It's strictly based on plot keywords and all it uses it Solr's MoreLikeThis feature.

Clearly it works not so well.... Maybe its not one of the major components of the API but I think you should rethink it (maybe base it on genres and actors instead of keywords...). I'm trying to develop a prediction algorithm wich will recommend movies based on user ratings. The Similar feature is crucial part of this.

So what I'm actually asking is do you think you'll have time to deal with it?

I don't have any immediate plans on improving this. It is very low on my priority list.

Thanks anyway.

I said it before but I will say it again - you're doing incredible job!!

No problem. And thanks. I have many features to add still and am working on some higher priority issues. Cheers.

I think the problem is the movie's keywords. We can solve the wrong film's relations writing better keywords and the same that other related movies. For example, The hobbit has "orc" but the lord of the rings has "orcs". If the algorithm uses the title's words as keywords, maybe it will be better.

Is there any guideline to write keywords?

However, great job and thank you!!