It shouldn't be too hard to locate publication information on papers and extract the year. That could then be used to ignore old papers.