Without using a community detection algorithm, suppose that i have a telecom social network of 877 nodes and 37025 edges. Roughly speaking, what would be the distribution of nodes and links? Put in other words, what would i expect from this network?
Can i say for example that since i have 877 nodes, i should expect to see roughly 87 communities with each community consisting of 10 nodes?
Any suggestion will be appreciated.
It all depends on several factors. Most importantly is what do you consider a tie between two people/nodes. If you put the bar low enough then everyone is connected and you have 1 giant hairball (1 poorly connected community). The second factor is what type of data do you have: one-way or two-way? One way (X calls Y) and Two way (X calls Y and Y calls X). Do you have frequency of calls? X calls Y daily and X calls Z weekly? The more data you have about the calls the better you can filter the communities. Finally, what was your data source? Was it one telecom company, or all of the telecom companies working in a specified area/country? If your data is only from one provider, you will be missing a lot of connections and whatever communities you find will be incomplete!
Hi.,
A community can be defined as a group of entities closer to each other in comparison to other entities of the dataset. A community is formed by individuals such that those within a group interact with each other more frequently than with those outside the group.
For more info:
https://pdfs.semanticscholar.org/1c09/50130580842a6e574d62cab23d2f89724594.pdf
Add some mathematics here.
You are looking for a tree from each social network user. Keep all connections from/to this user.
Then from this tree extract a minimum spanning tree (MST).
Compare MSTs of two users connected by an MST.
Selecting the strongest connectivity,you can aggregate communities
- Badges
- Science topic
- Similar topics
- Political Science
- International History and Politics
- Globalization