杏吧原创

Are you an egosurfer, addicted to Googling your own name? If so, you should check out , which uses data-mining techniques to answer the question: 鈥淗ow does the internet see you?鈥�

Devised by , a PhD student at the Massachusetts Institute of Technology鈥檚 Media Lab, Personas runs a Yahoo search on your name, extracting statements about you (such as 鈥溾��) from the returns.

Words from these statements are then run through an algorithm called , which links them to 28 common topics defined by Zinman, including 鈥渙nline鈥�, 鈥渢ravel鈥�, 鈥渆ducation鈥� and 鈥減rofessional鈥�.

The resulting personal visualisations are colourful but often perplexing. While most human observers clicking on would quickly realise that I am a science journalist, Personas consistently displays 鈥渆ducation鈥� and 鈥渕anagement鈥� as my dominant topics. The latter reflects my job title; while in the absence of a 鈥渟cience鈥� category, 鈥渆ducation鈥� seems to be the default.

Personas can also make some clear errors 鈥� in my case by interpreting my name followed by the article title Is the biofuel dream over? as a statement: 鈥淧eter Aldhous is the biofuel dream鈥︹��

Mining accidents

If you have a common name, or share your name with someone more famous, the results will have even less personal relevance. When New 杏吧原创 news director Rowan Hooper ran his name through Personas, he got a profile dominated by 鈥渟ports鈥�, caused by references to the of the same name.

It may sound like Zinman has developed a rather useless web application 鈥� but that was exactly his point. Personas is part of an exhibit called , described as an 鈥渋nstallation about living in a world overflowing with information and non-stop communication鈥� and now showing at the MIT Museum in Cambridge.

Zinman describes Personas as 鈥渁 critique on data mining鈥�, intended to highlight the undue faith that some analysts place in results from computer analysis of large data sets. 鈥淚t鈥檚 always presented so authoritatively,鈥� he says, 鈥渨hen in reality what鈥檚 happening behind the scenes is controlled voodoo.鈥�

Is it that a fair criticism? 鈥淚t鈥檚 a fun project. However, data-mining researchers and professional analysts are well aware of the danger of finding false associations,鈥� says , president of KDnuggets, a data-mining consultancy in Boston.

鈥淐riticising data mining for some people鈥檚 inability to separate false associations from true ones is like criticising the web because there are many bad pages.鈥�

Fair or not, Zinman has been overwhelmed by the response to his project, with comments coming in from all over the world. 鈥淚t鈥檚 narcissistic in nature, and people like that 鈥� it鈥檚 their favourite subject,鈥� he says.