Newsblur as an Intelligent Agent Used for CI

Newsblur (www.newsblur.com) is a Web-based RSS feed reading service. This is a review of how I found the service useful while working on some competitive intelligence (CI). It’s convenient and conducive for tracking issues, trends, commentary, and news.

Newsblur has a set of features that I find make it worth paying for an annual subscription (it also offers a no-cost option). I’ve long used RSS readers, initially preferring desktop readers like Akregator or RSSOwl. The quantity of feeds that I follow is so large now that it can take desktop readers a long time to update, whereas Web services can use their resources more efficiently to deliver the feed content. I currently use Newsblur to follow about 1300 feeds but I’ll just cover a few examples in the context of how it might be used in CI work.

Newsblur sets itself apart from many other feed readers with its “Intelligence Trainer” functionality. The Intelligence Trainer enables the system to proactively filter new entries by marking key words, tags, authors, and sites that should be raised as a priority for reading because they match user-specified characteristics. The Intelligence Trainer also allows a user to decrease the priority of entries based on the same criteria.

A user can go through a step-by-step process of training the service for every feed—but that can be time consuming so users can also train it on an ad hoc basis, whenever there is a good opportunity. The Intelligence Trainer proactively filters results that would otherwise require manual searching.

Examples with the Intelligence Trainer

In order to research some issues related to the subject of open innovation (OI), I created a folder in Newsblur to store feeds from any site with potential to publish information on the subject. I also created subfolders for feeds from organizations that might offer information even though their domain was not my primary interest. This allowed me to distinguish and pay better attention to the context of my sources.

Intelligence Trainer example
Newsblur Intelligence Trainer example

I trained Newsblur to identify key words or phrases, examples I’m showing for this post include “open innovation”, “r&d”, “management practices”, “patent”, etc. These varied depending on the particular feed. Some feeds, for example PubMed, can be pre-filtered for the open innovation topic so the Intelligence Trainer could be focused more on specific issues within that subject. This technique can produce useful results.

Although it’s possible, I wouldn’t recommend manually re-entering those terms to search the feeds for relevant information.

Newsblur "focus" screen
Newsblur “focus” screen shows entries the Intelligence Trainer filtered.

Instead I’d click Newsblur’s “focus” button (after setting up the training) and scan the results that it provides me—these are essentially automated search results based on the intelligence training. I would then use the search function if the recall was too high and I wanted to narrow the list of results. For example, I trained my feed for PubMed OI on following terms:

  • partners
  • collaborative
  • shared governance
  • open source
  • patent
  • management practices

This resulted in about 10 entries from a list of over 100. From within that pool, it’s possible to search on specific terms but that can produce overly narrow results. A sampling from scanning the results on their own (without the deeper search term precision) produced a good number of entry titles, which suggested useful intelligence. When I reviewed results this way, they strongly matched my need.

How it Works and Other Notable Functionality

title text intelligence training
Newsblur Intelligence Trainer example 2, with title text.

The Intelligence Trainer (as far as I understand) picks up words based on the entry title, body text, the feed publisher’s tags associated with entries, authors’ names, and the feed name itself. It allows the user to specify different terms based on the fields you want to use. I discovered a “cheat” to train the system for words that haven’t yet been included in any entry’s text, by typing the word into the editable title field. That’s useful to prepare for potential future interests.

Newsblur has other functionality that can be useful for CI work.

Newsblur example showing how it displays edits to an entry
Newsblur example showing how it displays edits to an entry trained on the term “pharmaceutical”

If a user wants to see changes to entries over time, it is possible to turn on a change tracking feature. For example, a feed of press releases might seem static: the release is issued, the reader picks it up, and the user reads it. However it’s possible that a PR person made an error and later corrected the release. In such a case, Newsblur can be set to show a red strike-through of the original text, and have the new text in green beside it.

Newsblur also provides a “save” feature and a Blurblog. Both of these features enable different ways of capturing feed entries for later review or collaborating with other people. In both cases you can add notes to the entry and review other users’ notes.

Previously, Newsblur’s search was somewhat confined but now it supports searching across all of a user’s feeds. It would be nice if it had advanced search functionality that made it possible to designate which fields to search (title, body, etc.).

Assessment

Although the intelligence trainer is a very nice feature, it requires some amount of effort to use it well. Newsblur would be more useful if it offered a management tool that pooled all the terms and tags you’ve identified, to prioritize or deprioritize. It currently requires that you manage them feed-by-feed or story-by-story. It should also enable users to organize terms by folder groups, to apply to all the feeds within, or across folders.

Generally users need to paste the URL of an RSS feed into Newsblur. The system makes a valiant try at discovering feeds on a Web site if the user doesn’t know the exact URL. If the site has no available feeds however, Newsblur cannot track its content and this of course is one of the limits in using RSS for competitive intelligence.

A risk with any third-party, Web-based service is the lack of control: it can be shutdown (as Google demonstrated with its Reader). While Newsblur has proven reliable and convenient with regular refinements to features for over a year (in my experience), it is a small organization so there is little guarantee of its longevity (of course, Google is big and that didn’t guarantee Reader’s longevity). It is easy however, to export your list of organized feeds into an OPML file to transfer to another reader if the site were ever to shut down.

Overall, I find Newsblur a very efficient way to track a large quantity of updates and news related to a CI subject but I would not recommend relying on any RSS reader as the sole means for finding and tracking information. Applicable intelligence results depend on the utility of the feeds that you’re able to find, as well as your resourcefulness and precision in refining the Intelligence Trainer.