An interesting tidbit that came out of Microsoft’s Worldwide Parrtner Conference last week. One of the company’s chief researchers, David Heckerman, is coming to South Africa to help test a new vaccine to treat HIV/AIDS. What makes it doubly interesting, however, is that as well as being a medical doctor, Heckerman was also one of the original designers of the Hotmail spam filters – and he reckons that his experiences with fighting the plague of spam have directly influenced the way he applies data analysis to healthcare.
The problem with spam, says Heckerman, is that in the early days companies like Microsoft used simple word lists to filter for sentences that contain viagra adverts, for example. But spammers quickly wise up to which mails are getting rejected, and so change the spelling of the words – perhaps to [email protected]@.
“Eventually, we figured that out, made our junk mail filters see ‘Viagra’ with the ‘@’ sign,” he says, “And then spammers would try something else clever. This time, they would embed the word “Viagra” in a bitmap. Again, we, the humans, can see the message, but our filters can’t.
“So this went back and forth for quite a while. And at some point, we said, “We’ve got to step back here and think a little bit more strategically.”
The one constant in all spam, Heckerman explains, is that at some point someone is going to try and elicit money. So instead of parsing mail for words, Microsoft’s filters now parse for links and cross-reference that with a black list of sites known to be used for fraudulently extracting cash and card details.
“It turns out there’s a striking similarity between fighting spam and fighting HIV. Just as spammers change or mutate their messages to work around our filters, just as I showed you, HIV mutates itself to avoid attack by the immune system,” says Heckerman, “And Bruce Walker and I are now essentially using the same high-level strategy to go after HIV: namely, to go after its weak link or weak links.”
Heckerman says that his team began looking for the points in the HIV molecule that don’t mutate, and by comparing data from the immune systems of people who have an unusual resistance to HIV with those of people who get very sick, they’ve been able to identify six points of weakness that could prove vulnerable to a vaccine. The last challenge, Heckerman says, is overcoming differences in people’s immune systems to encourage individual’s bodies to fight HIV in the right way.
You can read a transcript of Heckerman’s presentation here (it’s about halfway down) and there’s more information on the project at his blog here. We’ll be getting in touch to find out more as soon as he lands in South Africa.
In the meantime, here’s a video of Heckerman on this subject from a couple of years ago – good to see that the research is coming to fruition.
(Main image: Shutterstock)