myFeedz Development Blog

After myFeedz was released on Adobe Labs, we have been seeing tremendous feedback on blogs, technology websites and in our e-mail inboxes. Whether it was positive or negative, we look at it as constructive criticism and we assure you that your comments and opinions get incorporated into the myFeedz roadmap. We track everything people say about myFeedz, whether it’s something we enjoy hearing or not.

The best way to see what people think of myFeedz is myFeedz itself. Or you could try Google’s Blog Search and Technorati. But just for the sake of it, we’ll list some of the reviews myFeedz has been getting on the blogosphere:

• Amit Agarwal from Digital Inspiration covers the main features of myFeedz and wonders how Adobe will monetize this.
• JD discusses how article clustering can be helpful and why myFeedz is different from other similar web applications out there.
• Marco Casario thinks myFeedz is a smart move from Adobe, and we should go into Flex. (We’ll keep that in mind!)
• Ryan Stewart is wondering whether Adobe is becoming Googlesque. (We’re flattered by the comparison, but we’re not exactly aiming in that direction).
• Stephen Shankland from Webware offer an in-depth tour of the features, but wonders whether the automatic tag extraction really builds a relevant profile of interests.
• Robert Scoble publishes an interesting feature wishlist for a RSS reader. (Thanks, Robert!)
• Jeremiah Owyang offers his take on the ultimate RSS reader on his Web Strategy blog.
• Kim Cavanaugh thinks we should improve the article relevance algorithm. (We’re working on it.)

Keep it coming, guys! We love feedback - it’s what pushes us forward and makes us get out of bed in the morning. Oh, and speaking of feedback, if you’re the lazy kind, you can simply vote in the myFeedz survey.

What are “connections” in myFeedz?

Well, basically, they are tags related to another tag. As you’re browsing myFeedz for new articles, you’ll notice a small, blue pod in the top right corner, that changes as you swap between tags. Say you’re reading news about “Adobe”. This is how the Connections pod would look like for the tag “Adobe”:

Essentially, these are hot topics about “Adobe” (or any other tag) that you could dive right into. (It’s interesting to see that “web” and “myfeedz” are now related to the name “Adobe” ;) ).

What are connections useful for?

Connections are just another way of browsing through myFeedz. You can use your tag cloud as a starting point, but why not wander off to one of the connected topics? Plus, you get to see which are the hot topics people talk and write about with regard to a given topic.

How are connections determined?

myFeedz looks at the freshest articles for a tag and extracts a set of the most frequently encountered words from those articles.

The Library Clips blog has two very interesting posts about clustering, popularity and personalization. The first post explains how “clustered stories” help bloggers:

“Just say you have 1000 unread items in your RSS Reader, you boil a cup of chai and get into rss zen mode.
You come across a really cool story you want to blog about, but before you publish a post you want to see who else may have posted on this same story, so you cruise around your RSS Reader clicking on your favourite feeds to see if they have also reported on the story.”

Indeed, clustering stories, which basically means grouping related articles and blog posts, can be useful for both the power users/bloggers, because it helps them stay updated with the latest blogosphere buzz, and the persons who just want the essential stories, and need to skip the details. Of course, clustering is powerful if it is combined with a mechanism for detecting and eliminating duplicates: articles which are just identical copies of the same story. MyFeedz manages to do both things, by detecting which stories link or refer to other stories, and which are simply copies of others, and then presenting this information in an organized way for the readers.

In the second post, we find out the advantages of having a personalization engine (that learns from each user’s reading habits - just as myFeedz does).

“What I’m saying here is that I still want clustered items, but I want stories ranked according to me, not just how popular everyone thinks they are. In fact, I want both, show me two streams:
Stream 1 - stories ranked based on my attention
Stream 2 - stories ranked by blogosphere popularity”

Of course, it often helps to see what is also popular for the community. MyFeedz offers the best of both worlds: top articles for your profile and top articles from the community of users.

Today, we released myFeedz as a technology preview on Adobe Labs. This release marks the rebranding of myFeedz under the Adobe identity. So here’s the new myFeedz logo:

We’ve been busy working on several bug fixes, improvements and features:

• Default user profile for Adobe products and technologies. People who sign up have the choice of automatically building a tag profile made up of Adobe product names and technologies.
  
• Guest users who are not logged in see a more user-friendly message. (Yeah, we admit the old JavaScript pop-up message was ugly and annoying.)
  
• We’re not republishing the entire contents of articles anymore, even if the aggregated feeds are distributing the full story. We’re trimming republished content to a 300-character summary, while maintaining the link to the original story, because we want to encourage users to read the full story on the publishing site.
• Article details are not indexed anymore by search engines. This was done to address the “page ranking” issue raised by several bloggers.
• myFeedz now has a forum and a presentation page on Adobe Labs. You can share your ideas and discuss features with other users (you need to have an Adobe ID).

We hope you enjoy your reading on myFeedz. Oh, and don’t forget about the survey!

We want to hear how you are currently using myFeedz and how we could improve it in the future to make your experience better. That is why we kindly ask you to fill in this short survey. It has just 17 questions, and should take you less than 5 minutes to fill in.