| The incredibly large and fast growing amount of biomedical literature causes a strong demand for effective IR systems in the life science domain. Terminology related problems such as polysemy, homonymy, frequent use of multi-word terms and abbreviations make effective retrieval a challenge and incorporating external knowledge essential. In the past, various approaches have been proposed to incorporate conceptual knowledge from biomedical ontologies. However, it is still non-trivial to actually improve retrieval performance using this knowledge. I propose to investigate the problem of biomedical IR from a cross lingual perspective, in which two languages are present: a textual and a conceptual. The problem of incorporating conceptual knowledge then comes down to integrating translation and matching the different representations. |
