top of page
Writer's pictureAsaf Shapira

Episode 8: The Episode for the Aspiring Network Scientist w/Michele Coscia

Updated: May 19, 2021

What's the rumpus 😊 I'm Asaf Shapira and this is NETfrix, The Network Science Podcast. Today we're going to have a talk with the one and only Michele Coscia, an associate professor at IT University of Copenhagen by day, and by night, the author of the "Connecting Humanities" Blog for Network Science. In between, he managed to squeeze out a monumental book titled "The Atlas for the Aspiring Network Scientist", which we'll be on our to-talk-about-list.


We'll also review some of his recent papers, talk about the upcoming Network Science conference and the satellite event he's involved with. Together we'll make history by telling the first network science joke in a podcast and lastly, we'll try to uncover the mystery behind Italian's obsession with Network Science and podcasts.

 

Here are some of the concepts that were mentioned in the episode:


Bipartite network: A network with 2 sets of nodes/vertices in which a node from set A can only be linked to a node from set B. Usually it means that each set has a different attribute.

Personally, I don't like bipartite networks because their unique nature can change the basic assumptions that many network algorithms rely on. In order to get things "back to normal", we can use another concept we mentioned which is "Projection". Projecting means turning a network to a one-set-of-nodes kind of network, or just a plain network. To do so, we'll need to get rid of one of the sets. In the example above, we'll need to get rid of the squares (customers) to see which products are connected to other products. The rationale behind it is that connection to the same node is a common trait and can be considered as a link or an edge. We'll achieve it by connecting products only if they are connected to the same node (customer). That way we'll get a network that consists only products, and now we can: - Apply the standard network algorithms to the network.

- Use it to as a recommendation system to customers ("people who bought "A" also bought "B"). I strongly recommend reading the relevant chapter in Coscia's book we mentioned about the subject.

SBM or Stochastic Block Model: Honestly, neither Michele nor I felt like explaining it, or in Michele's words: "Aaron Clauset's is the person I trust more when it comes to SBMs, so his slides should help".

Time Windows: A concept we've covered in NETfrix's episode 6A about dynamic networks.


BFS or Breadth-First Search: A search strategy from graph theory that explores all of the neighbor nodes at the present depth prior to moving on to the nodes at the next depth level (The "opposing" strategy is DFS - Depth-first-Search)

Dijkstra Algorithm: A great way to find the shortest path in a weighted graph.

You can check this site for cool visualizations of Dijkstra at work.

 

And here is a list of some of the name-dropping that has been carelessly done throughout this episode:


Here are the rest of the links (papers & datasets that were mentioned):

Did you enjoy, and want to share? Have you suffered and you do not want to suffer alone?

Tell your friends or rate us on apple podcasts, follow us on Twitter or like our Facebook page. If you're from Israel, rate us on Podcasts of Israel. The music is courtesy of Compile band. check them out! See you in the next episode of NETfrix (:

195 views0 comments

Comments


bottom of page