Jump to content

Finding a proper solution for movie recommendations

horstepipe

By horstepipe
in Non-Emby General Discussion

 Share

Recommended Posts

horstepipe

horstepipe   356

Posted
Posted (edited)

Hello
I’m pretty sure most people here suffer from it. As bigger ones movie collection gets, as harder it is to find something what to watch next. So while digging around what 3rd party service is best to help me out with that, I stumbled about FilmFish
https://www.film-fish.com/

Their lists look really promising. So what if Emby Team gets in touch with them, developing a FishFilm Emby server plugin which syncs the movie lists via tags or genres to Emby Server’s database?

IMO this would be a real killing feature.

If that won’t work for some reasons, what about setting up something similar only for Emby usage?
Sure, lots of handcraft at first, but the data is there. Maybe the data could be created/cared by the community?

Maybe other people have some ideas/workflows how to handle this first world problem lol

Edited by horstepipe
  • Like 1
Link to comment
Share on other sites
adamstewiegreen

adamstewiegreen   147

Posted
Posted

This site is a little lacking in titles.  There's nothing from either the Koker or Apu trilogy - and I wouldn't even consider those obscure.

 

Personally - just like music recommendation sites - I find movie recommendation sites more miss than hit.  I'm not much of a "genre" watcher.  There is no one genre that gets me excited, I just enjoy good storytelling.   So a list of "Great Southern Gothic" movies probably won't intrigue me just because I liked Wise Blood. 

 

I read about what influenced directors as well as their recommendations.  I also take notes anytime a movie I'm unfamiliar with is mentioned.  My wife frequently asks me "where did you find out about this movie?" and 99% of the time I have no idea because my growing list comes from all over the place.

 

This method gives way broader material.  For instance, I searched for Kill Bill.  I know one of the biggest influences for QT was a movie called Lady Snowblood - but I couldn't find it in any of the lists (nor clan of the white lotus).  It really show be in the top 10 recommendations.  Generally though, those types of movies don't show up in recommendations, only movies like the new Hari-kiri or Crouching Tiger - neither is anything like Kill Bill, their similarity is purely surface level (e.g. martial arts movie).

 

Somethings just need to be handcrafted.  And part of the fun (for me at least) is the hunt.

Link to comment
Share on other sites
horstepipe

horstepipe   356

Posted
  • Author
Posted

I agree with that. But these lists provide a good base you can start working on. And for the mainstream users they’re very useful.

Link to comment
Share on other sites
  • 5 months later...
horstepipe

horstepipe   356

Posted
  • Author
Posted

I just want to bump this.

Maybe somebody got some ideas now about other services, possibilities...

Link to comment
Share on other sites
JasonNalley

JasonNalley   25

Posted
Posted

I can't say I'm against it, the "Recommended" only goes so far, it would be wonderful if it was much deeper, instead of surface level stuff as adamstewiegreen mentioned.  I would bet money that a User curated database could be used for something like this.  But to get enough data to be meaningful would probably take a really long time, and it would take a large userbase who are completely dedicated to doing it.

For instance:

Initial data gathering
-Popular Movies by amount of times watched over entire userbase (Trakt probably has this or similar information)

Secondary data gathering (This is where the dedication comes in)

-Survey's that show a movie or movies and asks someone to list 5 movies that they feel are similar or related

-Scrape or Gather information from IMDB's related movies field
-Scrape or Gather other list infromation from other sites

 

This project would be a rather large undertaking, it would involve a large mass of people, and would likely take some machine learning/AI to be really effective.

I'm not saying it can't or shouldn't be done, just that getting this kind of information and packaging it in a user friendly and easily digestible manner takes some doing.

Link to comment
Share on other sites
adamstewiegreen

adamstewiegreen   147

Posted
Posted

The first big mistake in recommendations is trying to recommend something "similar".  This is why computer generated recommendations are always going to be shallow and will never have that special ... je ne sais quoi...  Because computers need something to compare.  Either a genre or an actor or something of that sort.  Computers don't recommend based the feeling of a movie or the style.

 

I stick by my previous comment.  If you love watching movies put in a little hard work and research and you'll discover movies that you would have never seen otherwise.

 

That's why I really dig services like Criterion Channel and why I'm very annoyed with Netflix, Hulu, Amazon.  It's quite obvious (IMO) that Criterion has hired people to personally assemble movie collections based on themes, directors, mood, genre, actors etc and haven't relied on computer algorithms to do the heavy lifting.  Netflix takes the exact opposite approach and everything is done by computer which is obvious when scrolling through there selection (which - if I'm being completely honest - has been getting better over the last month or two).

  • Like 2
Link to comment
Share on other sites
Jdiesel

Jdiesel   1113

Posted
Posted

This is why I have wanted better Trakt integration for a long time now. Sure it is a way to keep track of what you watched, but it is also a great way to get recommendation on what to watch next. You can see what is popular and trending at the moment or look at user curated lists. I agree that you need to do your own homework to find interesting and new content.

 

https://trakt.tv/search/lists

  • Like 1
Link to comment
Share on other sites
Spaceboy

Spaceboy   2492

Posted
Posted (edited)

The first big mistake in recommendations is trying to recommend something "similar". This is why computer generated recommendations are always going to be shallow and will never have that special ... je ne sais quoi... Because computers need something to compare. Either a genre or an actor or something of that sort. Computers don't recommend based the feeling of a movie or the style.

 

I stick by my previous comment. If you love watching movies put in a little hard work and research and you'll discover movies that you would have never seen otherwise.

 

That's why I really dig services like Criterion Channel and why I'm very annoyed with Netflix, Hulu, Amazon. It's quite obvious (IMO) that Criterion has hired people to personally assemble movie collections based on themes, directors, mood, genre, actors etc and haven't relied on computer algorithms to do the heavy lifting. Netflix takes the exact opposite approach and everything is done by computer which is obvious when scrolling through there selection (which - if I'm being completely honest - has been getting better over the last month or two).

but surely Netflix, like Apple when it comes to music, are using user profiles to say well, user a watched films x, y and z and user b has watched x and y so let’s recommend him z. It can’t just be as simple as comparing genres like you say. But they have the benefit of massive data sets Edited by Spaceboy
Link to comment
Share on other sites
Jdiesel

Jdiesel   1113

Posted
Posted

I'm pretty sure Hollywood has been running a neural network on past popular movies to greenlight new projects for years now. Probably why every blockbuster seems so familiar.    

  • Like 1
Link to comment
Share on other sites
horstepipe

horstepipe   356

Posted
  • Author
Posted
That's why I really dig services like Criterion Channel

 

 

Thanks, didn't know them before. Do you have more like these?

Link to comment
Share on other sites
adamstewiegreen

adamstewiegreen   147

Posted
Posted

Thanks, didn't know them before. Do you have more like these?

Criterion Channel is an off-shoot from the now defunct FilmStruck (when it closed the likes ok Nolan, Speilburg and DiCaprio etc wrote letters to Turner to ask them to keep it open). To my knowledge there is no other streaming service like it. Closest would be Fandor or Mubi - but neither have the depth.

  • Like 1
Link to comment
Share on other sites
WilhelmStroker

WilhelmStroker   96

Posted
Posted

Looks like Plex already has support using webhooks

 

https://blog.trakt.tv/plex-scrobbler-52db9b016ead

 

That is their scrobble solution which only works one way so it won't pull any data in from Trakt (unlike the trakt addin in Emby)

Besides, the Trakt recommendations I get are not great anyway as my taste in films is all over the place so I'm with adamstewiegreen on this.

If only Criterion was available in the UK.....

Link to comment
Share on other sites
adamstewiegreen

adamstewiegreen   147

Posted
Posted

I'm not in USA either ... I took a risk years ago and bought a lifetime sub (which they don't offer any more) to playmo.tv.  I've used it on and off for services like Mubi, Fandor, Amazon Prime (US), Criterion Channel, Hulu, BBC, US network streaming etc...  It's been super reliable and new services seem to get added at launch.  I probably wouldn't pay monthly for it but the lifetime made it worth it in the long run...

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share
Go to topic listing
×
×
  • Create New...