Thursday, February 25, 2010

Keeping Out the Trolls: Relevancy in User-Generated Content

lunch_relevance_feb10.jpgIn the summer of 2008, J.R. Johnson sold Virtual Tourist to Expedia for $85 million dollars. While Johnson seems like the type of laid back Los Angeles entrepreneur that would take some vacation time, his quest for relevancy had him launching a new community the following March. Lunch.com is Johnson's attempt to cut through the noise that has proliferated since he first started in the user-generated-review space in 1999.



Says Johnson, "When I started, people asked me why anyone would want to read an amateur review. Now the environment has changed and there's even pay-per-post happening across the net. Virtual Tourist is travel-specific and you increase relevancy by picking a niche topic on which to base your community. With Lunch I'm trying to solve something new." Johnson spoke to ReadWriteWeb about some of the ways he's managed to ensure that his community is more than just search engine bait.


Sponsor



1. Identity and Interest: In addition to requiring that individuals use their real names and specify topics of interest within the site, Lunch uses an interesting member matching system. Similar to OK Cupid, users rate a series of topics and the "similarity network" matches them with like-minded members. In order for a spammer to target a specific user, they'd have to answer multiple questions in the same way as their target and trick the matching algorithm into displaying a higher percentage of compatibility. In this way, spammers are deemed less relevant while passionate users are matched by the percentage of overlapping interests.



lunch_relevancy_feb10.jpg



2. Opinion History: Johnson explains that just because people share common interests it does not mean that they share common opinions. Even at ReadWriteWeb we've seen Republicans and Democrats converge on the same comment thread in completely different ways. Lunch lets users view opinion ratings, past reviews and popularity rankings. From there you decide whether or not to follow others or look for additional commentary.



3. Top Review Ratings: The Lunch users rate each other and can review that history of rating over time. Ratings can be about a specific topic, on a specific review and on a specific user. In some cases a contributor with lackluster cooking reviews produces one standout piece about a specific type of food. It's important to be able to find those gems and weed out the reviews that are less relevant to you as a community member.



4. Existing Networks: Lunch allows users to pull information from their social graph via Facebook Connect in order to follow existing friends. Users can track topic reviews, member reviews and article reviews created by friends. I was actually surprised to see how little I had in common with my own social graph in terms of topic interest; however, where we converged was our opinions on other reviewers.



5. Frequent Contributors: By listing the top contributors to a community, Lunch is able to ascertain those with the most interest in a specific topic area. Johnson gives the example of Wikipedia's community as one where frequent contributors are also an indication of topic-based expertise. Explains Johnson, in Wikipedia, a community member that takes ownership of a page and can be seen as one of the page's top contributors (without deletions) is likely to offer more relevant content than those that do not contribute often. Coupled with opinion ratings this adds an additional level of user legitimacy.



While Johnson continues in his quest to thwart spammers and trolls, his site is already flourishing. For those attending SXSW Interactive, he is hosting a March 16 panel entitled, Trolls Suck, where he will continue to explore ways to ensure that thoughtful online contributors remain the majority in community sites.


Discuss





http://bit.ly/a633QC

No comments:

Post a Comment