Mine Safe

Mine Safe

Mine Safe is a cutting edge swear filter based on the latest technology. If you are looking for more than just a simple black list this is what you need. The team behind it has been doing content filtering since 1999 where we worked on Penguin Chat (1999). Through the years we have worked on billions of lines of chat affecting over 300 million users. Through working with multiple very large companies we have created the most powerful filter available.

It is our mandate to make the world a safer place for kids to grow up online. As such it is our joy to offer a free version of our PottyMouth lite services to keep you and your friends safe online.

How it works

Every chat request and sign created calls our webservice in the cloud. There we have advanced artificial intelligence and natural language processing examining the phrase to measure it's risk in context. There is very little logic in the plugin to keep the CPU load off your server as all it does is make a web request and then act on the results.

In addition whenever a user logs in their username is checked and determined if it is a bad name. If it's bad then you can either reject that user or replace their name with something auto-generated.

How to Install

Drop the jar file on your server, restart your server and execute the following command

/minesafe

This will connect to our cloud and generate a unique key for your server. It will also provide you a unique url you can go in and login and customize the settings in our dashboard. The dashboard is optional as the plugin will begin to work immediately.

If you call this command a second time it will just tell you the admin url again.

Joining our Project

It is our mandate to make the world a safer place for kids to grow up. If you log into your online dashboard at the url given on setup you can see your admin page. This will allow you to customize the plugin.

We also track words that our Artificial Intelligence detects might be misclassified in your install. If that happens it will show these words one by one with examples from your chat world (only yours). You can then vote on if it should be allowed or not. This allows us to crowdsource and get feedback on how our team should be building out the rules.

Features

  • The latest technology in chat filtering
  • Filter chat
  • Filter signs
  • Filter usernames (with option to ban or rename offending players)

Why Blacklisting is a Bad Idea

When we set out to make this people told us that there were plenty of filter plug-ins already. We give a special thanks to all the hard work people have put into them and invite them to talk to us about adding ours as an option into theirs. However there is a huge difference between a simple filter that handles a few hundred bad words and a cloud of computers using the latest technologies backed by years of experience.

When we started down this path in 1999 we thought that all we had to do was stop 4 letter words. We quickly learnt that in every site there is a small group of people who are far more creative than that and ruin the fun for everyone else. Imagine if a simple word like snarfel became a bad word (it's not but lots of people read this page and we need to be polite). Blacklisting needs to catch not only "snarfel" but "s.n.a.r.f.e.l", "SNAR fel", "sLnLaLrLfLeLl", "lefrans" (backwards), "snar (newline) fel". Some of that can get tricky and a lot of things can either be missed or lead to false positives.

What makes it worse is that it does not solve the problem. So we stopped snarfel but what about the thousands of other ways people will try to imply the same thing. As soon as you have blocked that they will resort to innuendos and subtle references. Now phrases like "sniffling falling thing" become a problem. As soon as you catch "sniffling falling" you have to deal with "wheezing falling". Of course these are all silly but the point is that it is a cold war race and takes a tremendous amount of time. For example in our 7th prototype for PottyMouth it took us less than a year to expand to over 1 million rules. I really don't think everyone has time to build over a million rules and keep it current just so they and their friends can play and not have a few people ruin it for everyone.

The new solution is to plug into an entire cloud of computers that are using machine learning, artificial intelligence, human moderation, human quality control. The more sites we get plugged into it the better the system becomes. We all grow together and our team does all the hard work.

Filtering Results

Choose one or more of the following

  • Put a warning on their screen teaching them about proper behavior.
  • Strike them with lighting
  • Lose in-game health, with losing more health more severe cases

We are open to adding more features as the community requests it.


Comments

Posts Quoted:
Reply
Clear All Quotes

About This Project

  • Project ID
    71643
  • Created
    Dec 31, 2013
  • Last Released File
    Never
  • Total Downloads
    0
  • License

Categories

Members

Recent Files

No files uploaded yet.