The amazing things about great graphs of our times is their enormously rich structure. A single link on a web page does not mean much by itself, but hundreds of billions  and trillions of them add up to something truly great, which is what Google brought to the world.

The power of graphs comes from their structure, of many links and paths connecting nodes.  It is the multitude of these paths and different ways connecting nodes that enables us to do things such as ranking based on eigenvalue analysis.

Contemporary AI graphs, such as the Knowledge Graph, have nodes and some edges associated with them, but not much structure in terms of paths. The nodes are concepts such as people, places, movies etc. which have associated attributes such as names, locations, directors, actors but there is not much beyond that.

How do we go from one movie to another, apart from clustering techniques such as collaborative filtering? Or from a movie to another node? And how do we get many more nodes that are part of the world we live in, with its immense richness?

The answer is reasoning, with its vast endless richness built up by our civilization. It is high time we start leveraging that.

Categories: AI

Leave a Reply

Your email address will not be published. Required fields are marked *