Location based services yield more accurate rankings than algorithms used by top review-ranking sites

It’s always fun to find cutting edge research that—when applied—has the potential to disrupt established business models. The subject matter of this research (.pdf download) has the potential to do just that: disrupt review-ranking services like Yelp and Zagat.

The researchers point out several negatives with these review-ranking services: they’re subject to expert bias, spoofing the system (i.e., marketers posting the reviews as opposed to genuine users), etc. As an alternative model the researchers developed a service called “SocialTelescope” which leverages location based services to track user interactions.

SocialTelescope used geo-tweets generated by Twitter users who also used services like Foursquare to discern popularity of locations. The basic premise is that the more check-ins a place has the more popular it is. Users are not taking any other action aside from checking-in to demonstrate their vote for a place.

Understandably, there’s also a qualitative issue to address: even though someone checked in they may still have hated the experience and, thus, a review site is more accurate (the premise being that a purposeful and explicit action taken by a user is more meaningful and accurate). However, the researchers tackled this issue by focusing on a inferred methodology:

 SocialTelescope does not consider all users to be the same, when computing popularity of a location. Instead, users are assigned a score based on their expertise on the search keyword. For a given search key- word, SocialTelescope assigns expertise scores to users as the number of times that user has visited any place that matches the search keyword. The intuition behind computing expertise is that, say, when ranking restaurants that serve seafood, people who visit lots of seafood restaurants can be considered seafood connoisseurs, and so their choices can be given a greater weight.

The researchers compared their results against the review-ranking sites and found that SocialTelescope was at least as accurate as these sites. The potential for disruption resides in the fact that a new company can now leverage location based services at a fairly low cost and return results that are just as accurate as their competitors. Thus, the barrier to entry in this space has been significantly lowered. I highly recommend everyone who is an analysis junkie to read the entire paper, as the researchers have done an excellent job describing their methodology and made great use of visuals.

Photo credit: TechCrunch

 

Leave a Reply

Your email address will not be published. Required fields are marked *