HTML Preview Technical White Paper page number 1.
Mollom technical whitepaper
1 / 4
Technical whitepaper
The web is changing. User contribution is now what makes or breaks a site. Allowing
users to react, participate and contribute while still keeping your site under control can be a
huge challenge. Mollom is a web service (Software as a Service) that helps you identify
content quality and, more importantly, helps you stop comment and contact form spam,
and cap protect registration forms.
Mollom analyzes the quality of content posted to websites and tries to determine
whether this content is unwanted or not. Websites that allow visitors to contribute or post
comments are constantly being flooded with inappropriate, distracting or even illegal
commercial messages, many of which are uploaded by automatic "spambots." Mollom
screens all contributions before they are posted to participating websites. We use Machine
Learning techniques, Language Analysis and a reputation system to ease moderation and
improve the overall quality of your site’s content.
Mollom also provides a centralized CAPTCHA service that allows to protect e.g. user
registration forms using both image and audio CAPTCHAs. Mollom constantly monitors
and tweaks its CAPTCHAs so they are still easily solvable by humans, but cannot be solved
by automated scripts.
This document elaborates on the technical aspects of the Mollom services, but does not
go into technical detail on how to implement the service’s open API. For this we refer to
the developer API documentation which can be found on our website.
1. Mollom products and services
1.1. Text filtering and content analysis service
The Mollom filtering service is a hosted web service that analyzes the quality of content
posted to websites. This includes comments, contact form messages, blogs, forum posts, etc.
Mollom specifically tries to determine whether this content is unwanted - i.e. "spam" - or
desirable - i.e. "ham." Websites that allow visitors to contribute or post comments are often
being flooded with inappropriate, distracting or
even illegal commercial messages, many of which
are uploaded by automatic "spambots". Mollom’s
text filtering and content analysis service screens
all contributions before they are posted to
participating websites.
Websites using Mollom send data they want
checked to mollom.com, and Mollom replies with
either a spam or ham classification. If Mollom is
not certain, it will return "unsure", typically
prompting websites to ask Mollom's CAPTCHA
service for an audio or visual CAPTCHA
challenge to present to the user.
The fact that Mollom can reply “unsure” makes Mollom unique compared to other
services. Thanks to the "unsure" reply and the CAPTCHA challenges, Mollom avoids
incorrectly classifying legitimate contributions as spam. The strategy of combining text
classification with occasional CAPTCHAs has two important benefits:
Mollom statistics
As of October 2008, Mollom
protects over 6,000 websites
and is used by companies such
as Sony BMG (more than 80
of their web sites), Acquia,
Adobe, IDG, Fast Company,
Now Public, LinuxJournal,
Jupitermedia, The New York
Observer, and many more.
Mollom has blocked over
10,000,000 spam comments
since its start, has an average
spam-stopping accuracy of
approximately 99,83%.