One of the things I love about where I live is that there is constant opportunity to learn interesting new things. My town has several universities that pull in interesting lecturers on a variety of topics, as well as an engaged and intellectually generous business culture.
Last week I had the opportunity to hear Sep Kamvar, director of the social computing group at MIT’s Media Lab, speak at seminar hosted by the University of Vermont’s Complex Systems Center.
Kamvar discussed a number of topics, including Dog-Lang, a computer language he is creating that assumes programs will be social, require connecting to a wide variety of APIs, and utilize asynchronous state management. He also provided some insight into the guts of Google’s personalized search ranking algorithm, which he helped create while at Stanford. Unless you love linear algebra as much as I do, this discussion would make your eyes glaze over.
The topic Kamvar approached that I want to dig into today is data and humanness.
Kamvar has created a series of projects aimed at discovering the “people behind the documents” of the Web. These projects explore the words people use to describe their feelings online.
To grossly oversimplify, what the “I Feel Fine” and “We Feel Fine” projects do is scour the Web for phrases that include “I feel.” These statements are stored and sorted.
Additional data can be attached to these simple statements of feeling. For example, many people who publish things online include information in their profile that identifies age or geography.
Kamvar and his associates created an API (application programming interface) so that others can make use of the data that is being gathered by the We Feel Fine project.
By putting all of this data together — statements of feeling and demographic stuff like age and geography — a number of insights can be gleaned.
For example, words that describe feelings can be categorized, sorted and ranked. By doing this, we learn about the words people use to describe their feelings. We might also learn whether there is more or less happiness in online writings.
We can tie in information about geography and discover whether there is any relationship between happiness and geography.
Also, what other words occur in statements about feelings? For example, if someone writes about feeling happy, what else is in that same statement? Does happiness start to relate to something specific, or a certain topic?
Kamvar noted that “excited” and “peaceful” are two words often used in conjunction with expressions of happiness. Where it gets particularly interesting is that when that data is examined to see if there is a pattern in which people are excited and happy, or which people are peaceful and happy.
It turns out that, using about 70,000 expressions of happiness, there was some correlation between excited happy vs peaceful happy based on age. Older people who write online are more “peaceful happy” and younger people who write online are more “excited happy.”
All very interesting, yes. But where do we turn back to marketing, technology and real estate you ask?
First, I am often told how real estate transactions often occur around life events. Some of these life events are predictable patterns of behavior. Understanding the kind of happiness that audience feels might be useful.
For example, data like that generated by We Feel Fine might help us distinguish the emotional cues we want to include in branding or marketing efforts targeting first time home buyers, or buyers who are downsizing after the kids move out.
This is especially important for those who are developing marketing efforts for segments that are not themselves: Millennial agents and brokers pursuing their grandparents as they retire, or Boomer agents and brokers going after the Millennials and their younger siblings as they begin to buy their first homes.
Data like this helps us to see beyond our own built-in emotional filters, and reach customers at their own emotional set point.
Another important thing to consider when examining data projects like We Feel Fine is that you can develope these kinds of observation points yourself. Your own Web platforms can begin to assess for emotional cues if you build them to do so. Your own team can learn to consider this sort of data as important if you train them or reward them accordingly.
Even if you don’t build your own platforms and projects to consider these kinds of sentiment data possibilities, using an existing one like We Feel Fine can help your understand more about the minds of customers.
So much of “social” is about emotion. And as social continues to weave itself into the online world, an understanding of emotion and how it works becomes increasingly important.
The tools and data sets are there to help us expand beyond our own personal, anecdotal “feelings” about feelings. We can know happiness and then we can work to deliver it.
Gahlord Dewald is the president and janitor of Thoughtfaucet, a strategic creative services company in Burlington, Vt.