More like this

[ChunChun]

Can managdex get a feature that suggests similar manga when you finish one / go to the page of a manga

This used to be the main way i found manga way back in the day, but now, rarely any website has this anymore

 

[seekermoc]

Probably because the similar suggestions many sites have aren't very good. The website doesn't know the contents of the manga beyond the tags it's labeled with, so you end up with the suggestions for "similar" manga that aren't even close to being similar. I hate this on many of the large manhwa group sites (looking at you Asura, et al).

The one way to make sure you end up with at least somewhat accurate similar suggestions is to do something like how myanimelist does it, where users have to manually link what they consider to be similar series, but that's a lot of work and subject to user trolling.

 

[yojacket]

The one way to make sure you end up with at least somewhat accurate similar suggestions is to do something like how myanimelist does it, where users have to manually link what they consider to be similar series, but that's a lot of work and subject to user trolling.

imagine being recommended porn while reading little shark's lol

 

[Grellow]

the thing about tags is that since they're not always accurate (or sometime too broad) and mangas are missing some from time to time, finding hidden gems that I like often ends up being hit-and-miss on the search page. I think the "best match" feature already works in a similar way that a "for you" page would work but it's still too janky for me to get good results from it. For exemple feature that takes in account what users who have read x manga also has in their "marked as read" for the last x days and then ranks the amount of time a title re-appears and shows it to you anonymously (could probably be done better but I'm just saying that from the top of my head)
(edit: could be somewhat costly but there's probably some solution you could find like only doing 1 check per user and removing it after x amount of time and such)

mod note: merged into thread

 

[Random01]

Currently, it feels like the scoring system & tags do not do a good job of helping you find series you might like, partially because tags are used so liberally. for example One Piece has 12 tags, Berserk has 15 tags, Fullmetal Alchemist has 19(!) tags. I'd love to see a recommendation tab for each manga, that suggests series that people who have liked this series also have liked.

(Algo should probably be weighted with some Bayesian inference towards more niche titles - having the algo for a dark fantasy series recommend Fullmetal Alchemist because it is the most popular fantasy series is not very helpful, instead it should recommend some fantasy series that might not have huge amount of fans, but many who like the original series also like the other one too)

mod note: merged into thread

 

[Basic_human]

Aight someone bring it up again, I once make thread asking about this recommendation but no response

Hope the staff able to do something about this, it's also make the help new comer doesn't stuck on home page only reading "popular new titles"

mod note: merged into thread

 

[Teasday]

Yeah, this would be really nice, and obviously it's come up before. I believe we've done a bit of testing on recommendation algorithms in the past (must've been during the v3 days) but I don't remember if there was anything we could've realistically implemented. Probably not.

Even if we had a working solution, it'd have to take an amount of computation power that I have no idea if we can or can't afford.

mod note: merged into thread