Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Semantic similarity from natural language and ontology analysis
Harispe S., Ranwez S., Janaqi S., Montmain J., Morgan & Claypool Publishers, San Rafael, CA, 2015. 254 pp. Type: Book (978-1-627054-46-1)
Date Reviewed: Apr 25 2016

Harispe et al. offer a coherent and most-welcome unified view of the vast literature on semantic similarity, covering both corpus-based and ontological methods. It fills a gap in the current literature and is rich with references and interesting insights into the state of the art, from the theory to the experimental evaluations. Researchers and students will find it a great resource to quickly get started in the area.

Semantic similarity is the problem of determining how related two concepts are. It has many well-known applications in search, data analysis, and artificial intelligence, to name just a few areas. While intuitively simple, this problem has many nontrivial nuances, starting from the actual definitions of concept, similarity, and semantics itself. There are two prevalent schools of thought. The ontological approach advocates the explicit definition of meaning through the use of some logical formalism such as an ontology, while the corpus-based approach states that one can infer the meaning of a term by collecting statistics of how the term is used (for example, from a large corpus). These two seemingly diametrically opposed approaches have been studied separately despite sharing many things in common. This book offers a coherent, unified view that is interesting on its own and allows for a better understanding of the problem of computing semantic similarity.

Keeping with the tradition of the “Synthesis Lectures” series, the book is concise and quite technical. It starts with an overarching chapter setting the nomenclature on the topic, defining and contextualizing what is actually meant by semantic similarity and how different communities view and talk about it. This unifying perspective is quite welcome and certainly will benefit many students and newcomers to the area. Next, the book discusses corpus-based semantics, covering a lot of ground and going into detail on the most prominent methods; the reader is left wanting more on some topics, however, especially latent Dirichlet allocation (LDA) and similar methods. Then the reader is taken into the ontology-based notions of semantic similarity, which are more familiar to most and based on graph traversal algorithms and other algorithmic manipulations of graphs. There is quite a lot to cover when it comes to ontology-based methods, from more abstract discussions about the expressiveness of the ontology language (and the associated computational cost for reasoning) to actual methods that have been implemented and validated in the literature. The reader will find a very comprehensive survey of the area and details of the most prominent and more recent work.

Another great contribution of the work is a conscious effort in balancing advantages and disadvantages of the various methods, which is done in every chapter, as well as discussions and examples of approaches that leverage both corpus-based and ontology-based information to define semantics. The book also offers a very thorough chapter compiling the methodology and many of the resources commonly used for experimentation in this field, which will be especially welcome to those starting their research in this area.

In summary, this book, which is quite unique in its coverage, is a welcome addition to the libraries of every graduate student and researcher in the area of semantic similarity. For the practitioners out there, it will probably seem a bit rough around the edges and lacking in examples and/or supplemental material, but most of that has to do with the nature of the series (that is, short books or very long surveys). Even for the novice, the book might serve as a concise starting point with lots of references for further reading.

Reviewer:  Denilson Barbosa Review #: CR144350 (1607-0480)
Bookmark and Share
  Reviewer Selected
Editor Recommended
Featured Reviewer
 
 
Language Models (I.2.7 ... )
 
Would you recommend this review?
yes
no
Other reviews under "Language Models": Date
A framework for investigating language-mediated interaction with machines
Zoeppritz M. International Journal of Man-Machine Studies 25(3): 295-315, 1986. Type: Article
Oct 1 1987
Prolog and natural-language analysis
Pereira F., Shieber S., CSLI/Stanford, Stanford, CA, 1987. Type: Book (9789780937073186)
Jun 1 1988
Competence and performance in the design of natural language systems
Bara B., Guida G., Elsevier North-Holland, Inc., New York, NY, 1984. Type: Book (9789780444875983)
Dec 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy