SEP "Movie Genie"

Assistant: Silviu Homoceanu
Classification: Software-Praktikum im Bachelor Informatik und Bachelor Wirtschaftsinformatik

And the winner is..."Movie Genie"!



KickOff will be on Wednesday, 11.04.2012 at 14:00 hours, room 218 (Informatikzentrum 2OG).



Today: These days if someone wants to rent a movie he doesn't need to leave home anymore. One can comfortably just rent a movie online from the many services available (e.g., LoveFilm, Videoload, Maxdome, etc.).
After a short while of using the service, its usually the case that users run out of known movies to watch, and have to rely on some recommender system for choosing new movies. Of course recommendations are not quite transparent, so users often end up asking themselves if the recommended movie would really be a good find. In order to decide upon the movie, one usually on ranking systems like IMDb.
But what usually happens is that most users won't be happy with the first recommendation, and go back to the recommender system for a second guess. And the process goes back and forth, until the user finds something interesting, or, more often than not, he gives up and chooses whatever movie to watch.

The good old days: Before Netflix and online movie rental took of, the user had to go to the movie rental store. Although not as comfortable as choosing your movie from sitting on the couch, this more traditional way had the advantage that one could speak with the trained personnel.
Typical dialogue snippet between a client and the video store personnel:


Movie Genie: Why not build that into a system? Of course a complete system, capable of reacting to unpredictable answers/questions, is not really possible. But what we could do, is to implement a system that can "read" queries about movies, written in natural language, as they would be addressed to a human video rental sales person:

The system should interpret the query, extract hard facts like the movie genre, and soft wished like a "good story" and generate a movie rank. Furthermore, feedback in form of 'I have seen this movie and liked it'/'I have seen this movie and dis-liked it' from the user will be considered: the so marked movies will be eliminated from the ranking, and the ranking will be restored considering what the user liked and did not like.

SEP presentation slides.


File mgenie.png07/02/12 6:52 pm61.01 KB
File dialog.png07/02/12 7:04 pm148.26 KB