Obsidian’s Graph Analysis plugin

This is a follow up article to Using Obsidian’s Graph view for real. Now I want to look at a massive enhancement using analytics: the community plugin Graph Analysis!

The very detailed videos and documentation might feel a bit daunting, so I’ve broken down what you can get out of Graph Analysis, along with some primary testing. Note, I also used the nlp add-on that allows Graph Analysis to compare notes based on the words in them, not just their place in the structure of the graph.

Use-case 1: Find very similar notes / clones

There are six sorts of algorithms to find similar notes, they are:

• Co-citations
• Adamic Adar
• Jaccard
• Overlap
• Bag of Words (nlp)
• Otuska Chiai (nlp)

To investigate them, I first set up an original note (“Original”), a note with a different title but exactly the same content (“DTSC” for different title, same content), and a third exactly the same as the original, but with a “2” at the end of the title (“STSC” for similar title, same content).

Then I selected Original and checked how the algorithms DTSC and STSC in terms of their similarity to Original. 5 of the 6 algorithms correctly identified the other 2 notes as the most similar. Co-citations did not, probably because the incoming links did not also point to DTSC and STSC. (here’s a very in-depth video…