“…There is a large body of work on re-ranking an initially retrieved list using graphbased methods that model inter-document similarities within the list (e.g., Dani lowicz & Baliński, 2000;Baliński & Dani lowicz, 2005;Diaz, 2005;Kurland & Lee, 2005Zhang et al, 2005;Yang, Ji, Zhou, Nie, & Xiao, 2006). As mentioned in Section 2, our fusion methods could conceptually be viewed as a generalization of some of these approaches (Dani lowicz & Baliński, 2000;Diaz, 2005;Kurland & Lee, 2005); specifically, of methods that utilize both retrieval scores and inter-document-similarities for modeling relevancestatus propagation within the list (Dani lowicz & Baliński, 2000;Diaz, 2005). A similar relevance-status propagation method was also employed in work on sentence retrieval for question answering (Otterbacher, Erkan, & Radev, 2005).…”