Context in the Web of Trust

by Dan Zambonini

My last blog entry was a throw-away, personal opinion (perhaps too opinionated) on the contents of Computing courses. What I didn't expect was for it to get Slashdotted, and the number of personal insults - both online and in my email.

This got me thinking to how the Semantic Web could improve on the situation. Within a few days, when Google has caught up, a search for my name may return a number of Slashdot comment results with terms such as "idiot", "ignorant" or "doesn't know what he's talking about". They're probably right, but that's not the point. An average person, looking at these results, will conclude that I am indeed an idiot.

With the Semantic Web, the Web of Trust will theoretically allow us to state, through machine-readable metadata, who we do and don't "trust". So, for example, my bank and my employers could create files that declare their trust for me. If you then needed to ascertain whether or not my ramblings are trustworthy, your browser/agent could - in theory - analyse the chain of trust statements that lead to me, and present you with information to help you decide.

But trust has context. My bank may trust me with financial decisions. My employer may trust me with systems architecture decisions. But nobody may necessarily trust me on my ability to analyse the state of Computer Science...

So, this is more of a question really for the Semantic Web crowd; do the current plans for the Web of Trust logic include different contexts of trust?

Do you have any thoughts on the Web of Trust? Is it open to abuse? Can context be accounted for?