This is also useful for Wikivoyage.

However, I don't think it's a good idea to hardcode a list of frequent section titles.

What do you mean by "hardcode"?

The way I see it, the headings will be added by default if any text is written in following paragraph, and the translator will be able to edit them just like any other text.

How can we get the list of frequent section titles?

For a precise number we could analyze dumps, although it would probably take a very long time given that they are so huge in some languages.

Or maybe @ssastry and @cscott have a way to query Parsoid data quickly?

Of course, we could start from some intuitive ones: "Biography", "History", "Early life", "Personal life", "Awards", "Bibliography", "External links", "References", "History", "Legacy", "Death", "Geography", "In popular culture", etc.

Is this the sort of thing you're looking for? Takes few minutes to generate with something like find /public/dumps/public -path '*/201506*/*pages-articles.xml.bz2' -print0 | xargs -0 -n 1 -I '{}' jsub -cwd LC_COLLATE=C bzgrep -HE '^==' {} | sort | uniq -c | sort -nr | head -1000 >> frequent-headers.txt -> F191888.

As part of the upcoming work on Section translation, we may have access to section mapping data which may enable to translate sections in this way.