This is the first of a series of monthly guest posts which will include content from people working both inside and outside of NewsGator as well as in the widget/tech space in general.  This week, we're lucky to have mSpoke's Sean Ammirati write on our collaborative Related Content Widget.

Over the last several months, NewsGator has partnered with mSpoke on several initiatives. Our first jointly-developed product "the Related Content widget" is now coming to market.  While many companies can recommend related articles, we provide a superior experience by combining NewsGator's attention data and reporting with mSpoke's content analysis capabilities. In this post, I'll give a quick overview of how the Related Content Widget works and some of the highlights of our unique approach.

How Does it Work?

On page-load, the Related Content widget displays other related articles. As the publisher, you can choose to recommend related articles from either across the web (which allows you to blacklist some sources), a defined set of white-listed sources, or exclusively from your own content.  You also can monetize this directly by including an advertisement or other sponsored content as part of the widget.

As with most widgets, the installation is simple. Select the sources to be considered for recommendations, then add a little HTML snippet to the sidebar of your template. The widget will then automatically recommend appropriate related articles for each item.

Why Do We Believe Our Approach is Better?

We all know this isn't a new use case as many solutions exist for recommending related content for articles. Yet intuitively we also know most of these solutions could provide better recommendations.  Our Related Content Widget will deliver that experience using a superior approach built on four unique properties:

• Richer understanding of the concepts in an article
• Improved measures of an article's popularity
• Flexibility to recommend multiple types of content
• Detailed reports on visitor usage and interaction
• A number of flexible display options

Richer Understanding of the Concepts in an Article

Leveraging the mSpoke's platform lets us annotate both an article and the sources available for related recommendations with appropriate metadata. For our widget, mSpoke tags articles with three different types of content attributes: 

• Categories: a modified subset of the Wikipedia categories which reasonably reflect the kinds of material often found in RSS feed items;
• Topics: events or ideas that have been mentioned in multiple feed items in recent days or weeks. Each topic is algorithmically identified and named; and
• Named Entities: identify the people, organizations, and locations mentioned in each article.

For example, the article "Build A Custom Search Engine Using Your Social Bookmarks" written by Sarah Perez at ReadWriteWeb would be annotated with the following pieces of metadata:

Content Attributes

Categories :: social bookmarking & SEO

Topic :: "bookmarks"

Named entity :: Google

The Related Content widget would then consider other articles identified as being about Social Bookmarking, Search Engine Optimization, bookmarks and Google  for recommendation.

Improved Measures of an Article's Popularity

In addition to the meaning of an article, we need to understand the popularity of the content being considered for recommendation. At mSpoke, we have investigated many different measures of an item's popularity. We have found the aggregate attention scores across the NewsGator platform to be far more predictive than any other single measure of popularity for relevance. The Related Content widget uses the NewsGator attention scores to evaluate the popularity of any article being considered for recommendation.   

Side Note! We have been so impressed with the predictive power of the NewsGator's attention scores that we are incorporating them into our demo application, FeedHub. For those of you who aren't familiar with FeedHub, it allows anyone using an RSS reader (including NewsGator Online, NetNewsWire or FeedDemon) to filter the most relevant set of articles from a set of feeds chosen by the individual.  (For example, the set of feeds you just can't keep up with and ever few weeks mark as read.)  FeedHub provides each user with a personalized feed they can subscribe to in the RSS reader they choose. Each FeedHub user has their own set of memes (or content preferences) which automatically adjust based on what content attributes are most predictive for them.  In the next few weeks, we'll be adding a NewsGator meme. So, go ahead, create a FeedHub account now, and watch as the new NewsGator meme starts recommending popular content.  We think you'll find the NewsGator meme very predictive for you!

Flexibility to Recommend Multiple Types of Content

Beyond targeting related articles, many publishers want to provide hyper-relevant commercial content to their readers by using the Related Content widget. This can be advertisements we provide and share part of the revenue with the publisher, or a publisher may have their own ads, sponsored white-papers, or other content they''d like to promote. Either way, our system can attach appropriate metadata to this commercial content, increasing visitors' interaction with both the recommended articles and the sponsored material.

Detailed Reports on Visitor Interaction

If you have used any of the NewsGator widgets, you already know that the level of transparency provided in their reports on visitor interaction is second to none. We are very excited that publishers using the Related Content widget will get that same level of insight into reader behavior.

Conclusion

We hope you have enjoyed seeing how the Related Content widget can help you increase visitor interaction and loyalty by providing superior related content.  At this time, we have a few pilot publisher's in the process of deploying the widget.  We will certainly announce here and on the mSpoke blog as they launch. If you would like to incorporate this functionality on your site, please contact us or leave a comment below.