Can a machine recommend a (good) movie?

The two things that you need to know about Nanocrowd (Nano = one = you,  and crowd = many = the whole web) are a) that it is a recommendation search engine and b) that it uses sentiment analysis, although they prefer their trademarked name Reaction Mapping™ .

One way to appreciate what a search engine is doing is to go back, way back, to pre-Internet days (if you can).  You see, there  used to be stores called “video stores” filled with nothing but shelves full of movies.  The trick was to go inside and return with a movie in the shortest amount of time possible.

Some people would just wander up and down the aisles, and when asked if they needed any help would always answer, “no thanks, just looking,” and they meant it.  They would pick a movie based upon the picture on the cover and the plot summary on the back.  Chance of enjoying the movie? 50%.

Other people would rely on their friends.  Tricky.  I recommended District 9 (my highest rating – you really must go see it!) to a good friend and he (gasp!) didn’t like it!  This can put a strain on your friendship(s). A much better choice for him would have been to ask the video store clerk.

Clerk: Can I help you?

You: Yes, I’m trying to find a (good) movie.

Clerk: What kind of movies do you like?

You: Well, I like science fiction movies…

Clerk: Which ones have you seen (that you liked)?

You: Well, I liked Star Wars and Close Encounters.

Clerk: How about Alien?

You: Is it scary?

And so it goes, back and forth, until you conclude that Independence Day and Men in Black are both light sci-fi comedies with Will Smith (you like Will Smith because he was in some other movie), so you rent them both.

Now flash forward to present day and Nanocrowd wants to do all of that for you.  You start the process by naming a movie, an actor, or a director. I chose Zooey Deschanel. To test their technology I clicked on this icon and it brought up the “nanogenres,” three word combinations that try to express the dramatic range of this particular actor.

Now a word about what it’s doing when I clicked on “Go.” Nanocrowd is a true search engine that has crawled the web grabbing every single movie review that it can find – but that’s only the half of it.  This is not traditional keyword search where you have to type in “movie drama Zooey Deschanel” and then go to IMDB and start reading the reviews for all of the movies she’s ever done.

No, this is Web 3.0 folks, and for many people that means the Semantic Web, and that means that Nanocrowd “reads” those reviews and “understands” whether they are positive or negative reviews, and how the movie is described. The final result should be just like an all-knowing video clerk.

Using the phrases in the image above, I say, well I hate formulaic movies, and I don’t want a downer, so I’ll take the intelligent, complex movie with a twist. There is one “exact match” that stars Zooey, Abandon.  But in the same nanogenre it also recommends Saw II (based upon a large sample set), and I really must point that out as an indicator that Web 3.0 is still in early stages.  The clerk at my local video store would never hand me Saw II if I asked for an intelligent Zooey Deschanel-type movie! Now to be fair, there is a pop up box with a summary of Saw II which would keep me from clicking on “watch now.”

In the end, whether I drove to the video store or just logged on to Nanocrowd, I would probably end up watching Zooey in Abandon.  The obvious advantages of Nanocrowd are that I don’t have to leave my home, don’t have to worry about returning the damn thing on time, and that over time it will have processed essentially every review for every movie ever made, and that’s a tough act for even the best video clerk to follow.