v1.1

Improvement: The Social-Textual method is now based on a TFIDF-based method (ClassicSimilarity) instead of a language-modeling-based method. The previous implementation was probably inconsistent with the Social-Textual paper.

v1.0

Some fixes