We're about to make community management a whole lot easier.

Get Notified

Find Out More

"Don't Read the Comments"

Interesting content and engaging discussions are the lifeblood of community sites. However, they can be drowned out by both anti-social comments and mediocre contributions.

Common approaches to comment moderation include:

Manual Moderation

Curation by dedicated staff members is an option, but it doesn’t scale and it’s expensive.


Up- and down-voting can work well, but is also subject to sabotage.

Word blocking

Key words can trigger blocking. However, words can mean different things depending on the audience.

A platform for community managers

Promote productive conversation
while discouraging anti-social behavior.

  • Automatically identify and promote high-quality contributions to discussion
  • Where possible, convert anti-social commenters into productive participants
  • Block inflammatory, anti-social comments

Article and comments are both automatically analyzed

Comments are scored on:

  • Relevance of comment to article
  • Quality of contribution to discussion (both supporting and opposing views)
  • Inflammatory potential

Comments are ranked

  • High-quality, non-abusive comments on top
  • Both supporting and opposing views are represented
  • Mediocre quality comments (e.g., “Great article”) are ranked lower.

  • Moderate to highly inflammatory comments are delayed, sometimes indefinitely.
  • During that delay, users can retract or rephrase their contribution.


Shawn Holland and Kate Dobroth have worked together previously to produce ChefTap, a popular cooking app for iOS and Android that uses machine learning and artificial intelligence to extract recipes from unstructured text. Users can clip recipes from any website or blog. ChefTap processes web pages from thousands of recipe sites, extracting millions of recipes a year.

Classifying the text of blog comments is not very different from classifying the text of recipes. It's all the same math.

Shawn Holland has been developing telecom, mobile and natural language processing software for over a decade and a half. He is the technical lead on ChefTap, and developed the machine learning technology behind ChefTap's Recipe Recognizer.

Previously, Shawn worked for PatientKeeper where he was the engineering lead for the security team responsible for keeping patient medical records secure on physician's personal mobile devices.

Earlier, Shawn worked at Lernout & Hauspie where he led a team of developers involved in producing an advanced mixed initiative VoiceXML speech user interface server system that leveraged natural language processing to enable voice IVR systems to conduct more natural conversations with people. This work was part of the foundation for the technology behind Apple's Siri.

Kate Dobroth has been a user experience designer and user researcher for over two decades. She has designed UXs for a variety of products, specializing most recently in mobile and web applications. She has worked as an independent consultant for much of the past 15 years, and has also worked at Verizon/GTE, BBN Speech Products, and American Institutes for Research. She has a PhD in cognitive psychology from Northeastern University.