5 Questions with Aaron Barnett, Director of Information Science for RealAge.com
|
Written By Reprise Media | August 14, 2006 | Share This
|
|
We recently had a chance to chat with Aaron Barnett of RealAge.com - you might know them from their only-slightly-scary test of the same name. We talked about how the site’s Smart Search feature is similar to a puppy (and yet so much more), taking the dread RealAge test, and yes, even thyroid concerns.
1. RealAge clearly takes pride in the fact that its health information is vetted by individuals before going into its database. How specialized are the people who review the articles, e.g. Would a piece about kidney diseases always get looked over by someone with a background in nephrology before being ok’d for the search results?
Although we don’t have nephrologists on staff, we do have a top-notch team of MDs and PhDs from a variety of science and medical fields on our Scientific Advisory Board. These are the experts who joined heads to develop the criteria that all materials OK’d for inclusion in Smart Search must meet.
Once these criteria were established, our research team went through extensive training to learn how to put them into practice. This evaluation process, combined with the intuitive technology behind our Smart Search engine, makes sure pages including misinformation, link farms, and redundant material all get weeded out. And, since we don’t include commercially-driven sources in the review process, users also get to avoid the growing number of website advertorials and well-disguised commercials.
If questions or issues come up during their review – on a specific topic such as kidney disease, for instance – our researchers can turn to the RealAge Scientific Advisory Board for their guidance and expert opinion.
2. RealAge VP of corporate communications Jennifer Perciballi has said that “Once people trust the results they have a tendency to click ads more frequently.” Has RealAge found a user’s trust in the results to be as important as the relevance of the ads in determining how often the ads are clicked on, and if so how is a user’s trust measured?
It’s no secret that if users like and trust content, they are more receptive to the ads within it. And since we operate on an advertising-based business model much like Google, Yahoo!, WebMD, and many other Web sites and print magazines, it’s important that our users place confidence in us and the information we provide.
One way we try to build trust is by making the line between advertising and editorial content very clear. In fact, we recently received the highest rating of “Excellent” from Consumer Reports in the category of advertising and sponsorships for our identification of paid information.
Another way we develop user confidence is by building the tools and information people need to protect their health and live life to the youngest.
The Web is chockfull of health resources – some good, some not-so-good, many full of misinformation – users say that they aren’t finding the really useful information. In fact, I just read a study by JupiterResearch that said although nearly 75% of Web users search for health information online, most of them report being disappointed with what they find. Only 16% claim they find the information they’re seeking.
That’s why we’re so thrilled that the revenue we’ve generated from ad placements is helping us provide our users with the smartest, most efficient way to find health information online.
3. Smart Search uses the ‘thumbs up’ and ‘thumbs down’ buttons to gauge how useful and relevant a given result is believed to be by a given searcher, but results that are bad for one search (results for “separation anxiety” when the user is interested in general anxiety, for instance) might be good for another (a later search specifically for “separation anxiety”). Do the thumbs ratings affect results for all searches by a particular user, or do they weight results more on a search-by-search basis?
Actually, it’s a combination of the two. Yes, query terms do carry weight in Smart Search, but keep in mind that much of what a user likes about an article is not clearly related to his query. That’s why Smart Search focuses more on what a user likes about his results overall than what the particular health topic or category is.
The intuitive power of our search tool really comes alive after analyzing feedback across a broad range of searches.
For example, a person with Graves disease who’s searching for fitness information may prefer resources that mention thyroid concerns even though he may never have searched using the specific term “thyroid.” However, Smart Search knows his search history. It knows he prefers articles that include thyroid-related issues, so it presents his results accordingly.
Smart Search also tracks how our users are ranking resources overall. This lets people see which specific articles and information people like the most – information many Web users are apparently craving. The JupiterResearch study I mentioned before also found that almost half the people who seek health information online want to know which sites other people thought were useful.
4. One of the features of the personalized Smart Search on RealAge.com is the ability to recognize whether a user likes or dislikes the “language patterns” found in articles. Could you further explain exactly what’s meant by a “language pattern” and how Smart Search determines how a user feels about it?
Sure, but first I want to talk a little about how language patterns are just one of the features that Smart Search considers.
Smart Search uses an intuitive algorithm to analyze several features on every page a user rates. Doing so helps it identify the kinds of articles the user will like in future searches. Each time a user gives a thumbs up or thumbs down rating, it’s added to a body of evidence on his or her preferences.
So by “language” we mean the way in which information is presented on a Web page.
And, the more a user searches, the smarter Smart Search becomes about that user’s preferences. Basically, the ratings teach the engine to be smarter about users’ likes and dislikes, and to arrange the results accordingly.
It’s kind of like training a puppy. As it grows, it learns to avoid doing what will lead to hearing reprimands like, “Bad dog!” in favor of actions that lead to being praised with, “Good dog!”
5. RealAge is known for the RealAge test, which makes predictions about a person’s health based on their self-reported current habits. Was there a time when you took the test and it produced a surprising (or even alarming) result?
For me, taking the RealAge test - answering questions about my health, history and behaviors - drew attention to things I didn’t realize would affect my health and wellness.
I know I don’t exercise enough and could certainly eat better, but what I didn’t expect from the test was how it inspired self-reflection beyond my physical health and into my emotional and social well being as well. The simple act of answering questions about how many close friends I have and what sort of regular social activities I do made me wonder, am I keeping myself healthy socially? I learned that by developing my network of family and friends I could make my RealAge up to 3.5 years younger.
Topics: Interviews, Search: Vertical |
Alexa Adware
Adware - Alexa toolbar on purpose? I thought for the simple affiliate links or alexa adware. Axacta