The intelligent web: Search, smart algorithms, and big data

by Gautam Shroff

The internet has increasingly helped produce and provide access to overwhelmingly big data, from page clicks to shopping transactions, and there is great commercial interest in exploiting it. Gautam Shroff, the author and Chief Scientist at Tata Consultancy, coins the phrase "web intelligence" to describe the ways in which the increasingly complex algorithms which operate upon these data work together in order to draw meaningful inferences; a process which can be considered a form of intelligence. This book does not consider the kind of intelligence that is comparable to that of humans; nor is it interested in the philosophy of what it actually means to learn. Instead, it is an introduction to the machine learning techniques which are used to produce non-sentient artificial intelligence.

Each chapter of the book is dedicated to one of the abilities which Shroff argues is necessary in order to progress towards an ability to make inferences about the world. The first chapter, look, introduces how data is sorted and indexed by algorithms, with a focus on Google's page rank algorithm. Next comes listen: an algorithm must be able to retrieve useful information from the surrounding noise. Learn discusses algorithms which can discern structure in datasets and connect looks at the ability to put together multiple datasets to derive new conclusions. With this ability also comes the possibility of being able to predict future behavior of a system. The final chapter is correct, which explores how predictions can be used to correct current behavior, as in the case of a self-driving car. The distinction between such systems and ourselves is purpose, for we must provide the purposes of any machine learning system, and it is this which is the subject of Shroff's epilogue. This epilogue is more philosophical in nature than the main body of the book, and posits that it might be possible to pass our purposes to a set of algorithms in a much more general sense, and have the machines work independently from there.

Although the author is also known for his Coursera course Web intelligence and big data, this book is not a computer science or statistics textbook. While complex methods such as neural networks are discussed, issues of implementation are seldom mentioned. Nor did I think that this book would make a good first introduction to the area — although concepts were certainly introduced in roughly the order of difficultly, from Bayes' theorem in the initial chapters through to genetic algorithms in the final chapter, I felt the pace to be a little fast for a first encounter. Nonetheless, I greatly enjoyed the book, and would particularly recommend it to students whose interests lie in complementary areas. It provides a broad and easy-to-read overview of the field, enabling insight into how different concepts come together. While technical detail is lacking, the extensive list of references provided is more than sufficient for those who are curious.

Book details:

The intelligent web: Search, smart algorithms, and big data

Gautam Shroff

hardback — 320 pages

OUP (2013)

ISBN: 978-0199646715

About the author

Mary Fortune is a PhD student at the University of Cambridge. She is working on statistical techniques for analysing genomic datasets.