Hello, Wikidata dumps (e.g this http://dumps.wikimedia.org/wikidatawiki/20140612/) have an annoying plus one named Yahoo abstracts, It has more than 16 GBs (mainly because it's not zipped) and because content of Wikidata pages are saved in term of numbers and codes instead of wikitext (e.g. this https://www.wikidata.org/w/api.php?action=wbgetentities&ids=Q42), usually the abstract is as the same as whole page.
Please fix it. Best
xmldatadumps-l@lists.wikimedia.org