ICTNET at Session Track TREC 2012
CHINESE ACADEMY OF SCIENCES BEIJING INST OF COMPUTING TECHNOLOGY
Pagination or Media Count:
In this paper, we describe our solutions to the Session Track at TREC 2012. The main contribution of our work is that we implement the learning to rank model to re-rank the documents retrieved by our search engine. We notice that Huurninket al. have used learning to rank algorithm to model session features at last years Session Track. Due to lacking of training data, their model did not outperform substantially than others. Intuitively, we use last years session data for tuning the weights of ranking features. Meanwhile, we define several useful features to model session search intent.
- Information Science