LSIS at TREC 2012 Medical Track - Experiments With Conceptualization, a DFR Model and a Semantic Measure

In this paper, we present our participation in the Medical Records Track of TREC2012. We focus on the impact of combining the word space and the concept space in the information retrieval process. For this track, we submitted a baseline run by employing the In_expC2 weighting model implemented in the Terrier platform, which achieved fair results (0.304 MAP, 0.51P@10). Then, we expanded the documents by performing automatic text conceptualization using UMLS and the MetaMap software on medical records. These textual and conceptual representations, still using the DFR model, led to precision (0.29 MAP, 0.47 P@10). We also automatically extended the topics with UMLS concepts. This led to a lower precision (0.27 MAP, 0.46 P@10) Lastly, we experimented the usage of semantic IR measures only (0.21 MAP, 0.41 P@10)..