Extracting information from text using FS technology

Abstract

Nowadays, with the advent of the Internet, we are all living in the information era, but computers are not yet powerful enough to make full use of the information available. This dissertation describes an attempt to extract information from text written in the Maltese Language to construct a domain knowledge base. The procedure is to convert a sentence written in Maltese into a logical form, which can be used to derive the concepts and relationships to be entered in the knowledge base. The system makes use of Finite State and Statistical techniques to perform shallow/light/incremental parsing of text. The restricted domain of the text corpus (cookbook recipes) was exploited to convert more than 70% of the sentences into logical forms, which correctly showed the semantic binding between the elements.