Data mining touches my life every day in the form of targeted marketing efforts. If I weren’t already aware of this, the point was brought home as I was listening to the radio the other day. Terry O’Reilly was being interviewed by Leonard Lopate. O’Reilly, a former advertising copywriter, currently co-hosts with Mike Tennant a radio program called “The Age of Persuasion” on the Canadian Broadcasting Corporation and Sirius Radio. They also co-authored a book called The Age of Persuasion: How Marketing Ate Our Culture.
Lopate: […] Now we’re hearing about things like data mining. And I guess those are the things that also scare a lot of people.
O’Reilly: Yeah. And I get that, too. I’m very ambivalent on that issue, Leonard, because on one hand, as an ad man, I want as much information about you as possible, not to pry, but I want to understand you so the advertising I create is relevant. On the other hand, I’m a consumer, too, and when I go and buy a pair of socks at a store and they ask me for my phone number it drives me crazy.
And I look at my daughters. I have three teenaged daughters, and they so easily give away their information online. And my wife and I are always so fearful of that. But then I realize that one of the reasons they do it is because they know instinctively that if they give away information, what they get back will be much more relevant to them.
Personalized shopping experiences
For quite a while now, when I go to the grocery store, targeted coupons appear with my register receipt. And, of course, when I shop online, I invariably see the phrase “people who bought [this], also bought [that]. But, according to an article that appeared in the Novelties column of the New York Times last year, I might also expect kiosks to appear in other stores where I shop. These stations would help me choose the right hair color and style, make-up color and application technique, and outfit.
Prototypes from Intel were department store oriented. Rather than wandering from department to department in a large store, customers would flash an identification card at a kiosk to enter the system. Based on prior knowledge of the customer’s purchases and information from current choices, new recommendations would be made. For the customer, shopping is more efficient; for the retailer, there is the opportunity to upgrade or increase purchasing.
The Virtual Mirror kiosk from IBM and EZface captures and stores an image of the customer along with a list of the products in which he or she was interested. By scanning a cosmetic product, the virtual customer can try it on and evaluate the new look.
I had clipped this article a year ago because of its connection to pattern recognition. Sadly, I have not, as yet, seen any cool kiosks in my favorite stores. I did, however, find the EZface Virtual Mirror online. (I wonder…how would I look with purple hair?)
Measuring a Nation’s Mood
If multiple measurements of a physical object can vary, imagine the difficulties involved in measuring an emotion like happiness. Peter Dodds and Chris Danforth, two applied mathematicians at the University of Vermont, have done just that.
In the interview segment quoted above, Terry O’Reilly tells Leonard Lopate that his motivation for using data mining techniques in his work is relevancy. Similarly, Dodds and Danforth say in their abstract:
The importance of quantifying the nature and intensity of emotional states at the level of populations is evident: we would like to know how, when, and why individuals feel as they do if we wish, for example, to better construct public policy, build more successful organizations, and, from a scientific perspective, more fully understand economic and social phenomena. 
So what did they do to obtain relevant data about population level happiness? In the paper cited, they studied song lyrics (from www.hotlyrics.net), song titles (from www.freedb.org), first person blog sentences containing the word “feel’ (obtained from www.wefeelfine.org ), and State of the Union addresses given by U.S. Presidents (from the American Presidency Project at www.presidency.ucsb.edu/). The words from these sources were given a happiness score based on the “psychological valence” assigned to the words by the Affective Norms for English Words (ANEW) study.
The blog study results showed that Election Day 2008 was the happiest day in the preceding four years, with elevated use of the word “proud”. On the other side, the day Michael Jackson died was among the least happy.
The methods are now also being applied to Twitter messages.
Pattern Recognition in the Media: Data Mining
by Linda J. O’Gorman (USA)
For more information:
Personalized shopping experiences
Virtual Mirror from EZface, Inc.
Measuring a nation’s mood
Joshua E. Brown, “If You’re Happy, Then We Know It: Research Measures Mood”, University of Vermont, University Communication, 23 July 2009
Margaret Bradley and Peter J. Lang, “Affective Norms for English Words (ANEW)”
“Persuasion”, The Leonard Lopate Show, WNYC, New York, April 19, 2010
“Thinking of Going Blond? Consult the Kiosk First”, The New York Times, March 28, 2009
Measuring a nations’s mood
“Measuring a nation’s mood”, Health & Science, The Week Magazine, August 21, 2009
 Dodds, Peter Sheridan, and Christopher M. Danforth, “Measuring the Happiness of Large-Scale Written Expression: Songs, Blogs, and Presidents”, Journal of Happiness Studies, 17 July 2009.