Constructing A Dataset of Support and Attack Relations in Legal Arguments in Court Judgements using Linguistic Rules.