ICTNET at Microblog Track in TREC 2014

Microblog track was first introduced in 2011 and we have participated in this task for 4 years [1,2,5]. This year s microblog track has two tasks. The first one, namely ad-hoc search task, is the same as usual. This task needs to retrieve all the tweets that are relevant to query Q before time T. Participants can access the corpus by official APIs. The second task is Tweet Timeline Generation(TTG) task. It is newly introduced this year and the main goal of it is to detect and remove the redundant tweets the first task retrieves. This report is organized as follows. Section 2 mainly focuses on the data preparation. Section 3 is our methodology and framework of the ad-hoc search task. Section 4 focuses on the methodology of TTG task. Section 5 gives the final results of the two tasks. Presented at the Twenty-Third Text REtrieval Conference (TREC 2014) held in Gaithersburg, Maryland, November 19-21, 2014. The conference was co-sponsored by the National Institute of Standards and Technology (NIST) and the Defense Advanced Research Projects Agency (DARPA).