We are pushing harder on the development of k-collector. Murphy willing we should reach version 1.0 in a few more days.

There's a new feature we are testing which seems quite promising: auto topic matching.

Until now RSS feeds being aggregated needed to use ENT to transport topics to the server to be properly categorized. Now we can subscribe regular RSS feeds and the new topic matching code parses the posts to find topics already created by other users in the cloud.

Topics are still created by users of a cloud via their topic management tool in their weblogs while writing posts, all topics end up in the OPML topicRoll and the topicRoll is used to categorize posts coming from other sources.


I think that this is an interesting approach: only posts containing topics that have been used by active users of the cloud are aggregated.

The topic matching code is not only parsing words but also considering their context, especially links. While the word "Paolo" will give no result, the linked word "Paolo" will be identified with the "Paolo Valdemarin" topic thanks to the url.

It's not 100% accurate at the moment, but we think we'll be able to improve it significantly shortly.