Follow

I wonder if there is some way to automatically determine hashtag synonyms like art/mastoart or cats/catsofmastodon/catstodon...

If I look at a 1,000 posts with the art hashtag, 158 of them are also tagged with illustration, 146 with photography, and 140 with mastoart

On the other hand, among 1,000 mastoart posts, 533 are also tagged with creativetoots, and 295 with gameing (wtf?)

Cursed 

Cursed 

@Gargron depending on the purpose i'd try using a fancy regex to find like hashtags. you could also look for posts tagged with synonyms like #art #mastoart #etc

@Gargron well a number of folks will use synonyms, perhaps a "Related Tags" for tags commonly used together?
@Gargron otherwise it'd need to be a manually updated list in my experience.

@clarjon1 @Gargron ☝️ this. I think the best you can do without human intervention is try to rank related hashtags and display them when they reach a certain threshold of being used together. But that also kind of encourages people to use multiple similar hashtags to get them to be grouped.

A human-curated list (which I think is how some of the Twitter “trends” work) would probably be more immediately accurate for the covered hashtags, but would be a LOT more work.

@Gargron If that would make it easier to block all the cats, that would be purfect!

@Gargron
This deserves an auto tag feature.
But Heh, it will make already existing combinations even more strengtheneded.
Maybe we should suggest lists with similar tags.

@annika @Gargron that wouldn't exactly be equivalent to "dev" in general yeah

Cursed 

Cursed 

@Gargron admins manually linking hashtags together?

or you could disambiguate to something like wikidata. for example

"tag": [
{
"id": "wikidata.org/wiki/Q146"
}
]

@Gargron on a more extreme level you could rewrite the links for certain hashtags, e.g. instead of /tags/cats you use /tags/Q146 but eh, that's just overcomplicating it trying to retrofit onto existing Hashtags

@Gargron yes, there are natural language processing algorithm designed to do extract similarities between words or other features. It could be from a network of tag or something more fancy like word2vec

@Gargron I think there are different levels of this. Your example seems tricky but some would be simpler, like plurals. For example I see little reason for #fountainpen and #fountainpens to not be indexed together.

@gargron Speaking of new features: I think it would be great if you could set any link text in a toot (e.g. <a href="mastodon.social">favourite instance</a>). To avoid phishing, a popup would show the URL before visiting the website…

@Gargron Although it is not directly related to this problem, I would like you to consider emoji hashtags. cat / cats = 🐈, beer / bier = 🍺.

@Gargron Please don't, sir, I don't want to be hauled with the cats.

Sign in to participate in the conversation
Mastodon

Server run by the main developers of the project 🐘 It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!