A first attempt at a constituency parser for Maltese

Abstract

The aim of this dissertation is to conduct a first attempt at building a functioning constituency parser for Maltese, along with the creation of a probabilistic grammar in Chomsky normal form. The majority of this dissertation is in the code which was written for it to function, however detailed explanations of every step taken are explained throughout the paper. The data acquired to create the probabilistic grammar was retrieved from the Maltese Language Resource Server, and transformed accordingly. The parsing algorithm implemented is the CYK algorithm, adapted to take probabilistic grammar rules. An evaluation of the entire system is conducted using the Parseval standard of evaluation, and is done automatically through an off-the-shelf module. The results show that the parsers evaluation is not so high, which may be attributed to the use of a greedy search, but is commendable nonetheless.