Thursday, 5 December 2013

Mining the social web [book review]

The social web is a phenomenon of our times when the web started to reflect our interactions and communications.

Who speaks to whom, who says what about what, how many people talk about what. Information that marketers want, the information underlying the altmetrics movement in academia, and it would appear,  the various security agencies.

Mapping out interactions is not new, the Republic of Letters project did much the same by analysing the correspondence of eighteenth century savants, but it is both the scale of the social web and the complexities of the analyses made possible by cheap processing power.

This book covers the major social networks such as Twitter, LinkedIn, Facebook, and Google+, with an emphasis on Twitter. The author also discusses mailbox corpus creation and analysis, and the analysis of semantic web data, and also interestingly, GitHub as a social platform.

This book is not a book for the dilettante. More than half the text consists of Python code and the reader really needs to work with the code examples to gain full value from the book. The book also provides a rapid introduction to OAuth, and ranges over topics as diverse as simple text analysis, cluster analysis, natural language processing,  and the use of applications such as MongoDB.

This is however a very good book for anyone seeking to work with the social web and would serve as a very useful primer or as a textbook for a module on data mining. The code examples are clear and nicely structured, making them easy to follow and work with.

Mining the Social Web: Data Mining Facebook, Twitter, LinkedIn, Google , 
GitHub, and More 
O'Reilly Media (2013), Edition: Second Edition, Paperback, 448 pages 
- also available as an ebook in most common formats

No comments: