Using Keyqueries to Reduce Misinformation in Health-Related Search Results

Maik Fröbe, Sebastian Günther, Alexander Bondarenko, Johannes Huck, Matthias Hagen
2022 European Conference on Information Retrieval  
In the scenario of health-related searches, we investigate whether explicit relevance feedback by experts can guide query expansion methods to formulate queries that return fewer misleading or wrong results. In contrast to standard query expansion methods that pay no attention to the ranks of the feedback documents in the results of the expanded query, we experiment with a keyquery-based approach to identify expanded queries for which the feedback documents are ranked as high as possible.
more » ... ments on the TREC 2019-2021 Decision and Health Misinformation tracks show that our keyquery-based method substantially reduces the portion of harmful results and improves the overall retrieval effectiveness.
dblp:conf/ecir/FrobeGBHH22 fatcat:gxt4zabryze5vlytzxgcaqaofi