Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
Tags

Is there a way to blacklist tags when searching?

A topic by Nex T. Kig-Yar created Jan 03, 2024 Views: 566 Replies: 29
Viewing posts 1 to 4
(+2)

I don't want to scroll through hundreds and hundreds of either sex games, horror games, or some fetishistic amalgam of the two.

Can anyone tell me how to blacklist tags, so I won't be bombarded by breasts and beasts? 

(1 edit) (+6)

For NSFW you can uncheck from your user settings:


After doing that, use this link to remove horror games: https://itch.io/games?exclude=tg.horror

Unfortunately "?exclude=" does not seem to work with multiple tags, but that may be enough to get rid of the two most common types of games in here from your search (I do that myself too).

(1 edit) (+1)

Someone on Reddit found a way to include multiple exclusions. For example:

https://itch.io/search?q=Train?exclude[]=tg.horror&exclude[]=tg.adult

EDIT

Finally home from work so I could do a little research. Back in 2017, Leafo implemented a quick fix in response to the post Can I use exclusion filters?. This fix allows users to exclude one tag from the browse page. There is no way to exclude multiple tags.

I also looked into how to exclude tags in Search. I'm no web developer, but from what I can tell you can't exclude tags from your search results. The website just isn't designed to do that.

So https://itch.io/games?exclude=tg.horror will hide all games tagged as "horror, but https://itch.io/search?q=game&exclude=tg.horror will not.

hum... unfortunatelly, that does not seem to work, it shows results for as if you put "Train?exclude[]=tg.horror" in the search box.

Deleted 117 days ago

It does not work for me either.
https://itch.io/search?q=Train?exclude[]=tg.horror&exclude[]=tg.adult
and

https://itch.io/search?q=Train?exclude[]=tg.horror

They show exactly the same games, and in both cases there are games with the adult tag in the result.


That person just claimed it works. It does not. The implementation would have to treat exclude like an array and loop through it's elements. It does not, as you cannot even do a single element exclusion with []=.

Also, that exclude thing does not seem to work, if not logged in. Would explain the complains I see sometimes, that it does not work. Single tag exclusion does work. But not in search. Search does not even limit itself to games or assets, you cannot specify what to search. That might change in the future, as the whole search system will get an overhaul. Maybe they even implement official tag exclusion, maybe even multiple tag exclusion.

Yes? That is what I clarified with my edit?

I'm not sure how to reply to this comment. I think I have said all that I can say about this topic, so I will no longer be replying in this post.

Deleted 117 days ago

That's the thing. You are not looking at search results, when looking at the browse section. I doubt that all those unwanted games show up, when looking for a search term.

Using browse for random games that are unspecific, yet "popular" is not reliable. There are complain threads about the ranking being unstable, unfair and or wrong. And other complain threads about tags being used incorrectly. And of course, complains about things not showing up in actual search.

Deleted 117 days ago
(+1)

A naive version would simply remove the entries that don’t match, with nothing taking their place, so search results could be empty a lot of the time.

A non-naive version might be against Itch’s ToS, but I am not sure.

Deleted 117 days ago
(+1)

Actually, looking at the search results page, tags aren’t shown there either, so even a naive version would need extra HTTP requests to find them.

Oof, yeah, that may put what I'd like to do outside of the scope of my programming skills currently. Thanks for checking that out

(+1)

I believe there was some css magic script that would do basically this. But it would "simply" load every game page in the grid and look at the tags and then filter it out. Not very nice to the servers.

It would be a trivial matter to filter stuff that actually is on the browse page. Tags are not, unfortunately.

@-moz-document url-prefix("https://itch.io/games") {
div.game_cell:has(a[href*="paste_full_link_to_game_inside_quotes"]) { display: none }
div.game_cell:has(a[href*="//or_ignore_a_publisher.itch"]) { display: none }
div.game_cell:has(a[href*="itch.io/or_ignore_games_starting_with"]) { display: none }
}

I'm hesitant to suggest doing anything that would ultimately cause issues for the website. 

I have noticed that many horror projects include key phrases like "horror" and "scary" in the title or tag line. Would you be able to hide projects with such key words or phrases using CSS Magic Script?

If it is in the result page of browse it is as trivial as I posted. Use a custom css and you can filter it out. This is all client side. It just tells the browser to not show it.

If you use Chrome, just right click an element and select inspect. Then a new window will appear, showing the source code of the web page. The inspected element is highlighted. Click the highlighted part and right click again. Select hide element. Look at the website again.

This is, what display none does in the code snippet.

The filter I used is a wildcard matching that will fit the link to the url that will be in the game cell. That is the grid element that displays the game thumbnail and title. Matching the url was easiest, but it should be possible to filter other parts of the game cell as well, and hide it.

I can recommend the browser addon Stylus for that.

INB4 this becomes machine learning classification, which sounds interesting in itself.

itch.io doesn't have a large enough team to go through every project and ensure it is using the correct tags

Tags are not ensured by itch staff at all. And there is no such thing as correct tags. To elaborate: for any given game you could "correctly" apply more than the allowed 10 tags, so how would you rank the tags to get a "correct" 10 tag combination. Steam seems to do it by popular vote of the players.

You might complain about blantantly and misleading tags with the report button, but I believe that is more for music categorized as games or for obvious tag spamming. Most of tag usage is in the eye of the beholder, and the one selecting the tags is the devs and they will have different opinions than many users.

I bet I would not even tag half the "horror" games here as horror.

Projects are not immediately indexed

While search and browse are affect from not being indexed, I was talking about search. Browse is not search. If you select a tag in browse you will see all available items indexed with that tag, with a dubious popular ranking that is not really accuratly reflecting popularity. Or rather, they do not tell how they calculate and how they define popularity. There is random in it and boosts/weighting for new/random stuff despite there being a separate new&popular page.

So to say it again: if you use search you will not see all matching results. What you get is truncated and randomized. So yes, there were complaints about indexed games not showing up in search. There is no guarantee to show up in search, even for a specialized search term.

Just try searching for horror (not tag horror, entering horror into the search box). Currently there are 72 results, despite there being 44000 games with the tag horror. To make it more complicated, just because you tag horror does not mean you will show up with a search for horror. You need also mention "horror" in the game description, better in the name of the game, at the very least. The search algorithm does not look at tags.

For example, I can currently find your "cottage cats", but not the "monster girls" assets. Despite using partial title match search term and only 28 results showing up for the monsters and 60 for cats.

Deleted 117 days ago

Let me say it again: search is truncated and random. But not random per search. More like per day or whatever. I assume the servers build a nightly search index, so it does not take a minute for each search.

You will show up on one day and not on the other. It is supposed to cycle the relevant items. Your dog or was it cats even was on place one of search. But itch will not tell how it works, because people will try to game the system.

I remember a thread where the dev complained that the game was not showing up in search despite searching for the exact title.

Do you have absolute proof that this is the case? 

That it is truncated and random, as in, not the same every day and you do not know how it is ranked?

I do not know if staff or faq said it somewhere. I seem to remember the mod explaining it some time ago in response to a complaint that an indexed item could not be found by exact title search, but I do not find the thread among all the threads about games not indexed. Maybe I misremember the details. That staff will not tell how those things work exactly you can read several times.

But you can observe and test this for yourself. The capping of results is obvious. And if you look now and in a few days and will have a different ordering and different items you have proof that the results change. What you do not know, is the mechanism of the change. Is it pure rng. A cycling of a pool of available items. A combination of those and other factors, inlcuding popular ranking?

We do know that the search was recently overhauled and will be again. There is an announcement by leafo in developer updates subforum about this.

What I still do not really get, if the search is just optimized for title search or if it can only search/find title matches. I would expect it to not be a pure title search, but maybe it is. It certainly does not help that there are so many games on itch that for practically every word you will have games with that word in the title.

Moderator(+2)

In principle, searching by the exact title should work, but many games have extremely generic titles. Sometimes dozens of games are named essentially the same. (Can't think of an example right now, sorry.) When that happens, a search might surface other games than the expected one.

(+1)

Tetris. There are at least 123 games named exactly Tetris. What I do not know, if there are more games named Tetris, that did not appear in the search for tetris. It yields 180 items. There are 1600 games tagged tetris, but only 71 of them are named exactly Tetris.

I do know that searches will not give all results of titles containing the word(s). I found two examples created by pop shop.

Maybe it is different for single word searches matching single word titles. Those examples had multiple words and the search was for a subset of the words.

(+2)

I have not been able to find any instances of an itch.io staff member directly stating that search results are truncated (the only person I found who suggests this is you, oddly enough). What I did find was multiple instances of the staff clarifying that search results are optimized for searching for exact titles.

"The box is optimized for exact title searches, that's what. Searching by keywords is a matter of luck." (No Time To Play, Why is the search function of the store so limited?)
I've said this in other posts, but essentially search is suited for finding things by direct title, not by topic. Tags are how people browse things by topic. People do not browse by topic...through our search engine. Hope that helps. (leafo, My game doesn't show up in search or tags)
"Our search engine is optimized for Direct title search since that is how the majority of people use search. For users who are browsing for new games, we try to push them into our browse pages with tags, which are surfaced on the top of the search results. This means that the text in the body of your project page and any tags you add will not become search queries for your project. Tags are only for our browsing pages, which are separate from search." (Limitations of Search, Getting indexed on Search & Browse)

So no searching of description. Maybe tag searching will be added, if I understand leafos announcment correctly. The wording is vague.

But if search results are not capped, why could we both not find your monsters asset pack? I tried several of your packs, most I can find with the search button.

I will also not find the pack Slimes | Pixel Asset Pack by searching for slimes. I did get a result of 72 items. It can be assumed that there are at least 73 items with that word in their title. This is capping/truncating.

And the results for that monster thing was capped at a lower number. It is still capped at 61 if you activate adult items. 

The capping for tetris is 181. Of the 1600 games tagged tetris, most have tetris in their name. So I searched some of those games with tetris in their name along other words in the search 180 search results. Many do not appear there. And only 123 games are named exactly tetris in the search result, so there should be 60 games with tetris plus other words. Only 60.

Those 60 are what I assume to be a subset of matches that will rotate in some manner to expose every game once in a while to such a search. I am not confident, if it will always return all exact title matches, but I am confident, that it will not return all matches where multiple words are involved.

(+3)

Now I think it *is* something for the administrators to look into that multiple tags can be excluded.

(+1)

Absolutely. Excluding tags is a pretty standard feature on many websites. It makes me wonder if there's a specific reason why they haven't implemented it yet

(+3)

I did some research actually. The sticky is now closed, but I posted it there. No good answers. Bascially I wanted to know any game site, any at all that would have that feature. Excluding Steam. And even on Steam, this feature is so hidden, that I used Steam for years and did not even know it was possible.

The fact is, most game sites do not even have tags to begin with.

Maybe it is different for asset stores?

As for the reasons, the single tag exclusion was implemented rapidly after a feature request in community some 5-6 years ago. The fact that it was neither removed, nor implemented or expanded on the site as a documented feature instead of undocumented tells me something.

The calculation of actual usefullness, actual demand, accuracy and such vs. cost of implementing it in current framework is unfavorable. Or in other words, if it were cheap to implement, it would have been done.

Steam has a fixed pool of tags. And they are very accurate because of curators and or popular vote. So tag filtering is not only cheaper for Steam, but more usefull. 

(1 edit) (+3)

They might not, but do they have as much diversity and freedom in tagging?

I see Itch very much as the “booru of video games”, and a good tagging system is vital for those.

I researched this for a whole two minutes, so correct me if I am wrong.

Those items on those boorus are tagged by a dedicated enthusiast doing the tagging for many, many items and usually in an environment with a rather limited and standardized local flavor of the usual tags seen on similar boorus. Not only is the tag pool small, limited or not, the method of tagging is kinda by some formula.

The contrast on itch is this: there is like 100k persons tagging 10 items. Each with different views and opinions about word and tag meaning. And no guideline on the method how to categorize the item and putting importance on different aspects of the item. They just freely associate their item with tags and have perspective bias, because they created the item and also because they might select tags they think will make the item more popular, while an enthusiast would emphasize different aspects of the item.

I did not understand your first sentence, because I do not know to whom the "they"s referenced and what might not be done. If you were talking about steam, you can suggest tags for games. It reads "Popular user-defined tags for this product:" and shows a short list of tags you can expand to up to 20 tags where you can suggest tags yourself (from a fixed pool of tags).

(4 edits)

Each booru does things its own way, but I can’t think of one with a small tag pool. Most are actually giant. IIRC Danbooru actually allows anyone to create a new tag, or edit the tags of other users’ posts. As of 6 years ago it had approximately 317,760 tags according to here.

There’s also volunteers who moderate tag misuse, but that’s much harder for video games than it is for single-frame images.

Now I’m not saying Itch should erase the tag limit or 100% become a booru, but if it already allows people to make up their own tags, then a higher limit and tag exclusion is basically expected.

Do they at least have guidelines how to tag? Because I believe that is a major issue on itch. There is no step-by-step checklist what to consider for a game. A concept in that direction is the prompt to select a genre from the available list of genres. If one fits, that should be selected as the main genre. If several fit, they can be selected as additional tags.

Also, you can tag different types of information. Like the content, theme, topic, stuff about the player character or lack thereof, features of the game, even things about the developer ("indie" comes to mind. While implicit on itch for all games, that is a popular tag on Steam), art style, general game mechanics, type of story, and so on. So which to chose from? If you ask many people, there will be a trend what seems to be catching about the game. If you ask one person for many games, a certain standard will emerge. But if you ask devs to tag their own games without some guidance, it will be hard to find games, even with tag browsing and exclusion.