In order to make trakt appeal to all users, a performing, personalising, watch time ranking, finding individual movies/series both for individual users & trakt’s crowd, A.I. algorithm is necessary. Personalising recommendations by putting individual users into a content cluster through analyzing movies/series keyword metadata & implicit engagement.
While developing a performing recommendation algorithm, simultaneously develop api latency performance.
Does that chart mean twitter, Facebook, instagram etc in the end appeal to all users???
God no. It’s all becoming more and more trash.
Also I use couchmoney for recommendations. Try it out!
1 Like
‘it finds people with similarly weird tastes to you, and recommends films and TV that they like more than average people.
It looks for “diamonds in the rough”: lesser-known titles you may have not heard of that you may like. It recommends these more than well-loved blockbusters that you probably know about.
It uses tens of millions of film and television ratings to compile its recommendations for you.’
Two things why couchmoney has less potential to appeal to all users next compared to trakt:
- Couchmoney’s algorithm does the opposite from watch time ranking. Trakt’s algorithm, if it doesn’t already, already comes closer to watch time ranking
- Couchmoney recommends based on rating. In order to get closer to appeal to all users by analyzing explicit engagement, likes outperform ratings. However, the majority of engagement is implicit. Therefore, in order to appeal to all users, implicit engagement must be analyzed.
No offence, on the topic of your preference towards twitter/Facebook/insta/etc., appealing to all users is, as it says, about all users, and therefore your individual taste is irrelevant.
And that is were those sites fail. Instead of focusing on the individual it focuses on a group. They try to push stuff that would “appeal” to everyone and it just ends up being the same regurgitated name brand garbage that you always see. Instead of looking for what would appeal to you. I am very much against this. Sorry 
I personally would rather see closer integration with couchmoney.
1 Like
This does just what I mentioned, by putting individual users into a content cluster instead of focusing on the individual, it focuses on a group identity and puts them in one. For example because two individuals went to comic con that does not mean that they both like Friday the 13th and Lord of the rings. Or if they like scifi that they both enjoy Star Trek and Avatar.
As I said, it’s in order to appeal to all users, and, as I said, no offence, your individual taste is irrelevant. You want nothing about appealing to all users.
People comment to this post, by telling about their individual preference. If that’s what you want, then go to couchmoney, because couchmoney’s recommendations are for individuals, which will never appeal to all users. Trakt is developing a social recommendation algorithm, to give personalised recommendations. As far as I know, trakt desires to appeal to all users.
For those who do not want such algorithm, which will appeal to all users, it just means trakt is not the place for you. You’re better off with couchmoney.
I wrote a reply that goes point by point in response. But then I realized that you must be using an add-on/app to watch content and the add-on/app uses Trakt lists to recommend “totally legal content” to you. (Not judging)
For the record Trakt has nothing to do with any of that.
I realized that you must be using an add-on/app as you have not used this site even once but have a VIP account(not even one play/list etc.). You also keep mentioning watch time, engagement and likes which Trakt does not have. And you think couchmoney works separate from Trakt. etc.
If I am incorrect I am sorry, but I have a feeling I may be right on this one.
LOL no.
I’m fine by you denying that individual taste is irrelevant. If it means that you’ll stop telling me otherwise, I’ll admit that individual taste is relevant. Happy now?
Besides, it doesn’t matter whether trakt currently analyzes watch time and implicit engagement (likes are explicit. Again, I’m talking about implicit). What matters is whether trakt wants to appeal to all users. Oh, wait, no, I was gonna agree with you, so you would stop telling me otherwise. So, yes, there is no such thing as trakt wanting to appeal to all users. And there is no such thing as analyzing watch time and implicit engagement to appeal to all users.
Also, what difference does it make whether I talk about a trakt add-on or trakt directly? It’s both about trakt’s recommendation algorithm.
Also, if you’re not judging, then I’m a monkey.
It means you are barking up the wrong tree. You should be reaching out to the add-on/app developer not Trakt. Also I guess you are a monkey then LOL. Sorry, I probably should have added a winky face. Tone does not translate well via text. 
The thing with recommending movies/series, is that in order to appeal to all users, it’s dependent on copyrighted movies/series. When you say you’re not judging, then basically you are pulling the leg of netflix’s audience, which is all the movies/series users. So, therefore I say, if you’re not judging, then I’m a monkey.
About the add-on/app, what does it do other than what trakt itself does? If trakt recommends without analyzing watch time and implicit engagement, does that mean a trakt add-on/app is gonna recommend through analyzing watch time and implicit engagement?
Trakt does not stream any content. 
Trakt is much more like IMDB than Netflix. It is made for keeping track of what you have watched. Not as a content platform/streaming service. And as such, can never analyze watch time and implicit engagement the same way a streaming service could.
It could perhaps. You would have to ask them about it. What software were you using that brought you here? It could help shine some light on the subject at hand.
I was hinting at pirated movies/shows. And that I am not judging if you were. 
That is what I meant by “totally legal content”. 
You’re killing me smalls. 
I’m talking about the trakt stremio add-on. Personalised recommendations are currently non-existent on stremio.
So, is it not possible for trakt to analyze stremio’s implicit engagement as an add-on? How about a reversed add-on? So one that pulls data from stremio. Can an add-on work that way? Or can trakt in another way pull data from stremio?
Also, if torrents are not legal, then why do judges not say differently? So, you admit that torrents are legal, only you put in sarcasticly? I still say that if you’re not judging, then I’m a monkey. Oh, and you can respond with another facepalm emoji. You won’t need to, if you’re clear on what message you wanna give.
Also, you took my question differently. I asked what a trakt add-on/app is gonna do differently from trakt itself.
I already knew that trakt ain’t a streaming service, and that trakt catalogues movies/series.
Since trakt catalogues content, and already has (under-performing) personalised recommendations, I figured trakt will make the performing personalised recommendations through the stremio add-on.
So, tell me, what now? What is gonna do the job? What possibilities are there to analyze stremio’s implicit engagement?
Trakt does not make that add-on. Nothing Trakt could do would help you in any way.
Again, you take my question differently, I ask about trakt, not the add-on. I only mention the add-on, because that’s how stremio can connect to another host. Basically anything the host can do, stremio can pull.
If it’s at all possible for trakt to analyze the implicit engagement, then stremio can pull that information through the add-on. So, it comes to this: is there any way in which trakt can analyze stremio’s users?
Trakt can recommend to its users. An add-on can do this to users using the trakt stremio add-on. First, trakt itself analyzes watch time and keyword content metadata, from wherever. Cinemeta already analyzes watch time and ranks movies/series, and shows this to stremio through the add-on. Second, the stremio trakt add-on dev applies trakt’s recommendation algorithm to stremio’s users. So stremio users get movies/series recommended. The more the trakt add-on learns about someone’s taste, through the add-on, the better it learns someone’s preferred keywords, and the better it can recommend.
question, Trakt has a recommendation system?
is that just like the “If you like…, check out…” section at the bottom of every individual tv/movie page?
I’m not aware of trakt recommending. According to their api, they do. The stremio add-on lists ‘trending’ and ‘popular’. I don’t use trakt itself. I only know about the stremio add-on.
About a ‘if you like, check out’ section, that is not a recommendation. It’s a related section. If trakt refers to that as a recommendation, then they just put words down that don’t belong there
Recommendations are at the bottom of your website dashboard. The iOS app has recommendations card in the home tab as well. API also includes these for other apps to use.
Currently, it looks at people you follow and what they’re recommending. We plan on expanding this algorithm.
1 Like
Wait currently my recommendations are based on the recommendations from people I follow? Lol, I follow no one and dont intend to change that in the future. When you say youre planning on expanding this algorithm, what exactly does that mean? Does that mean I can expect to soon have recommendations based on something else?
Will it eventually work like how I imagined it would here: New 50 Item Recommendation Cap? - #17 by cloverhoney?
1 Like