Artificial intelligence predicts which films will succeed and which not

© Provided by FREED Information Technology Networks S.A.

The anxiety of the actors, director and production company about whether their film will do well in the box office when it is released may be diminished in the future, as artificial intelligence promises to reduce uncertainty (and) in this area.

On the other hand, some films will probably never be shot if the forecast is negative, not because of the intuition of a producer, but the predictability of a machine.

Although it certainly cannot predict the future, a new algorithm analyzes written scripts and estimates how many chances a movie will have for the audience and the critics, or rather, to make a mess of it.

South Korean researchers at Sungkyunkwan University presented their system at a conference in Florence, Italy, according to “Science”.

The system was “trained” with summaries of 42,306 scripts from around the world, which correlated with how well each of these films did commercially (several scripts were collected from Wikipedia). The artificial intelligence system evaluated each scenario with the help of mechanical “sentiment analysis”.

One key finding was that successful films have much more frequent alternations of emotions than what was considered a failure. This rotation is considered a better predictor of success than whether the film has a pleasant beginning or a good ending. Most importantly, according to the researchers, is the film to offer a frequent alternation of emotions.

The system scores each movie with a sentiment “score” and finally decides whether the movie -thriller, comedy or drama- will prove popular or not. Predictability is not perfect, of course, but it’s better than guessing at luck.

Researchers are optimistic that in the future, once they have improved their performance, they will be able to predict even the future earnings of a movie, which will help investors decide whether to bet on the film or not.

The question is whether such a system will prove to be good for art and for the work of scriptwriters, directors and actors.

Researchers prefer to see the positive side (eg that their system will be unbridable and give equal opportunities to everyone, known and unknown, provided they have a good idea presented in writing), but they are not certain that everyone will see it that way.

If, however, viewers of the future are spared two wasted hours of watching “rubbish”, the system will have earned “points”.

See the scientific publication at:
https://www.aclweb.org/anthology/W19-3414

sofokleousin.gr
Source: Sofokleousin
(Visited 9 times, 1 visits today)

2

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.