Natural Language Processing for Online Applications: Text Retrieval, Extraction and Categorization

by Isabelle Moulinier and Peter Jackson

Book Overview

Technology practitioners in the Internet space are constantly faced with difficult decision-making as to what research has been done and what the best practices are. This lucid text covers the emerging technologies of document retrieval, information extraction, and text categorization.

• Format: Paperback
• Language: English
• ISBN: 902724989X
• ISBN13: 9789027249890
• Release Date: January 2002
• Publisher: John Benjamins Publishing Company
• Length: 226 Pages
• Weight: 0.85 lbs.

Customer Reviews

An excellent overview

Work on Natural Language processing has been going on for at leastthirty years. In the past most natural language processing (NLP)applications where mainly in the research realm. The rapid increase in computer processing power and disk storage capacityhave moved NLP from research into the area of applied science.This gives NLP the feel of a new and vibrant area. Progress isbeing made rapidly, but the research literature can be difficultfor someone who has no experience with NLP. Simply learning theterminology can be time consuming.This book by Jackson and Moulinier provides an excellent overviewof several sub-areas of NLP applied to natural language text.Both Jackson and Moulinier have been involved in implementingNLP applications in a commercial context, so there is asome concentration on applying NLP in real applications, ratherthan artificial contexts like the Message Understand Conferencedata set.I purchased this book for its coverage of Information Extraction.Along with this book I read a number of papers from the researchliterature. One paper I found particularly interesting wasa paper on the FASTUS NLP system developed by researchers atSRI. I was very happy to see that FASTUS and finite automataapproaches were covered in more detail in this book.For my purposes I would have liked a book that moved froman overview of various NLP applications to more implementationdetail. For example, while I think that I understand push down automata from working on parsers for compilers,I don't fully understand them in the NLP context. This bookdid not go into enough detail to make this clear.I cannot really offer this lack of detail as a criticismsince I don't believe that it was the authors intention toprovide this level of detail. Their objective is to providea detailed overview and I think that they succeeded in doingthis. A book that provided this overview withdetails on implementation would be much longer, perhaps thesize of Manning and Schutze's excellent book "Foundations ofNatural Language Processing" (Manning and Schutze providea great deal of detail, but they do not cover information extraction).

Excellent

I recommend this very highly. The words "online applications" in the title suggest that this book is about NLP for websites, but it's much more general than that; certainly any of the technologies discussed in it could in fact be implemented on a website, but "online applications" should be interpreted as meaning something like "applications that are made possible or commercially viable by the availability of large bodies of documents over the Internet."The focus of the book is on technologies with commercial applications, and that aspect of the topics discussed in the book is addressed clearly and well. However, there's also plenty in the book that will be of interest to a researcher, especially one looking for an overview of a topic. In fact, the book reads much like a series of well-written review articles.The first chapter of the book discusses some of the general issues and challenges in natural language processing, on a levelthat should be accessible to pretty much anyone. (Actually, one of the really outstanding features of this book is its overall readability.) Subsequent chapters focus on information retrieval, information extraction, and text categorization. The final chapter has shorter sections on other topics, including summarization and named entity recognition.All chapters of the book are characterized by high readability, clear explanations of algorithms (and I say that as someone who struggles with ANY algorithm), and good explanations of the relevant evaluation metrics.This book would be a good starting point for anyone; if you're not a beginner in natural language processing, you'll still find much that's useful in this book.