The Inversoft team is excited to announce the formal launch of CleanSpeak KidSafety in advance of the Digital Kids Conference that we will be attending in California this week.

Send a note to adam [at] inversoft [dot] com if you will also be there and would like to connect in person.

• Press Release:  KidSafety-PressRelease-04-24-2012.pdf
• One-Page Info Sheet: CleanSpeak-KidSafety.pdf

- – -

April 24, 2012 | FOR IMMEDIATE RELEASE

Inversoft Strengthens CleanSpeak KidSafety Offering with Expert Moderation Services

Inversoft, the leading provider of profanity filtering software and moderation tools to the Kids, Tweens & Teens market, is pleased to announce the addition of specialized moderation and consulting services specifically designed to protect children in online games, virtual worlds and social networks.

Denver, CO (PRWEB) April 24, 2012—Inversoft’s Online Child Safety Practice announced today the immediate availability of expert moderation and consulting services to complement its market-leading CleanSpeak software tools. The combined offering named CleanSpeak KidSafety allows clients to safeguard their kid, tween and teen audiences with the industry’s only fully-integrated online safety solution designed specifically to protect children.

“The experience gained from working with many of the world’s most prominent kids-oriented online games, virtual worlds and social networks over the past four years has provided a strong foundation on which to build the CleanSpeak KidSafety offering,” said Brian Pontarelli, Inversoft CEO and parent of two. “Our unique combination of industry knowledge, powerful technology, and the highest quality child-specific moderation services allows our clients to focus on growing and engaging their audience, knowing that we will keep them safe.”

Inversoft’s KidSafety moderation team is entirely US-based, specifically trained to moderate younger audiences and has experience working on some of the industry’s highest volume and most successful kid-focused online properties. The KidSafety consulting services leverage four years of real-world client experience to provide advice and guidance around COPPA and regulatory compliance; moderation workflow and process optimization; and filtering technology integration, configuration and tuning. All services are individually tailored to each client’s specific needs and moderator-staffing levels can be quickly scaled up or down to meet changing audience requirements.

“We believe that our focus on providing high quality services exclusively to the children’s market makes the KidSafety offering compelling on its own,” continued Pontarelli, “but the true advantage here is realized through the use of our CleanSpeak software. The improvements in moderator efficiency and effectiveness with CleanSpeak are often as substantial as 400%-500%, which allows us to offer the entire suite of services at monthly rates that are very competitive with those of non-specialized providers.”

For more information about CleanSpeak KidSafety, and to schedule an online demonstration of the full end-to-end solution, please visit http://www.inversoft.com/kidsafety.

ABOUT INVERSOFT:
Inversoft, makers of CleanSpeak profanity filtering and moderation software, is a Denver, Colorado technology and professional services company that helps clients manage all sources of user-generated content. Widely deployed among gaming, financial services, kids & teens, entertainment, online publishing, and agency clients since 2008, CleanSpeak remains the industry standard for enterprise-scale content filtering and continues to set the market benchmarks for accuracy, performance, security and customization.

PRESS CONTACT:
Brian Pontarelli, CEO
Inversoft Inc.
720-352-1193

###

The sixth in a series of posts about the finer points of profanity filtering…

Embedded words occur when a dictionary word or proper name contain profanity:

1. Don’t assume profanity filters are inaccurate
2. Harry Lipshitz has a hard time creating accounts on web sites
3. This has been documented as the Scunthorpe problem

This case is actually quite simple to handle as a sophisticated profanity filter can look for dictionary words that contain profanity and safely ignore them, either  preemptively or during the filtering process. Poorly written filters will often get caught up on these simple cases and flag large number of dictionary words as profanity. CleanSpeak pulls from a large set of dictionary words and proper names in real time, over 140,000 in all, to correctly handle this situation and avoid a potentially large number of false positives without hindering performance.

Learn more about the Scunthorpe problem online at:

• http://en.wikipedia.org/wiki/Scunthorpe_problem
• http://www.dailymail.co.uk/news/article-2076336/Virgin-Media-censors-offensive-D–ens-A–nal-Hitchc–zealous-blunder.html

The second in a series of posts about the basics of user-generated content moderation…

There are a couple of basic challenges encountered when moderating user-generated content with a moderation team that consists of more than a single individual. First, you want to prevent different moderators from reviewing the same content. Secondly, when the team is not all at the same location, you want to make sure that the time-zone is accounted for when reviewing the logs of moderated content.

Your community moderation software should not allow a moderator to review content that has already been checked-out by another moderator. In technical terms, this is akin to a database lock of a record. But what if a moderator checks out content that is pending approval then falls asleep at the keyboard? There must also be a check-out duration so that content is returned back to the queue for another moderator in the case of the first moderator taking too long

CleanSpeak solves this issue with a content check-out system and configurable settings for check out duration. When the check-out duration expires, the following message pops up on the moderator’s screen:

Finally, CleanSpeak automatically accounts for different time zones by setting a system time-zone and allowing each moderator to set their time zone independently. This way, when reviewing logs of moderation actions performed, the times are always localized to the moderator that is currently logged in.

The first in a series of posts about the basics of user-generated content moderation…

When JoeBob submits a message in a public forum along the lines of “go to hell!” what action should you, as a moderator, take on his account? There is no single answer, of course, unless you know the context of the message. There are a number of considerations, including: JoeBob’s history, the conversation before and after, and the theme of the application.

Knowing the theme of the game or forum is the easy part. If it’s an online game where there’s an area that players refer to as “hell”, then go to hell! may actually be a constructive comment. Being able to quickly view the historical content generated by JoeBob over the course of the day, the last week, or the last couple of years is also essential.

JoeBob may simply be having a bad day, but has been overall very supportive and productive within the community, in which case a warning would suffice. However, for a repeat offender, a short-term ban may be appropriate. Finally, viewing the conversation in context is also extremely helpful as another user may have been goading him into reacting in such a way.

Your community members appreciate moderators keeping an eye on things and steering the conversations toward constructive paths. On the other hand, the community will lash out quickly and vehemently should moderators or automated software take an inappropriate action on a user.

The optimal community management platform or moderation tools will allow you to understand context with minimal steps and in real time. If a conversation starts heating up, you want to know about it now and review the context quickly in order to take the correct action 100% of the time.

]]> http://www.inversoft.com/understanding-context/feed/ 0 http://www.inversoft.com/separators/ http://www.inversoft.com/separators/#comments Wed, 15 Feb 2012 12:49:43 +0000 marshall@inversoft.com http://www.inversoft.com/?p=1648 Profanity Filtering 101: Separators

The fifth in a series of posts about the finer points of profanity filtering…

One of the more sophisticated attacks that users employ against profanity filters involves inserting separators, such as spaces or periods, between the other characters of a word so that the word can still easily be read.

The following examples illustrate how the simple process of inserting additional non-alphabetic characters between the characters of the word does not interfere with the reader’s ability to identify the word correctly:

1. s…….m…..u…..r……f
2. s m u r f
3. s….m u r….f
4. I’m going to smash it (false positive!)

It might be difficult to see the profanity in #4, but if you look at the last 4 characters on their own, you’ll see it.

Filters that do not intelligently handle separators will incorrectly identify this sentence as inflammatory and generate a false positive. Therefore, the filter must understand how word separators behave within sentences and how they can be used as an attack.

Separators can also be extremely complex. In some cases, other alphabetic characters can be used as separates without impacting the ability of the reader to identify the word:

• SaaaMaaaUaaaaRaaaaF
• She Might be Under a RainFall

CleanSpeak is capable of handling separators that use common punctuation marks such as periods, semi-colons and asterisks. It is also capable of handling spaces without introducing false-positives. It understands that word boundaries are used correctly more often than not, but that spaces can still be used to attack the filter.

The fourth in a series of posts about the finer points of profanity filtering…

“Repeated characters” is another commonly used filter attack that involves the simple repetition of characters in a word. This straightforward tactic still fools many profanity filters, most of which are not designed to ignore multiple instances of the same character:

• heeeeeeeeeeellllllllllllooooooooooooo

CleanSpeak is capable of detecting this type of filter attack and will correctly and automatically identify words regardless of repetition.

The third in a series of posts about the finer points of profanity filtering…

Character swapping and collapsing is the process of replacing characters with other alphabetic characters (or removing unnecessary characters) while still retaining the phonetic structure of the word. This tactic is often used to attack filters that do not understand phonetics:

1. Teech me guitar
2. Attak the main castle gate

Example #1 is a simple character swap of an “a” to an “e” that still retains the same phonetic structure of the word and allows the reader to infer the original word.

Example #2, on the other hand, is an example of character collapsing. In this example the “ck” in the word “Attack” has been collapsed to a single “k” character.

In some cases characters can’t be collapsed without changing the meaning of the word. For example, the word “been” can’t be collapsed to “ben”. Therefore, a filter can’t simply ignore multiple characters that are phonetically the same. It has to understand if the word can be collapsed.

This type of filter attack is a common tactic of users due to the fact that most filters have no knowledge of phonetics. CleanSpeak is able to correctly identify the majority of these types of phonetic attacks. It understands that many characters can be swapped, and in some words, multiple characters can also be collapsed. It also knows which words can’t be collapsed without impacting the meaning of the word.

The second in a series of posts about the finer points of profanity filtering…

Character replacement is the process of replacing certain characters with others, usually symbols, that look the same or similar. This is a popular method, often referred to as “Leet” or “L33t” speak, used to attack traditional content and profanity filters that ignore or don’t play well with non-alphabetic characters. Some examples…

1. $ally is my neighbor
2. |)on’t be a menace
3. |<nive$ can be dangerous
4. \/\/hat are you doing?

Examples #2, #3 and #4 illustrate a user using multiple characters to replace a single character. #2 is using the “|” (pipe) character and the “)” (right-parenthesis) character to create a capital “D” character. #4 is using a combination of forward and backward slashes to create a capital “W” character. Example #3 goes a step further. It is replacing two different characters in the text. Both the “K” and “S” characters are being replaced.

{IMAGE SOURCE: http://en.wikipedia.org/wiki/Leet}

Most of the filters in use today don’t understand character replacements very well and, if they do make an attempt, usually just swap back in the original characters prior to filtering. This can produce the dreaded “false-positive” if the user was in fact using a character (such as the dollar sign) correctly, so a filter must be careful not to assume that all non-alphabetic characters are filter attacks.

See some examples of how the CleanSpeak filtering engine automatically handles replacement characters here: http://www.inversoft.com/cleanspeak-capabilities/filtering/#smurf

And learn more about leet speak online at:

• Wikipedia: http://en.wikipedia.org/wiki/Leet
• Google Leet Search: http://www.google.com/webhp?hl=xx-hacker
• Leet Speak Translator: http://www.brenz.net/services/l337Maker.asp

]]> http://www.inversoft.com/character-replacement-leet-speak/feed/ 0 http://www.inversoft.com/the-grawlix/ http://www.inversoft.com/the-grawlix/#comments Mon, 30 Jan 2012 18:10:31 +0000 Inversoft http://173.214.160.119/?p=1399 Profanity Filtering 101: The Grawlix

The first in a series of posts about the finer points of profanity filtering…

You’ve seen it all over the @#$%&! place, but probably didn’t know that this ubiquitous string of characters has a name that was coined almost 50 years ago by cartoonist Mort Walker, the creator of “Beetle Bailey” and “Hi and Lois.”

“Grawlixes” is one of a series of great words (Agitrons, Blurgits, Plewds, Farkles, Digitrons,  and many more…) that Walker invented to describe the devices cartoonists employ to convey specific types of information in their comic strips. Loudly used to express obscenities right in the middle of the family oriented funny pages, the Grawlix reigns as the grandaddy of all profanity filtering options and practically defines the category of “replacement characters,” a topic we’ll explore in more detail in an upcoming post.

Learn more about Grawlixes online at:

• http://en.wikipedia.org/wiki/The_Lexicon_of_Comicana
• http://www.amazon.com/Lexicon-Comicana-Mort-Walker/dp/059508902X

Reducing Moderation Costs With An Intelligent Filter

Viewpoints has used CleanSpeak for several years as an add-on to their own proprietary moderation tools. The Viewpoints team moderates hundreds of thousands of consumer contributions across multiple communities and vastly different demographics annually. The CleanSpeak tool has proven to not only meet those needs but it has also allowed Viewpoints to post more consumer contributions than ever before.

“Adding a profanity filter such as Clean Speak has given our moderation team a safety net when moderating. They are now able to quickly assess poor content on the site and moderate it off instantly. The moderation team has increased their overall efficiency and also increased the throughput of content. We highly recommend a solution such as CleanSpeak for any site that works with a large amount of consumer generated content .”

Download the PDF →