What are you watching for your next movie night?


Imagine a movie with a female vampire, with a lot of chasing and stabbing, some axes and swords, maybe someone wearing leather pants who died at the end from leukaemia, and above all that it’s a sequel, based on a video game and you can’t recognize any of the actors. This is called a recipe for a disaster movie.

No one wants to spend a movie night with popcorn and drinks ready, to watch a bad movie. But how can you tell if it’s bad or good in advance? Well, thanks to Data Mining, now you can!

In the context of the participation to the Linked Data Mining Challenge [1], My colleague (Emir Munoz) and I ran an experiment on a dataset of 2000 movies to find movie features useful to predict whether a movie is good or bad. With the help of Linked Open Data: DBpedia[2], IMDB [3], Metacritics [4], we discovered that the most important feature is the critics review. If the critics love a movie, it’s great. That simple.

But what if I don’t like the critics’ point of view? They can complicate things sometimes and make a nice simple movie turn out to be a horrible one. That’s where more features come to help. The next important thing in a movie – which we all usually look for- would be the plot of a movie. A simple analysis from our study case showed that good movies plots contain keywords like: “family relationships”, “moral ambiguity”, “very little dialogue” and interestingly, people like drama. Deep dark drama, so keywords like: “frustration”, “crying”, “sleeplessness”, “schizophrenic”, “melancholy” are in the recipe of a good movie.

In addition, we discovered that the most popular genre for a movie is “PG-13”, which is something between “NC-17” movies with highly sexual/violent content and “G” movies for general audience and children appropriate content. Also, movies with famous awarded actors and directors are most likely to succeed if they contained the previous features.

In Summary, you can discover many interesting stuff with mining Linked Data sources, as we did with best movies features, which you can use later on with the aid of Machine Learning techniques to create a classifier for good/bad movies that could save you a disappointing movie nights and much more.

Our approach [5] will be presented in ESWC’15 [6] as part of the Know@lod workshop. See you there!

[1] www.knowalod2015.informatik.uni-mannheim.de/en/linked-data-mining-challenge/

[2] www.dbpedia.org

[3] www.imdb.com

[4] www.metacritic.com

[5] http://knowalod2015.informatik.uni-mannheim.de/fileadmin/lehrstuehle/knowalod2015/papers/KnowLOD_2015_submission_11.pdf

[6] http://2015.eswc-conferences.org


One response »

  1. Pingback: And the Oscar goes to.. | Letters from Galway

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s