Next Version of FeedHub Continues to Attack Information Overload

Feedback from engaged users drives enhancements included in the next release

Pittsburgh, PA - February 4, 2008 - mSpoke has introduced a significant upgrade to FeedHub (www.feedhub.com) as the company continues to attack information overload.

This upgrade includes improvements in three key areas:

Streamlining Feedback Interactions - In the most recent release, FeedHub has added thumbs up / down buttons to let users directly provide feedback on the relevance of its recommendations.

Dave Mawhinney, mSpoke CEO and co-founder, points out that, while FeedHub's new release lets users provide more direct feedback, the system will also still learn implicitly from observing user behavior. "FeedHub's attention engine learns from a user's multiple attention streams - ratings feedback, del.icio.us tagging, Digg voting, link sharing, etc. - and transforms them into user intentions, or what we call memes. Memes capture a user's intentions about content preferences in a machine learning-friendly form."

Relevancy improvements - There were numerous improvements to the underlying engine. Most significant were the transition from the open directory project (DMOZ) category hierarchy to a Wikipedia-based taxonomy for classifying items, and a new meme that recommends posts with significantly more comments than are typical for other posts from that source. According to Paul Ogilvie, Principal Scientist, "Estimating the relevance of a feed item is a complex task which involves accurate representation of aspects of the item and the user's interests. Our new Wikipedia categories and the comment count meme will help us provide more relevant content for a very diverse user base."

Infrastructure Improvements - Several infrastructure improvements ensure that FeedHub will continue to scale effectively as the user base continues to grow. In addition to implementing several standard best practices, the JavaScript library was upgraded from Dojo 0.4.3 to 1.0.

mSpoke's CEO and Co-Founder Dave Mawhinney summarized the common theme behind these diverse enhancements:

About FeedHub

FeedHub (www.feedhub.com) is a next generation personalization solution that works with popular feed readers to save precious time by delivering only the relevant stories people want to read.

FeedHub cuts through the overwhelming volume of content published today by creating an individualized RSS feed that intelligently selects the most relevant posts from a set of feed sources.

mPower Adaptive Personalization Engine

FeedHub is powered by mSpoke's highly advanced mPower Adaptive Personalization Engine, the first adaptive learning platform to demonstrate that monolithic "one size fits all" algorithms simply don't work for the masses.

mPower is an intelligent attention engine platform that can power and enhance the experience and value of any site, channel, or network. Much more sophisticated than existing discovery/recommendation technology, this unique and powerful adaptive engine evolves with people by dynamically adapting to their preferences and selecting content that specifically matches their individual tastes.

Leading publishers, including Reed Elsevier's SupplyChainDaily (www.supplychaindaily.com)have leveraged the mPower Adaptive Personalization Engine to deliver individualized content to their subscribers.

About mSpoke

Based in Pittsburgh, PA, mSpoke has been providing next generation personalization solutions since 2006. Founded by CEO Dave Mawhinney, CTO Dean Thompson, and Co-Founder & VP, Business Development & Product Management Sean Ammirati, the mSpoke team has deep ties with Carnegie Mellon University, a recognized world leader in advanced computer science technologies.