I can see mine, but, ironically, most of them are not relevant to my interests at all. I'm not sure what sort of algorithm they're using, but it's not working for me.
Have you rated many movies and TV shows for the system to use to make recommendations?
It's still not working. It's a shame, because I usually find good movies, new or forgotten, through the feature. Right now, I have 2000 movie favourites and no recommendations and 20 TV favourites, but the recommendations contain the shows I already added to my favourites. The recommended TV shows don't show as being in my favourites from the recommendations page, but they do show in my favourites.
That's weird, that it works differently for different folks. I'm having the same issue that Zürich Gnome is having: nearly all the recommendations are things I really don't want to watch. I have indicated lots of favorites, but the recommendations are just crazy. Does anyone know what sort of algorithm they're using to make recommendations?
Two things, why @MatchooW isn't seeing any data I am not sure about. I'll have to manually dig into it. I'm pretty busy with other things at the moment though, so I'm not sure when I will be able to find time to look into this.
With regards to the recommendations, it's an old system that relies on favorites. I have an open ticket here to look at moving to be more rating based. We've got a lot more ratings on the past few years ago this is a much better option today then it was back in 2015.
Travis, thanks for the clarification. But I'm not at all sure that migrating from favorites to ratings will help much, if at all. I think the problem is how the system uses the data to produce recommendations. So it's really about the "recommendation engine" you are using, e.g.:
Collaborative filtering
Content-Based Filtering
Hybrid Recommendation System
And about what sub-type of one of these you are using. Before I dropped Netflix--having seen just about all I wanted to see after several years and deciding I could get most of it cheaper--their recommendation engine used to drive me nuts w/its almost total inappropriateness. Eventually it seemed to run out of good recommendations completely.
And Netflix's engine is supposed to be rather complex and sophisticated. Same thing w/Amazon's engine, which almost never shows me anything I would be interested in buying. Maybe there's just a limitation to current recommendation engines. The only one that seems to work really well is Google's. But those guys are in a league of their own, apparently.
Travis, thanks for the clarification. But I'm not at all sure that migrating from favorites to ratings will help much, if at all. I think the problem is how the system uses the data to produce recommendations. So it's really about the "recommendation engine" you are using, e.g.:
Collaborative filtering
Content-Based Filtering
Hybrid Recommendation System
And about what sub-type of one of these you are using. Before I dropped Netflix--having seen just about all I wanted to see after several years and deciding I could get most of it cheaper--their recommendation engine used to drive me nuts w/its almost total inappropriateness. Eventually it seemed to run out of good recommendations completely.
And Netflix's engine is supposed to be rather complex and sophisticated. Same thing w/Amazon's engine, which almost never shows me anything I would be interested in buying. Maybe there's just a limitation to current recommendation engines. The only one that seems to work really well is Google's. But those guys are in a league of their own, apparently.
Yes, Google IS in a league of their own. I think they're the only ones who have figured out the whole recommendations thing and taken it to a higher plane. Other recommendation engines pale by comparison.
Maybe Travis will eventually give us some insight about the recommendation engine they're using here. I think he's got a lot of other stuff that's more important to deal with right now.
Zürich Gnome 的回复
于 2019 年 05 月 08 日 11:03下午
I can see mine, but, ironically, most of them are not relevant to my interests at all. I'm not sure what sort of algorithm they're using, but it's not working for me.
Have you rated many movies and TV shows for the system to use to make recommendations?
MatchooW 的回复
于 2019 年 05 月 09 日 5:01上午
About 2000.
Philippe LeMarchand 的回复
于 2019 年 05 月 09 日 6:18上午
I thought the recommendations were based on "favourited" films/shows rather than ratings. Could well be wrong, though!
MatchooW 的回复
于 2019 年 05 月 09 日 6:29上午
No, you're right. Right now the recommendations screen is telling me to add more favourites, even though I have 2000.
Zürich Gnome 的回复
于 2019 年 05 月 10 日 11:37下午
Yes, that's what meant. The Favorite designation is on the same page as the rating designation.
MatchooW 的回复
于 2019 年 05 月 17 日 6:38上午
It's still not working. It's a shame, because I usually find good movies, new or forgotten, through the feature. Right now, I have 2000 movie favourites and no recommendations and 20 TV favourites, but the recommendations contain the shows I already added to my favourites. The recommended TV shows don't show as being in my favourites from the recommendations page, but they do show in my favourites.
PT 100 的回复
于 2019 年 05 月 18 日 12:24上午
That's weird, that it works differently for different folks. I'm having the same issue that Zürich Gnome is having: nearly all the recommendations are things I really don't want to watch. I have indicated lots of favorites, but the recommendations are just crazy. Does anyone know what sort of algorithm they're using to make recommendations?
Travis Bell 的回复
于 2019 年 05 月 18 日 1:25下午
Two things, why @MatchooW isn't seeing any data I am not sure about. I'll have to manually dig into it. I'm pretty busy with other things at the moment though, so I'm not sure when I will be able to find time to look into this.
With regards to the recommendations, it's an old system that relies on favorites. I have an open ticket here to look at moving to be more rating based. We've got a lot more ratings on the past few years ago this is a much better option today then it was back in 2015.
Zürich Gnome 的回复
于 2019 年 05 月 18 日 4:19下午
Travis, thanks for the clarification. But I'm not at all sure that migrating from favorites to ratings will help much, if at all. I think the problem is how the system uses the data to produce recommendations. So it's really about the "recommendation engine" you are using, e.g.:
And about what sub-type of one of these you are using. Before I dropped Netflix--having seen just about all I wanted to see after several years and deciding I could get most of it cheaper--their recommendation engine used to drive me nuts w/its almost total inappropriateness. Eventually it seemed to run out of good recommendations completely.
And Netflix's engine is supposed to be rather complex and sophisticated. Same thing w/Amazon's engine, which almost never shows me anything I would be interested in buying. Maybe there's just a limitation to current recommendation engines. The only one that seems to work really well is Google's. But those guys are in a league of their own, apparently.
Sixties Holdout 的回复
于 2019 年 05 月 24 日 5:06下午
Yes, Google IS in a league of their own. I think they're the only ones who have figured out the whole recommendations thing and taken it to a higher plane. Other recommendation engines pale by comparison.
Moondoggie 的回复
于 2019 年 05 月 24 日 5:14下午
Maybe Travis will eventually give us some insight about the recommendation engine they're using here. I think he's got a lot of other stuff that's more important to deal with right now.