Customer Reviews

Understanding Search Engines: Mathematical Modeling and Text Retrieval (Software, Environments, Tools)

5 star:		(1)
4 star:		(3)
3 star:		(0)
2 star:		(2)
1 star:		(1)

 

 

I read this book because I am starting academic research on search engines. It was one of my first books on the subject. It actually deals with two aspects: (1) relevant issues in search engine design, and (2) a mathematically sound approach in building and querying large index strucutres. The explanation in the book on both aspects is short but to the point and explained in an understandable way. It also contains a short list + description of some key references. Great !

There are better books in the market, and even the author would be the first to recognize it. However, this book is one of the most clear and readable introduction to the subject that you can find.

The author fully acomplishes the objective: teach his reader, at undergratuate level, how search engines work. Even some difficult subject, such as LSI, are treated at a level one can easilly understand.

One of the most important characteristics of the book is that it does math. Every formula has an example, usually using small matrix that allow the reader to easilly follow them.

The book is suitable for an objective introduction to the field. It is not very "academic", in the sense it is rather informal. If it is not a textbook, it could help some bewildered student to grasp the inner workings. It could also help a teacher to find clearer ways for explanations and good examples for classroom.

This book is condensed knowladge about the topic of search engines - but only the basics. I am very dissapointed because I thought I will see something special, inspiring.
This book is good for beginners in search engines field but not for the money it costs now.

As others have pointed out, this book is very short. As a consequence, it leaves out a lot of details and forces the reader to refer to another book. This is more noticeable in the sections that do not relate to linear algebra (stemming, performance evaluation, and user interface design). If you want more information about these topics, it is best to look for another book.

However, the discussions about latent semantic indexing and querying based on link structure are more detailed in comparison and both topics are mentioned within the context of linear algebra.

Don't expect an introduction to QR or SVD matrix decompositions or what an eigenspace is. Also, don't expect a proper definition of what a graph is. For all of this, you will also have to refer to another book. If you do not need such an introduction, then you may not mind.

Overall, the book attempts to do too many topics in few pages and suffers from this. However, if you are looking for a "crash course in search engines"-type book, then this might be the one for you. You may end up buying another book afterwards if you want to know implementation details, though.

Other than showing the readers how to design a search engine, the authors, Michael W. Berry and Murray Browne of "Understanding Search Engines: Mathematical Modeling, and Text Retrieval," intend to fill the gap between applied mathematics and information management. In a latent semantic index (LSI) system, mathematics plays a major role in search engine performance. The term-by-document matrix of the system would be transformed to a lower rank matrix for conceptual indexing. However, nobody knows how low the rank should be for the best performance. The best technique so far for lower rank approximation is called singular value decomposition. In such a system, vectors model both documents and queries. The angle between the document vector and the query vector determines the rank-order of the document. The elements of the vectors are usually the weighted frequency of the term occurrence. Thus the searchers should list as many terms as possible in their queries for better search results.

LSI search engine is good for small document system only. Other searching methods such as HITS and PageRank are introduced. For the readers who have the background on linear algebra, numerical linear algebra, and search engine should find this book interesting.

Generally speaking, the book is brief. It has 117 pages and 9 chapters. The nine chapters are Introduction, Document File Preparation, Vector Space Models, Matrix Decompositions, Query Management, Ranking and Relevance Feedback, Searching by Link Structure, User Interface Considerations, and Further Reading. Chapter two (Document File Preparation) reminds the readers that the documents of the system needed to be "clean-up" and index. The works may require plenty of manual labor.

Good first (and short) book on the subject. Easy to follow and understand. Most suitable for reader who has some exposures to numerical analysis and/or numerical linear algebra because QR factorization, SVD, Semidiscrete decomposition (SDD) etc. methods are used in the text.

This little booklet is best to be read at the bookstore.

There is no need to buy. In fact, buying it is a waste
of money, given the $[money] price tag. That is about a $[money]
a page.

So, save your money and time...