UNED at WebCLEF 2008: Applying high restrictive summarization, low restrictive information retrieval and multilingual techniques

Abstract

This paper describes our participation in the WebCLEF 2008 task, targeted at snippet retrieval from new data. Our system assumes that the task can be tackled as a summarization problem and that the document retrieval and multilinguism treatment steps can be ignored. Our approach assumes also that the redundancy of information in the Web allows the system to be very restrictive when picking information pieces. Our evaluation results suggest that, while the first assumption is feasible, the second one is not always true. © 2009 Springer Berlin Heidelberg.