Structural Analysis of Social Networks: A Programmer's Approch
- At: Workshop on Data Science & Machine Learning (DSML-2017)
- Place: Center for Soft Computing Research, Indian Statistical Institute
- City: Kolkata, India
- Workshop Dates: March 28-31, 2017
- Title Presented on March 30, 2017
Social network analysis is the field of study which deals with the analysis of relational data using the help of graphs. Research scolars tried to used structural analysis of social networks to infer social behaviors of network actors. The tutorial presented in DSML 2017 was to demonstrate how to kick start with social network data sets. For coding demos, I used the networkx package of python. This tutorial avoids advanced algorithms for addressing the multi-disciplinary audience of the workshop.
Click inside and use ‘space' to forward the presentation. Use ‘q' for hide/show footer. As there is no audio, I have added some notes inside the slides when require. For a full screen presentation click here.
This part contains the data set information and hands on coding preview.
You can download the presentation notebook and access it from your local pc through jupyter. Or you may copy the url (http://sumankundu.info/documents/presentations/dsml-2017/DSML-2017-At-Lecture.ipynb) and view its content using the online nbviewer available here.
- Useful Links
- What is the difference between traditional graph and social networks?
It is the size, complexity and charecteristics it possess. For example, small world phenomenon, power law like degree distribution etc. Although it is a graph, but due to its complexity the analysis need different techniques.
- Six degrees of separation.
Although it is said that you can reach to anyone in the world by six hop in reality it is not exactly the six. Networks do shows a small world charecteristics. However, the diameter is not always six. Sometimes it is found to be 4 some times 15, but mostly less than 10.
- Suggest to include GUI tools
Gephi, Cytoscape, Pajek are some good tools for analyzing networks graphically.