Posts Tagged ‘algorithms’

False Positives: The Problem of Fake Customer Reviews

Friday, August 26th, 2011

Psst..hey have you heard? Someone’s putting fake positive reviews for companies. There’s a black market for people who are willing to write fake positive reviews of products, places or services. Since more and more people are looking online for reviews of products, places and services, the demand for good reviews is growing.  As more and more positive reviews are being written, the demand for more positive reviews increases. It’s a positive feedback cycle.

How to tell if a review is fake or not

So what about these fake reviews? Well most of them are hard to spot. Luckily for businesses, a team of researchers have come up with an algorithm to determine if that review you got on Yelp is fake or not.  So, how did they determine the algorithm? Well the long answer’s in the link I gave you. The short answer is this. The fake reviews tend to be more narrative, have vauge descriptions about whatever they are reviewing and tend to use the words “I”‘ and “me”‘ often.

So why need an algorithm at all? Well take a look at this. Don’t read the descriptions and ignore the highlighted words for a minute. Just read the review. Could you tell it wasn’t real? No? Well neither could I. The Cornell researchers have an explanation for this. Since humans have been communicating face to face for about 60,000 years, it’s much harder for us to pick up clues about deception in things such as reviews. You know this to be true, if you’ve ever had a misunderstanding over email.

So how do you prevent fake reviews? You can try setting up a review guideline. Try asking your customers to put the name of the person who helped them, describe the product or service in detail and point out the parts that they liked the best or had some problem with. If the review has passion, chances are it’s a real review.

A brief bit about negative reviews

Then there’s the flip-side, negative reviews. These are often posted by competitors or people who really don’t like the business. Surprisingly,  the trend is more towards writing positive reviews. But, that’s not to say that negative reviews aren’t important too.  Negative reviews can help you figure out what went wrong with that new pasta recipe you were trying out last Friday night. So keep an eye on those too. Also, if you answer negative reviews really quickly and fix/compensate for the problem, your bad review might just change to a good one!

Tweet This Post

Posts Tagged ‘algorithms’

Algorithms — They're Not Just For Big Businesses Anymore

Monday, October 19th, 2009

Thanks to the internet, online buyers and sellers have access to a lot of data to help guide their buying and selling decisions.  Finding the average price of a 1984 Chevy Citation (my first car) is a couple of clicks away.  But the problem with data is that it is everywhere and nowhere at the same time.  As I mentioned in a previous post, online reviews often suffer from grade inflation, and determining a reliable trend really requires thousands of transactions — not really possible for individual buyers and sellers.  So though it is easier than ever to gather information, that has just resulted in information overload.  This isn’t really a unique problem.  Big businesses have faced this issue for quite some time.  What’s their answer?  Scoring models with sophisticated algorithms.  Or stated more simply, math.

Though our interest in math prevented us from being very popular in high school, it has allowed us to create complex risk models for many types of decisions that big businesses face.  So why not use our nerdy interests to help small businesses sell online and provide comfort to shoppers that would like to buy from a small business, but isn’t sure of the shopping experience he/she will face?  That’s exactly what we did.  I can’t get into the details of the scoring m0del, or others (much like the popular kids in highschool) will just steal our homework.  But in general terms, KikScore takes public information about the online business (e.g. site security, traffic, links, domain registration, privacy policies, consumer reviews) and joins it with permitted information about the business owner (e.g. financial stability, public records)  — ultimately creating a tailored recommendation of the shopping experience to be expected.  This helps potential customers feel more comfortable doing business with a lesser-known small business.  Math…leveling the playing field for online sellers and bolstering trust for online shoppers.  Take that, Geography!

Tweet This Post