In this paper, we propose a wrapper to efficiently extract information from unstructured documents, containing plain text expressed with natural-like language. Our extraction approach is based on the automata formalism to describe the wrapping process running from text documents to Databases. As usual, relevant information in the text document are delimited by regular expressions, which define the extracting automaton. Each automaton is enriched by an output function that automatically generates SQL queries synchronized with the extracting process in order to insert extracted data into database records. We validate our extraction approach with automaton-based prototype to extract legal information about Lebanese official journal decrees and automatically insert them into a relational database.