In this paper we give a brief overview of the Webis group's participation in the TREC 2015 Tasks and Total Recall tracks. In the task understanding subtask of the Tasks track, we use di erent data sources (AOL query log, Freebase, etc.) and APIs (Google, Bing, etc.) to retrieve topics related to a given query. All sources are combined in our SQuare system. The task completion subtask is based on combining the results of our ChatNoir 2 for the di erent topics identified in the task understanding

more »

... subtask. Finally, for the ad-hoc subtask (similar to the previous years' Web tracks), we use an axiomatic re-ranking approach of the search results obtained from ChatNoir 2. In the Total Recall track, we employ a simple SVM baseline with variable batch sizes equipped with a keyqueries step to identify potentially relevant documents.