The　presence　of　a　schema　for　XML　documents　has　numerous　advantages.　Unfortunately,　many　XML　documents　in　practice　are　not　accompanied　by　a　schema　or　a　valid　schema.　Therefore,　it　is　essential　to　devise　algorithms　to　infer　schemas.　The　fundamental　task　in　XML　schema　inference　is　to　learn　regular　expressions.　In　this　paper,　we　focus　on　learning　the　subclass　of　RE(&)　called　SIREs　(the　subclass　of　regular　expressions　with　interleaving).　Previous　work　in　this　direction　lacks　inference　algorithms　that　support　inference　from　positive　and　negative　examples.　We　provide　an　algorithm　to　learn　SIREs　from　positive　and　negative　examples　based　on　genetic　algorithms　and　parallel　techniques.　Our　algorithm　also　has　better　expansibility,　which　means　that　our　algorithm　not　only　supports　learning　with　positive　and　negative　examples,　but　also　supports　learning　with　positive　or　negative　examples　only.　Experimental　results　demonstrate　the　effectiveness　of　our　algorithm.