Wednesday, April 9, 2014

SSR: Pattern recognition on Amazon (Not thee Amazon)

We've all seen on Amazon, after clicking an item, near the bottom of the page are suggestions for something else Amazon thinks you may be interested in purchasing. The little recommendation area generally consists of other things people have bought in relation to the item you're looking at. An example might be if you were looking at a table, what would probably appear in the recommendations are matching chairs. Not just because they were made to go along with that table, but because that's what most people who bought the table also bought with it. Hence this magical little recommendation area being named the "Customers Who Bought This Item Also Bought" Section.

Now stop, because if right now you're thinking: "What is this person talking about?! I've never been to the Amazon, but I'm pretty sure there's nowhere to shop around there!" Then you are currently thinking of the Amazon River, and not the online website on which people shop. I realize this little bit was probably unnecessary, but it's my way of recognizing that not everyone has, does, or will use Amazon. So...yeah.

But enough of tables, and tourists, because we're really here to discuss the recommendation system for books. Aha! I got you! You probably didn't think I was going to get to that! But I did. Unfortunately for examples I'm going to have to mention mainstream books (Or current bestsellers). Believe me it's not that I want to, but I feel that I won't be able to get the point across if I choose books not everyone has read. It's a sacrifice I'm willing to make.

The main system of book recommendations on Amazon is pretty much the system mentioned above of, whatever item your looking at, it'll show you what other people who bought that book have also bought. Usually other books by that author, other books in the series if it's a part of one, similar books, similar themes, just similar in general, books of or around the same genre, and if the book is mainstream, then other mainstream books a lot of people bought. You see where I'm going? For example, the first five recommendations for The Hunger Games include: Catching fire (Twice), Mockingjay (Twice), and Divergent. If you keep scrolling through the rest of the recommendations you'll see a few other well known books, a few classics, The Hunger Games movies, and merchandise. Now do you get why I had to use a mainstream book? Odds are if I used a not well know book, a lot of people out there wouldn't be able to see the connections a lot of the recommendations had to the book of subject.

My response is that it's a pretty good system. If I'm looking at a book series I like, it either gives me a good idea of something else I might like, or give me a quick link to another book in the series. But, like most "bullet proof" systems, it's not absolute. Sometimes items recommended seem to be completely unrelated and as random as finding a shopping mall by the Amazon river. Sometimes it's only slightly related, and sometimes you don't know where it came from. While looking at The Hunger Games recommendations, one of the first five items was a stack of note books with one of the generic Amazon pictures used, though the next time I looked at the same link it wasn't there anymore. A different time I was looking at a bored game for kids and one of the first items was...well? An adult item. My first reaction was somewhere between laughing and exasperation as to why that would be there in the first place.

I don't know if those, just happened to be what most people bought along with what I was looking at, or if it's just a minor site glitch, but you get the point, not everything in the recommendations is useful, or necessarily what you're looking for. Odds are most people don't even glance at the recommendations bar, let alone look through more than the first five. I usually don't, but now you know. As with most things, it has it's up's and it's down's.

Here's a link to go to Amazon. But not thee Amazon.

~Let the mayhem begin.

No comments:

Post a Comment