An Empirical Comparison of Goodness Measures for Unsupervised Chinese Word Segmentation with a Unified Framework