Construction of a Benchmark Data Set for Cross-lingual Word Sense Disambiguation