IR4.25 Why pseudo feedback works - Victor Lavrenko - AI深度探索：從機器學習到NLP精通