Creating a Gold Standard for Sentence Clustering in Multi-Document Summarization