Natural Language Processing for Prolog Programmers

Rating:
Summary: One of the best intros to NLP
Review: This book, while a bit dated, still the best discussion of a wide variety of (nonstatistical) parsing methods in print. The best part of this book is the table on p. 191, where the author compares the runtimes of the various parsing algorithms, and shows that the one with the scariest "big O" complexity is actually the fastest in practice. This is a very important lesson to impart to budding computer scientists: use "big O" analysis wisely, not as the voice of God, as to which algorithm is the best practical algorithm.

I've given this book 4 stars instead of 5 stars because it is a little out of date: look elsewhere for the best discussions of quantifier scope handling and discourse. Also, it doesn't contain any discussions on the what is today the more fashionable statistical-based approaches, but the jury is still out, IMHO, on whether this is a drawback or not.