A Peek at HiveFire, unlaunched news aggregator

Here’s a peek at HiveFire, for a virtual publication called “Free Tibet”.  They’re a very low key startup.  It looks like they are scraping articles from around the web, determining whether they belong in the virtual publication, or whatever they call it, extracting entities and quotes, and attributing the quotes in some cases.  They have some simple mechanisms for associating articles, made easier by the narrowness of the vertical.

Dalai Lama page:  http://truetibet.hivefire.com/entity/profile/dalai-lama—93149/

Quotes hub: http://truetibet.hivefire.com/entity/quotes/

And so on.  You can explore the site from there.  It looks like they have a number of virtual publications foating around.  Its a nice idea, good way to make sure related content you bring in stays relevant, and indexes stay small, although clearly the UI has a ways to go.  The analytics look good, entity extraction, quote extraction and attribution.  Quote attribution you have to approach very carfully, and aim for high precision.  You never want to misquote someone, even if you can blame it on an algorithm.  Its not too hard to get basic functionality there for entities and quotes, plenty of open source packages for the former, if you’re happy with low recall.  Its also a risky move to leave the site open at such an early phase, although it may be unintentional.

The virtual publications are nice, would help with performance, and should be easy to set up.  You can collect articles using a boolean search terms (although this technically is filtering, not searching), and compare them as they come into the system.  If they had broader topics like “politics” they would probably want to consider a different method, but given the narrowness of the verticals, I’d go with boolean search terms.  Mostly they need work on their UI, its not the prettiest thing.  As with some other’s I’ve blogged about, its the same old field emission problem of sucking users away from other sites, or capturing free ones from Google searches.  I’m not sure if the level of innovation is sufficient yet to really distinguish them.  They either need a great UI, or a better UI with some unique analytics or way of organizing things.  Even then, capturing online users is a crap shoot, and its a crowded field.  

They guy behind it, Pawan Deshpande, has an interesting pedigree.  His father Gururaj Deshpande founded Cascade, which was bought by Ascend, which was then bought by Lucent back when I was there.  It was a huge deal, and McGinn was constantly talking about killing Cisco.  Its amazing how fast things can fall apart.  Soon thereafter, Deshpande IPO’d his second company, Sycamore.  One of my cohort at Bell, Jinendra Ranka, went to them in 2000.  Then the crash hit.  Sycamore is still around although the stock is a tad bit off of its high.  Lucent merged with Alcatel, and now it’s getting pushed into the back seat.  

Back to HiveFire.  It’s interesting, they have a good start on the analytics, I’ll be curious to see how they evolve.