It appears that the friend of a friend was the driver in the terribly unfortunate tragedy in Santa Monica. I’ve been thinking about possibilities for Friendster and FOAF lately, and for some reason this sad event triggered a tangential idea – automated tools for tracking news about friends.
If many of your friends maintain weblogs, subscribing to their syndication feeds would obviously go a long ways towards keeping you up to date on their activities and ideas, or at least the ones that they considered blogworthy. Info in a syndication feed is fairly structured to begin with, and will likely become even more structured as Atom/Echo/Necho/Pie evolves and takes hold.
But, news sites typically contain less structured content, or at least the structure is wildly inconsistent from site to site. There are a few proposed XML standards like NewsML, but they are a long ways off from common adoption. Even once adopted, these standards will likely just simplify somewhat the search of the content. Of course, if enough of the news sites you want to search have syndication feeds, you could at least start with them.
Obviously, an incident such as the one in Santa Monica will draw a huge amount of publicity. If a friend of yours is involved in a major news event, you will likely find out soon enough. But, what about the smaller news stories? I want to know if any of my friends from high school or college are in the news, especially if it is good news, and even more so if it is a friend with whom I have lost close touch over the years.
On a less happy note, I want to know about the bad news, too, primarily to avoid putting my friends’ families and housemates into uncomfortable situations. I would hate to call up an old friend in a cheery voice, only to have his or her spouse/partner/children/etc. have to explain to me why my friend won’t be coming to the phone. It’s not like this happens often or has actually ever happened to me, but I suspect when I get much older it will become more common. Maybe I’m just being paranoid. Don’t worry, I don’t lay awake at night worrying about things like this.
In addition, I don’t want to manually search these feeds. I want a tool that will automatically search a large number of feeds. Feedster and other tools like it could be part of the solution. Google via the Google API could also play a role for sites that don’t have feeds, but you will need to separate old news from new news. If you can use the API to search just Google News, that might help.
So how do Friendster and FOAF come in? Well, if you currently use them, you’ve already gone to the trouble to identify some of your friends in a very structured way. The tracking tool I’m imagining could use that info for the source of names. Of course, if you have a friend named Joe Smith, you’ll probably get a lot of hits that aren’t about the Joe Smith you know. So, the tool would use the profile of your friend to estimate the odds the story is actually about your friend. If a threshold is exceeded, the news story is included in your “Daily News about My Friends” page.
One relatively simple piece of useful data is geographical address. Let’s say the Joe Smith you know lives in Des Moines. If the news story appears in the Des Moines newspaper, the score for this article would be increased. If you have included your friend’s address in their profile and the story also mentions the street your friend lives on, the score goes way up. Any other matches between your friend’s profile and the story would also add to the score.
Now, we can throw in some technology made famous by the war against Spam email – Bayesian filters. Next to each story on your “Daily News about My Friends” page, there would be a control for you to indicate if the story was or was not about your friend. Over time, the system would get better at distinguishing the Joe Smith you know from all the Joe Smiths you don’t know.
So, is this a good idea, a scary idea, or a good and scary idea?