From ``Identical'' to ``Similar'': Fusing Retrieved Lists Based on Inter-Document Similarities

Abstract

Methods for fusing document lists that were retrieved in response to a query often utilize the retrieval scores and/or ranks of documents in the lists. We present a novel fusion approach that is based on using, in addition, information induced from inter-document similarities. Specifically, our methods let similar documents from different lists provide relevance-status support to each other. We use a graph-based method to model relevance-status propagation between documents. The propagation is governed by inter-document-similarities and by retrieval scores of documents in the lists. Empirical evaluation demonstrates the effectiveness of our methods in fusing TREC runs. The performance of our most effective methods transcends that of effective fusion methods that utilize only retrieval scores or ranks.

Fusion from a Condorcet Perspective

“…We use the Best Runs and Random Runs settings as past work on fusion demonstrated different important properties for these settings which resulted in different fusion performance patterns. A case in point, the overlap between relevant documents in randomly selected runs is much lower than that for the best runs (Kozorovitzky and Kurland, 2011). This overlap affects considerably the effectiveness of fusion methods (Beitzel et al, 2003;Kozorovitzky and Kurland, 2011).…”

“…A case in point, the overlap between relevant documents in randomly selected runs is much lower than that for the best runs (Kozorovitzky and Kurland, 2011). This overlap affects considerably the effectiveness of fusion methods (Beitzel et al, 2003;Kozorovitzky and Kurland, 2011).…”

“…Another observation we make based on Figure 3 is that the ratio is higher for Best Runs than for Random Runs. This could be attributed to the higher overlap -specifically, at top ranks -between the most effective runs with respect to that for randomly selected runs (Kozorovitzky and Kurland, 2011).…”