Slashdot Mirror

it sounds like it's just a big ol' LSI System

A Perl implimentation of LSI can be found at Building a Vector Space Search Engine in Perl

However, there are at least three problems. First, it doesn't look LSI can answer questions like "Who is the Prime Minister of Canada?"

Second, the approach is patented by Telcordia Technologies.

Third, there are scalability problems with LSI. The author of the Perl article writes:

For all its advantages, LSI also presents some drawbacks. The poor scalability of the singular value decomposition (SVD) algorithm remains an obstacle to indexing very large collections. While techniques have been developed for making incremental updates to a scaled collection, these changes typically cannot exceed a certain threshold without triggering a rebuild [7,8]. These constraints make LSI ill suited to the kinds of large, rapidly changing document collections typically found on the Web.

A further disadvantage to LSI is the difficulty in interpreting the underlying reduced term space [4]. This makes it difficult to select an optimum number of singular values to retain in the SVD for a given collection, or allow domain exert adjustment of relevance values in the reduced space once the SVD has been calculated.

As a result, the author is now pursuing something called Contextual Network Graphs and has written a Perl module that was updated as recently as last August.

Bob Lucky of Telcordia/FCC has called it Grass-roots Networking. That's a more appealing name.

I know it's poor taste to post to my own thread ... but here's the article.

Not long ago Robert Lucky of Telcordia, who is a frequent columnist in IEEE Spectrum, wrote an article noting that technology predictions are worthless; that in his words, we're all just "bozos on the bus" waiting to see where technology will take us next. Even if someone "predicted" the impact of the internet ten or twenty years ago, because technology is so nonlinear, there's no reason why that same person's predictions today are any more valid than yours or mine.

Does anyone else remember this article? Any links?

A lot of these research projects are funded by the millitary, because ad-hoc networks are an obvious solution for situations where you don't have or cannot trust the infrastructured networks (be they wireless or wired).

Both wireless LAN and Bluetooth are capable of ad-hoc networks, but higher layers (the IP layer) must have some form of configuration to talk to each other. This is being developed in the IETF in the MANET and ZeroConf working groups.

Speed of these networks will improve over time, the other developments are at least as important and they will take some more years to mature I believe, so when it all comes together, heaven is upon us ;-)

Another interesting subject is ubiquitos or pervasive Internet. Meaning the accessibility of Internet in all (reasonably capable) devices and in all physical locations.
One complication is that Internet should not just be accessible to the rich, but also the poor and the people in the developing countries (this is important for a lot of reasons, but I digress...)

note these are hosts, not individual IP addresses

Umm, false. From here:

The Telcordia solution to quantifying Internet growth statistics is based on an internally developed unique sampling method. In this approach, over 150,000 randomly generated IP addresses are sampled on a daily basis and checked for their existence.

I haven't checked http://www.argreenhouse.com/netsizer ('cause it seems to be taking just forever to load), but they appear to be generating random quads and seeing if they exist, and running statistics against their findings over time.