HW 1: a Facebook network: Instructions Help Center

Getting started with SNA

For this first assignment, you can choose between working with my Facebook network or downloading your own. I believe that working with your own network will make it easier to grasp concepts. In order to automatically grade your assignment, we will need an uploaded copy of the network. We will not examine individual networks by hand (the assignment will be graded programmatically), the data will only be used to grade the assignment.

If you are not a Facebook user, or if you prefer not to work with you own network (or if you have >= 500 Facebook friends and the grading script can't handle your network), you can use my anonymized network (the names have been replaced, but there is an interesting literature showing that social networks cannot be completely anonymized).

My Facebook network is available as LadaFacebookAnon.gml

In order to get your own network, complete the following steps:

  • Go to http://snacourse.com/getnet
  • Choose which user data (e.g. "wall posts count") you'd like to include, for this assignment no additional data is necessary (meaning that you can leave all checkboxed unchecked), but whatever you do download, you can visualize/analyze (profile age rank: oldest profile = highest value, declarative intensity (length of text in fields like activities, books, etc.)
  • Be patient. If you have many Facebook friends, the app will fetch the data gradually, and report on its progress (e.g. (30 out of 148, 120 out of 148...). Once it is done, at the bottom of the page it will provide a download link. Right click on 'gml file' and save it to your computer.
  • Open Gephi and load your file using "File -> Open...".

If your network, like mine, has disconnected components, you will most likely want to use Force Atlas 2 as your layout algorithm. To keep the disconnected components from flying off, set "Gravity" under "Tuning" to ~50. If it is overly clumped together, you can try checking "Dissuade Hubs" and/or "Prevent Overlap". It's up to your own aesthetic sensibilities.

Hint on calculating the size (i.e. number of nodes) of the giant component: Go to Filters (in the middle/right of the Gephi window). Select Topology->Giant Component. Drag the filter down where it says Queries: Drag Filter here. Then click on the 'Filter' button with the green right arrow. You should see your visualization change (only contain the giant component). Look in the upper right corner for the number of nodes and edges that are in the giant component.
Note: Make sure you reset the filter (click on 'filter' again) to answer the other questions about the whole network (the number of nodes, edges, maximum degree etc.), otherwise you will be reporting on the numbers for the giant component and your answer will be incorrect.

Uploading your assignment

You will need to upload 2files:
(1)Fill in the template with values according to what you observe using Gephi. Upload the template under 'Output Submission', making sure it is still in plain text format. Tip for making sure your file is in plain text. In MS Word or TextEdit, select 'Plain text (.txt)' under 'output format'.
(2) Upload the GML file under 'Additional Submission'. The grader will only be able to understand GML files. Similar apps will generate GDF files. Although Gephi happily works with these, the grader for this assignment does not, so please use the GETNET app for this assignment. If and only if you are itching to learn igraph and/or NetworkX, consider doing the assignment with either of those instead. I've provided a starting script for networkX (in Python), and a starting script for igraph in R.