Minecraft PC IP: play.cubecraft.net
>:(
Well, "jew" is more often used in an offensive manner than those other terms. There's nothing wrong with "jew" in and of itself, it's just how it's commonly used by people that's an issue (and people don't typically use the words in your list in a way meant to offend). Also, "jew" doesn't automatically mean a follower of Judaism : P (going off of your list...).
I have nothing to do with the rules, just trying to explain why I think those are the rules...
ps:
Any theories on why the following is blocked?
Chat.
- Thread starter Zhypo
- Start date
- Status
Hopefully they'll fix it within whatever update, maybe the friends update? Seems like a good time. Only 335 more weeks guys!
The following words should be unblocked, as they can be used legitimatly (and it'll ruin the experience of legit players if they remain blocked):
Also, I believe that the cap filter should only work on lines that are more than n characters long, as it will make it hard to type acronyms.
EG:
(any more?)
Also, I believe that the cap filter should only work on lines that are more than n characters long, as it will make it hard to type acronyms.
EG:
Okay, I don't develop for CubeCraft, but I think I can tell you how it works and why certain phrases are being blocked.
Addressing Some of the Blocked Phrases Mentioned Above
Explanation
Players often try to find all kinds of ways to bypass the blocked words filter. For example, people will try to bypass a word by inserting spaces. Thus, a string is treated as though no spaces are between the words. That is why "als het" is treated as "alshet" which contains the substring "shet" which is blocked. If spaces were counted, users could simply insert spaces in cuss words.
Potential Resolution
Figuring out how best to filter words is tough. Here is my idea:
Tokenize the chat string, splitting by spaces and other non-alphanumeric symbols.
Build a spaceless (and alphanumeric-only) n-gram collection from n=1 to [token array length].
If an n-gram contains a blocked word, the chat is blocked.
EXAMPLE:
Since I want to avoid putting cuss words in the forum, let's pretend for example that the word "hit" was blocked from the chat.
Under CubeCraft's current system, then, the phrase "hi there" would be blocked for containing the substring "hi t"
However, using my idea, the following process would occur to "hi there":
Tokens = {"hi", "there"}
n-grams: "hi", "there", "hithere"
Are any of those n-grams the word "hit" by itself? NO. So the phrase is not blocked.
Now suppose someone tried to bypass the block of the word "hit" by putting it in chat like this: "h*i*t" (or with spaces or some other symbols). Now here is the process that would occur:
Tokens = {"h", "i", "t"}
n-grams: "h", "i', "t", "hi", "it", "hit"
Are any of those n-grams the word "hit" by itself? YES. So the phrase IS blocked.
The idea here is to avoid linking together a word ending and a word beginning substring to form a false positive for bad language detection.
To save time, you need not iterate through every n-gram to the end if a blocked n-gram is found. When a blocked n-gram is found, simply stop the process and block the chat.
Thus, as another example, "I'm going to hi t you" would do the following:
Tokens = {"I", "m", "going", "to", "hi", "t", "you"}
n-grams: "I", "m", ..., "tohi", "hit" BLOCKED WORD FOUND, stop finding n-grams and block the chat.
So instead of finding ALL n-grams, then checking for a blocked word, check for a blocked word after each n-gram until a blocked word is found (or until the end of the token stream is reached, in which case if no blocked word is found, the chat is good to go).
Final Remark
Does that resolution make sense to anyone? Does it seem like a feasible idea?
I don't know exactly what goes on behind the scenes of CubeCraft's source code, but that is just my thought on how I would probably try handling it.
Addressing Some of the Blocked Phrases Mentioned Above
"als het" contains the substring "s het" which appears to be a form of the word "sh*t"
"dankjewel" contains the substring "jew" which is sometimes used as an insult, thus it is blocked.
Explanation
Players often try to find all kinds of ways to bypass the blocked words filter. For example, people will try to bypass a word by inserting spaces. Thus, a string is treated as though no spaces are between the words. That is why "als het" is treated as "alshet" which contains the substring "shet" which is blocked. If spaces were counted, users could simply insert spaces in cuss words.
Potential Resolution
Figuring out how best to filter words is tough. Here is my idea:
Tokenize the chat string, splitting by spaces and other non-alphanumeric symbols.
Build a spaceless (and alphanumeric-only) n-gram collection from n=1 to [token array length].
If an n-gram contains a blocked word, the chat is blocked.
EXAMPLE:
Since I want to avoid putting cuss words in the forum, let's pretend for example that the word "hit" was blocked from the chat.
Under CubeCraft's current system, then, the phrase "hi there" would be blocked for containing the substring "hi t"
However, using my idea, the following process would occur to "hi there":
Tokens = {"hi", "there"}
n-grams: "hi", "there", "hithere"
Are any of those n-grams the word "hit" by itself? NO. So the phrase is not blocked.
Now suppose someone tried to bypass the block of the word "hit" by putting it in chat like this: "h*i*t" (or with spaces or some other symbols). Now here is the process that would occur:
Tokens = {"h", "i", "t"}
n-grams: "h", "i', "t", "hi", "it", "hit"
Are any of those n-grams the word "hit" by itself? YES. So the phrase IS blocked.
The idea here is to avoid linking together a word ending and a word beginning substring to form a false positive for bad language detection.
To save time, you need not iterate through every n-gram to the end if a blocked n-gram is found. When a blocked n-gram is found, simply stop the process and block the chat.
Thus, as another example, "I'm going to hi t you" would do the following:
Tokens = {"I", "m", "going", "to", "hi", "t", "you"}
n-grams: "I", "m", ..., "tohi", "hit" BLOCKED WORD FOUND, stop finding n-grams and block the chat.
So instead of finding ALL n-grams, then checking for a blocked word, check for a blocked word after each n-gram until a blocked word is found (or until the end of the token stream is reached, in which case if no blocked word is found, the chat is good to go).
Final Remark
Does that resolution make sense to anyone? Does it seem like a feasible idea?
I don't know exactly what goes on behind the scenes of CubeCraft's source code, but that is just my thought on how I would probably try handling it.
Last edited:
Okay, if you're going to block "jew", you might as well block:
>:(
What this mean :confused1:?
Well, "jew" is more often used in an offensive manner than those other terms. There's nothing wrong with "jew" in and of itself, it's just how it's commonly used by people that's an issue (and people don't typically use the words in your list in a way meant to offend). Also, "jew" doesn't automatically mean a follower of Judaism : P (going off of your list...).
I have nothing to do with the rules, just trying to explain why I think those are the rules...
I think the filter should not fully prevent people from bypassing it; all to often it makes it impossible to say legit messages.
ps:
Any theories on why the following is blocked?
@BalloonBoy2
Not sure how long you've been on cubecraft but when CCG released the Op skywars update, everyone was spamming the chests saying exactly things like vote for op or op pls etc
Not sure how long you've been on cubecraft but when CCG released the Op skywars update, everyone was spamming the chests saying exactly things like vote for op or op pls etc
- Status
Similar threads
Team online
-
MuhhMuhderator
Latest profile posts
This is YOUR daily dose of facts #41-
The moon has moonquakes, which are kind of like the earthquakes on the Earth, but weaker.
The moon has moonquakes, which are kind of like the earthquakes on the Earth, but weaker.
Reesle wrote on ii6xxq's profile.
Welcome to the Official Cubecraft Forums!! Hope you enjoy your stay :)
