University of Hagen at GeoCLEF 2008: Combining IR and QA for Geographic Information Retrieval

This paper describes the participation of GIRSA at GeoCLEF 2008, the geographic information retrieval task at CLEF. GIRSA is a modified and improved variant of the system which participated at GeoCLEF 2007. It combines results retrieved with methods from information retrieval (IR) on geographically annotated data and question answering (QA) employing query decomposition. For the monolingual German experiments, several parameter settings were varied: using a single index or a separate index for content and geographic annotation, using complex term weighting, adding location names from the narrative part of the topics, and merging results from IR and QA. The best mean average precision (MAP) was obtained by combining IR and QA results (0.2608 MAP). For bilingual (English-German and Portuguese-German) experiments, topics were translated via various machine translation web services: Applied Language Solutions, Google Translate, and Promt Online Translator. Performance for these experiments is generally lower than for monolingual experiments. For both source languages, Google Translate seems to return the best translations. For English topics, 60% (0.1571 MAP) of the maximum MAP for monolingual German experiments is achieved. For bilingual Portuguese-German experiments, 80% (0.2085 MAP) of the maximum MAP for monolingual German experiments is achieved.