• We want to know whether a group of similar refs is made up of free-form text, or if the body of the footnote is a template. This logic already exists in the scraper but needs to be crossed with the similar ref finding logic. Find a convenient format for outputting the result.
• If similar refs are made up of templates, can we identify the template parameters which differ? Find a convenient format for outputting that info.
• Currently we get footnote bodies from the reflist, which does not include data-mw. We need data-mw to analyze template transclusion, this should be available in the reflist item body data-mw.

Examples:

• {{Cite book|title=Book name|page=12}} vs. {{Cite book|title=Book name|page=24}}
• Book name, p. 12 vs. Book name, p. 24
• Series X: unique title of issue vs. Series X: yet another totally unrelated title