Validate DBpedia taxonomisation

[georgerichardson]

Assessing between methods

[ampudia19]

I presume it's best both Emily and I are assigned to this one.

[emily-bicks]

Methods to assess the quality of the taxonomy groupings:

• [ ] distribution of tokens per group (if the distribution is too skewed it likely won't be useful for our purposes)

[ampudia19]

• [x] Apply taxonomy to & validate on ARIA DBpedia entities.
• [ ] Make a selection on the final taxonomy.
• [ ] Code refactoring: Are we still opting for generators?

[emily-bicks]

Potential validation metrics:

• Skewness of distribution of entities/topics
• Depth of trees
• Number of topics / journal
• Relative difference in variance of embeddings / level
• Variance of resulting taxonomy based on initialisation / hyperparameters / introducing noise and checking if known words are always ending up in the same category
• Measure stability with different sample sizes: distribution of