The snippets we generate actually contain stuff from within tables, image captions, and headings. These don't look great. If we could smash those into another field then the snippets would be nicer. We could also use the sentence fragmenter in the experimental highlighter.
Note: we already do this for the headings. We should do it for tables and infoboxes and stuff. Maybe we should do it for a css class as well.
Version: unspecified
Severity: normal
See Also:
https://bugzilla.wikimedia.org/show_bug.cgi?id=61669