Persai is a site that automatically discovers articles you might like from around the web. Imagine it being the opposite of social news sites like Digg or Reddit, where users submit and vote on their favorite stories. Instead, it’s a personalized news aggregator, sort of like automated news aggregator Techmeme, but just for your interests.
With Persai, you sign up (it’s in private beta) and create a set of keywords that you’d like to discover articles about. Let’s say I want to track all the breaking news about Facebook and Myspace. I type in, “facebook, myspace, bebo” into Persai’s interest creation form and give the interest a name, such as “social networks.”
Then Persai uses machine learning, where it trains software algorithms to find articles on the web related to the keywords in your interest. It identifies and extracts the unique content from each web page it finds, matching each page up against the keywords to find what’s most relevant. (You can read more about machine learning here, if you’re interested.)
If you click on an article to read, Persai learns that you like that type of article. If you click on the “reject” button below each entry, that entry disappears and Persai learns that you don’t like that type of article. It doesn’t rely on user votes, tag clouds, analysis of metadata about an article, or anything else.
You can also get RSS feeds of each “interest” you create, as well as an OPML file containing every feed for every interest. I’ll be adding Persai feeds to my feed reader, to see how it stacks up against the other tools I use to find news on the web. I’ll be using it in place of Google Alerts, which emails or RSS feed notifications that you can sign up to have Google send you, linking to newly-published articles that mention keywords like “Facebook” or “Myspace”.
Emeryville, Calif.-based Persai also stresses that it’s not just a tool to help tech bloggers find hot new tech stories. It says it is developing tools for businesses and publishers, but won’t go into more detail.
The three-person company has been living off $35,000 in funding over the past eight months. It has also gained a certain amount of notoriety, as the publishers of Web 2.0 hate blog Uncov — which they’ve recently stopped doing in order to focus completely on Persai.
Meanwhile, at least one executive of a successful web company tells me he would like to hire the team for their know-how, “once Persai fails.”
If you want to try out Persai out, email firstname.lastname@example.org and include the word “VentureBeat” in your subject line.
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.