ANALYSIS OF A FINITE-STATE GRAMMAR FOR PARSING AVIATION-SAFETY REPORTS
César V. Dragunsky · Summit (Simon Fraser University) · 2004
With the growth of the World Wide Web in the nineties, alongside the increase in storage and processing capabilities of computer hardware, the problem of information overload resulted in an increased interest in finite-state techniques for Natural Language Analysis as an alternative to fragile, slower algorithms that would attempt to find complete parses for sentences based on general theories of language.As it turns out, shallow parsing, a set of robust parsing techniques based on finite state machines, provide incomplete yet very useful parses for unconstrained running text.The technique, however, will never provide 100% accuracy and requires that grammars be geared to the needs of particular data samples.In this project, we take a corpus of aviation safety reports parsed by Cass, an existing partial parser, with a particular given grammar, and look for instances of linguistic constructs whose treatment by the parser could be improved by modifications to the grammar.A few such constructs are discussed, and the grammar is edited to reflect the desired improvements.A parser accuracy measure is implemented and evaluated before and after the grammar modifications.First of all, my parents.Second, people like Verhica Dahl and Diana Cuikerman, whose help when I was just arrived in Canada was invaluable and probably a big factor in the eventual realization of this project.My great friend Rick Ouellet shuold not be left without mention either.I have to include all of the fantastic people at the School of Computing Science at SFU, all of whom have always been helpful and their help has many times been crucial in the concretion of this h4.S~.project.Finally, my supervisors, Fred Popowich And Oliver Schulte, whose seemingly infinite patience, faith, and guidance, were the most important factor in making this project a reality.Along with them I want to thank the remaining members of the committee,