Fix source--target node designations for code using Cora dataset

[geoffj-d61]

Description

The Cora README documentation from the dataset specifies that the column order is: cited-paper; citing-paper. However, at least one Jupyter notebook (namely cora-links-example.ipynb) assumes the column ordering: source; target.

Notwithstanding that this example assumes undirected edges, the constructed graph should accurately reflect the directed relationships inherent in the data.

How to reproduce (the error):

The relevant code snippet in the cora-links-example.ipynb notebook is:

edgelist = pd.read_csv(os.path.join(data_dir, "cora.cites"), sep='\t', header=None, names=["source", "target"])

Aim

To locate all code examples that load the Cora dataset, and ensure the loaded graphs utilise the correct target--source citation ordering.

• Bug fixed
• Branch and Pull Request build on CI
• Branch and Pull Request pass unit tests on CI
• Branch and Pull Request pass integration tests on CI
• Version number reflects new status
• Peer Code Review Performed
• Code well commented
• CHANGELOG.md updated

[geoffj-d61]

@adocherty, could you please review this ticket and let me know if I've made any goofs, or if it seems okay?

[adocherty]

Hi @geoffj-d61,

This looks good to me. I point out that the directionality here is dependent upon the semantics of the link, the current links are a "cited_by" relationship, you are proposing to change this to a "cites" relationship. When we make this change it would be good to add a comment about the semantics of the link.

Note also that a lot of the demo notebooks use CORA with the same code, so we should change all of them.

[geoffj-d61]

@adocherty, agreed - I intend to review and fix all occurrences in the demos. Also, I'm not exactly changing to a "cites" relationship - this is already explicitly assumed in the notebooks; I am merely correcting the loaded directionality. While I am about it, I am also testing demos in python 3.6/3.7 and trying to understand them.