Complex Network Analysis with Instagram Data

This project implements the creation and visualization of a Complex Network of hashtags (also refered here as tags) on Instagram posts about politics in Brazil. An article about it at Medium.com is available here.

The project works with some concepts about the process of data collection and construction, handling and visualization of Graphs.

Jupyter Notebook is required to collect and analyse the data. Gephi is required to open, handle and visualize the graphs (and it requires Java).

Contents

This section describes the contents in the repository.

Notebooks

• 2019-06-10-1-macj-instagram-tags.ipynb: designs and tests of functions to collect data from Instagram

• 2019-06-10-2-macj-collecting-data.ipynb: performing data collection and saving to JSON file

• 2019-06-11-1-macj-graph-scratch.ipynb: analysing the data, creating and exporting graphs

The data collected on 2019-06-11 is saved to the file data.json.

GraphML Files

The GraphML files represents the networks generated.

The number in the files means the number of posts in the analysis (must be changed in future works).

• edges_counted_100.graphml: graph with edges between all tags without node weight

• edges_counted_100_dropped.graphml: graph with insignificant edges dropped

• edges_counted_keys_100.graphml: graph with edges between key tags and all others without node weight

• edges_counted_100_nw.graphml: graph with edges between all tags with node weight

• edges_counted_100_dropped_nw.graphml: graph with insignificant edges dropped

• edges_counted_keys_100_nw.graphml: graph with edges between key tags and all others with node weight

All these files can be opened directly in Gephi to perform other analysis.

Gephi Files

Gephi files contains the networks with filters, colors, scales and layouts applyied.

Bellow are short descriptions for each one.

The following views use GraphML files without node weights:

• v01-all-edges.gephi

  • Workspace 3: original imported network

  • Workspace 6: Force Atlas 2 layout

  • Workspace 7: Force Atlas 2 layout

• v02-key-edges.gephi

  • TODO

• v03-key-edges-fruchterman.gephi

  • Workspace 2: Fruchterman layout

The following views use GraphML with node weights:

• v04-key-edges-node-weight.gephi

  • TODO

• v05-all-edges-node-weight.gephi

  • Workspace 1: Fruchterman layout with Expansion and Adjust Label

  • Workspace 3: Force Atlas layout

  • Workspace 4: Force Atlas 2 layout

  • Workspace 5: Fruchterman layout for long time

  • Workspace 6: original imported network