[hooande]

I've definitely noticed a trend towards graph based thinking in recent computer science publications. My half baked theory on this is as follows:

Graph based thinking is a result of the rise of social networking. The term "social network" wasn't common until I was a senior in college. Back then most computer scientists thought of things in terms of matrices - rows and columns. After 2004 when facebook became the most popular software in the college universe, people became much more interested in graphs (social and otherwise). I believe this lead many young computer scientists to start thinking in terms of graphs - vertices and edges. If you read comp sci papers written by people over the age of 30, many of them still express things in terms of matrices.

In my understanding, graphs can be faster to process and in many cases easier to traverse. But I believe that the shift in thinking has more to do with popular trends in software than it does any technical advantage of graphs over other ways of thinking.

[jey]

No, your premise is wrong. Graphs are ancient within CS.

Linear algebra and graph theory do have some deep connections though.

[kvs]

Perhaps people stopped talking about matrices because social graphs are sparse. Also, most of the time papers (esp. system papers) are written to solve a problem and if the paper proposes a solution to solving a social graph related problem then that's what the author will use because it's easier for readers to relate.