4 comments

[ 2.9 ms ] story [ 20.6 ms ] thread
> Axiom Zen is working with Timeline.com on a very ambitious new project to present a unique perspective on current events by providing important historical context.

Can DBpedia encode dependencies between historical events, e.g. event A @ location/time caused events B & C? Or Person M alleged a casual relationship between events A & C?

Semantic Databases can easily encode that kind of statement, adding any new kind of relationship is quite straightforward.

That particular relationship is outside the goals of DBpedia, which is primarily intended to represent the "summary blocks" that appear in the upper-right corner of major pages on Wikipedia.

Freebase allows anyone to make new relationships, though, so you could go into Freebase today and encode those dependencies if you had sufficient source data. This leads to it's own problems! It's common in Freebase to find the same conceptual relationship encoded using multiple different relationship classes. For example, we've seen more than ten different ways of encoding geolocations within this one dataset!

Semantic Databases are remarkably flexible, but -- as with Javascript! -- this flexibility can be its own curse...

The write up is about an engineering problem (comparing two large data sets) which is not evident from the title "Making a Big Difference" .