Microblog Track 2011 of FDU

Twitter provides huge amount of short messages, raises challenge problems to the research community. The Microblog Track of TREC detects the special behavior of the twitter dataset in the “real-time” retrieval task. This paper reports our participation in the Microblog Track task. Given the query topics, each participants are required to conduct a “real-time” retrieval task, which seeks for the most recent and interesting tweets for each query topic. Our focus in this task includes two aspects: (1)data preprocessing to remove non-English tweets, and (2)feature extraction for clustering the tweets into two categories. Given the huge interest in the microblog, there is lot of work to apply different linguist analysis techniques and data analysis methods to explore the behavior and special features in the Microblog sphere.