John Tropea, fellow Corante Web Hub contributor, has posted some thoughts on more advanced mechanisms to get recommendations, based on what you read and write. RSS Reading: recommendations (excerpted)
- Here are some more feeds from people who also read this feed
- Here are some feeds based on stories people like, these people read many of the feeds you read
- Here are some stories based on stories people like, these people read many of the feeds you read
- Here are some stories with similar keywords in the title
- Here are some feeds based on conversations with the blogs in your OPML
- Here are some feeds from people who use the same tags as you
- Here are some feeds that blogs you subscribe to read
- Here are some feeds from people who also read this feed
- Here are some feeds from people who use the same tags as you
- Here are some feeds based on conversations you have in the blogosphere
- Here are some feeds based on conversations people are having with you in the blogosphere
Obviously, one could do this manually, but that takes a lot of time. What I'd really want is automation and integration of this recommendation engine into my aggregator. But the people who create these tool will have to be smart about it, since I don't always want recommendations: this is more of a research service than it is a constant need. And don't forget: My best recommendation engine is the people I am already reading. If one of my trusted sources (trusted based on my assessment) suggests a new information source, I will have a look. If the recommendation comes from someone I don't know (trust), I'm much less likely to follow the trail.
Thinking about this some more, I wonder if there is a better way to write about or visualize this process. My normal reading process gives me a slice of the world in which I am interested. I place some level of trust in my sources to bring me additional interesting news from an even wider slice of the world, and they do the same with the sources they read. Sometimes, however, I find such an interesting many-steps-away source that I decide to bring it into my regular sphere of reading. Similarly, I will drop sources when my interest in what they are saying fades. This is a continual process, as I find myself adding and removing feeds on a regular basis. I find search feeds to be a particularly good source for new possibilities, and a number of my recent articles have been a result of those search feeds.
While I am actively reading, I will sometimes wonder what others have written about a specific article. I am particularly interested in what my friends are saying, but I don't necessarily want to skip down to their location in my aggregator to find out. When threading works well, this is a great way to see that conversation as it has happened over recent history. For longer stretches of time, this doesn't tend to work as well, and I rely on external services like Talk Digger or Technorati. These searches are typically based on the URL of the source article, which may be many steps away from the article I'm actively reading.
Sometimes, rather than a specific article, I am curious about the topic described in the article. At this point, the only way I know to do more research is to hit the search engines and ask with varying levels of sophistication. I can do this search based on keywords or tags or use Waypath's text analysis-based search. If this topic ends up being something I want to track over time, I usually set up one (or more) search feeds to bring that topic into my regular reading habit. One of John's suggestions above highlight something that has frustrated me about the feed search tools: for these topic searches, I want the search to ignore the feeds I already read.
Now what about when I have a little more time to look for new and interesting feeds? In that case, I might explore the blogrolls of feeds I have found recently. (The feeds I read regularly have already told me about their blogrolls.) Or I might do some of the more interesting things John suggests above: Given all my feeds, what sites have they collectively referenced that I don't already know about? What new-to-me sites to they link in their blogrolls? And adding more complexity, John suggests diving into searches based on shared tags. At this stage in my reading habits, I tend to do this only sporadically, generally when I come upon a new topic that I want to follow. And I am not sure if anyone has created such a beast yet. I would certainly be curious about the results of a tool that could provide recommendations based on commonly-used tags, whether those are tags I commonly use or tags that are commonly-used within my reading list.