1/ What’s my idea:

My idea is to add a filter in comments. This would allow studio hosts or project owners to filter what can be commented by activating predefined filtering options or by entering words in a custom filtering field

2/ What could be filtering options:

Filtering options could be:
-No links: Prevents users from posting any link in comments
-No @user: Prevents users from mentioning other users in comments
-No capital letters: Prevents users from sending messages written mostly in capital letters (to avoid “shouting”)
-Modified length: Allows the owner to set a new maximum character limit for comments

//The owner wouldn’t be able to define a limit higher than 500 characters.

-No special characters: Prevents excessive use of special characters (e.g. “!@#$%^”)

//A custom limit could be set by the owner.

-No nonsense: Detects and blocks comments that contain meaningless sequences like “asfjkld”, “zzzzzzz”, or random keyboard smashes. This would work by blocking comments with:
-Too many repeated characters or words (e.g. “aaaahhh” or “hi hi hi hi”)
-Random sequences of letters with no vowels or recognizable words (e.g. “dfgtrb”)
-Overuse of filler characters like periods or commas

3/ Custom filtering:

The custom filtering would allow the owner to add specific words to a “blacklist.” Any comment containing a word from this blacklist would be blocked and couldn’t be sent

4/ Honesty:

-Filtering options activated by the owner would be visible to all users
-The custom filtering “blacklist” would also be visible to everyone and could be reported if it is misused

//The blacklist would be reviewed during project or studio reviewing (if such reviewing exists)

5/ After defining the options:

-Once the filtering options are defined, the owner would be able to modify them at any moment
-Once the custom filtering is defined, the owner would also be able to modify it. However, modifying the custom filtering would make the project or studio “unreviewed”

//This means the project or studio would immediately disappear from public pages.

answers to your questions may be here:
https://scratch.mit.edu/discuss/post/8330586/
https://scratch.mit.edu/discuss/post/8336649/
https://scratch.mit.edu/discuss/post/8340608/
https://scratch.mit.edu/discuss/post/8347154/
https://scratch.mit.edu/discuss/post/8416531/

I like this idea, but there's a problem. What if someone filters negative comments about their project or studio?

(to avoid “hurling”)

whats that

hotcrystal wrote:

I like this idea, but there's a problem. What if someone filters negative comments about their project or studio?

also, people may unnecessarily add swear words to the filter; scratch already does that for them

How would the filter detect “nonsense”?

hotcrystal wrote:

I like this idea, but there's a problem. What if someone filters negative comments about their project or studio?

Even if it’s sad, and that in my opinion, it’s lying to yourself, nothing would be against Scratch rules.

Scratch rules wrote:

When commenting on a project, remember to say something you like about it, offer suggestions, and be kind, not critical.

——————

Cool_kid092 wrote:

whats that

Sorry, wrong word, I fixed it

8to16 wrote:

also, people may unnecessarily add swear words to the filter; scratch already does that for them

Yeah, even if it’s not a problem, a reminder could be displayed during setting the custom filtering

Za-Chary wrote:

How would the filter detect “nonsense”?

The “nonsense” filter would work by using a set of rules (trigram list) to detect patterns that are commonly associated with meaningless or random text. These rules could include:
1/Repetition of characters or words:
-Detects messages with too many repeated letters (e.g. “aaaaaaahhh” or “zzzzzz”)
-Detects comments with repeated words (e.g., “hello hello hello hello”)
2/Unusual letter sequences:
-Detects text that doesn’t resemble real words, such as random key mashing (e.g. “sdfgjkl” or “lkjhgfdsa”)
-Uses simple logic to detect a lack of vowels or meaningful syllables
3/Excessive filler characters:
-Detects overuse of punctuation or filler symbols (e.g. “………………..” or “?!?!?!?!?!?!”)

I hope I answer to your question, feel free to ask me if needed!
precision of trigram list added later due to a post of @AHypnoman mentioning it
Edit: fixed a misspelling

Za-Chary wrote:

How would the filter detect “nonsense”?

This isn't a perfect system, but trigram checkers can usually distinguish keyboard smashing from words.

It works by comparing every three characters of a comment to a list of common trigrams (three letter combinations), and reporting what percentage of the characters in the comment match at least one trigram.

So if you had a trigram list with only ‘the’ in it, you would get scores like this:

'then' - 75%
'the cat' - 50% (assuming spaces are ignored)
'asjkl' - 0%

And I assume there would a minimum percentage for how much of the comment has to match a trigram for it to be posted.

Bump

everwinner64 wrote:

The “nonsense” filter would work by using a set of rules (trigram list) to detect patterns that are commonly associated with meaningless or random text. These rules could include:
1/Repetition of characters or words:
-Detects messages with too many repeated letters (e.g. “aaaaaaahhh” or “zzzzzz”)
-Detects comments with repeated words (e.g., “hello hello hello hello”)
2/Unusual letter sequences:
-Detects text that doesn’t resemble real words, such as random key mashing (e.g. “sdfgjkl” or “lkjhgfdsa”)
-Uses simple logic to detect a lack of vowels or meaningful syllables
3/Excessive filler characters:
-Detects overuse of punctuation or filler symbols (e.g. “………………..” or “?!?!?!?!?!?!”)

Could any of this be a problem for those who write non-English languages?

Za-Chary wrote:

Could any of this be a problem for those who write non-English langauges?

Probably. A solution could be to implement language detection before applying the filter. This would ensure that the rules are adapted to the linguistic structure and norms of each language supported by Scratch, the risk of false positives could be significantly reduced

The language detection could be done by using existing tools like franc-min or CLD3 (JS libraries), which are capable of identifying the language of a given text, these tools are generally reliable. Once the language is detected, specific rules (trigram lists) for that language could then be applied. Although implementing this solution would require some effort and optimization to maintain performance, but it’s feasible.

i mean, i usually type «asdjkdhdjsak what did you do there????» or similar things, which is not necessarily spam

jmdzti_0-0 wrote:

i mean, i usually type «asdjkdhdjsak what did you do there????» or similar things, which is not necessarily spam

I never said it’s spam, it’s just very probable that you can’t post this comment if the user activated the « no nonsense » filter, as there’s one word without any sense

Bump

8to16 wrote:

hotcrystal wrote:

I like this idea, but there's a problem. What if someone filters negative comments about their project or studio?

also, people may unnecessarily add swear words to the filter; scratch already does that for them

It may just be adding minor swears, though, like “crap” (for some reason Scratch turns it into a * so I had to bypass) or “frick”, which the ST has said that they will never add to the filter. Also, can the studio host/project owner really add anything?

everwinner64 wrote:

1/ What’s my idea:

My idea is to add a filter in comments. This would allow studio hosts or project owners to filter what can be commented by activating predefined filtering options or by entering words in a custom filtering field

2/ What could be filtering options:

Filtering options could be:
-No links: Prevents users from posting any link in comments
-No @user: Prevents users from mentioning other users in comments
-No capital letters: Prevents users from sending messages written mostly in capital letters (to avoid “shouting”)
-Modified length: Allows the owner to set a new maximum character limit for comments

//The owner wouldn’t be able to define a limit higher than 500 characters.

-No special characters: Prevents excessive use of special characters (e.g. “!@#$%^”)

//A custom limit could be set by the owner.

-No nonsense: Detects and blocks comments that contain meaningless sequences like “asfjkld”, “zzzzzzz”, or random keyboard smashes. This would work by blocking comments with:
-Too many repeated characters or words (e.g. “aaaahhh” or “hi hi hi hi”)
-Random sequences of letters with no vowels or recognizable words (e.g. “dfgtrb”)
-Overuse of filler characters like periods or commas

3/ Custom filtering:

The custom filtering would allow the owner to add specific words to a “blacklist.” Any comment containing a word from this blacklist would be blocked and couldn’t be sent

4/ Honesty:

-Filtering options activated by the owner would be visible to all users
-The custom filtering “blacklist” would also be visible to everyone and could be reported if it is misused

//The blacklist would be reviewed during project or studio reviewing (if such reviewing exists)

5/ After defining the options:

-Once the filtering options are defined, the owner would be able to modify them at any moment
-Once the custom filtering is defined, the owner would also be able to modify it. However, modifying the custom filtering would make the project or studio “unreviewed”

//This means the project or studio would immediately disappear from public pages.

answers to your questions may be here:
https://scratch.mit.edu/discuss/post/8330586/


idk if this if a duplicate, I found nothing about what I suggest

The first two would be a bad idea, those features on here exist for a reason. The rest of them I'm a bit mixed about.

hotcrystal wrote:

Also, can the studio host/project owner really add anything?

No, they could add some of words you said and other ones, but in my opinion, they wouldn’t be able to add anything or to censure, as it would break, I think, the « be honest » rule. However remember that if someone doesn’t want you write a comment where he’s the owner, he will be able to delete it. if you think I make a mistake, feel free to tell me

Raw_Productions wrote:

The first two would be a bad idea, those features on here exist for a reason.

Sorry but I’m not sure to correctly understand, could you explain what you want to mean please?

AHypnoman wrote:

Za-Chary wrote:

How would the filter detect “nonsense”?

This isn't a perfect system, but trigram checkers can usually distinguish keyboard smashing from words.

It works by comparing every three characters of a comment to a list of common trigrams (three letter combinations), and reporting what percentage of the characters in the comment match at least one trigram.

So if you had a trigram list with only ‘the’ in it, you would get scores like this:

'then' - 75%
'the cat' - 50% (assuming spaces are ignored)
'asjkl' - 0%

And I assume there would a minimum percentage for how much of the comment has to match a trigram for it to be posted.

i saw a video about how animal crossing uses (with other checks) to determine whether a letter is nonsense or not (affecting the response)

Easy to add, just find the code that filters comments and add a JSX code to add custom things and scratch WWW to be able to filter it for that studio (which isn't hard, I believe you would just make it so if it matches the ID in the URL.) make a new API for that if needed, add buttons and JSX to trigger the custom filter add and remove functions, then get JSX code to only show to studio host or manager (I didn't read the OP, also we can just copy the same code that happens where only studio host can edit title and stuff.)

Bump :]

everwinner64 wrote:

-No @user: Prevents users from mentioning other users in comments

This would effectively block people from replying as replies automatically mention the user that posted the comment replied to.