Condorcet fusion for improved retrieval

By Mark H. Montague, J. Aslam. 2002

337 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Information Systems, Information Retrieval & Search, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

1. Craig Macdonald
2. Rodrigo Nogueira
3. Andrew Yates
4. C. Clarke
5. Wei Yang
6. Hong Yu
7. Jimmy J. Lin
8. Iadh Ounis

In Information Systems

1. Jimmy J. Lin
2. Rodrigo Nogueira
3. Wei Yang
4. Andrew Yates
5. Kyunghyun Cho
6. Craig Macdonald
7. Zeynep Akkalyoncu Yilmaz
8. Iadh Ounis

In Information Retrieval & Search

1. Jimmy J. Lin
2. Rodrigo Nogueira
3. Andrew Yates
4. Zeynep Akkalyoncu Yilmaz
5. Wei Yang
6. Kyunghyun Cho
7. Jamie Callan
8. Zhuyun Dai

In Signal Processing

1. Jimmy J. Lin
2. Rodrigo Nogueira
3. Andrew Yates
4. Craig Macdonald
5. C. Clarke
6. Wei Yang
7. Hong Yu
8. Iadh Ounis

Citations in the corpus

• Pretrained Transformers for Text Ranking: BERT and Beyond (2020, methodology): “. While the effectiveness 105https://ir.nist.gov/covidSubmit/archive.html 102 of ensemble and fusion techniques is not a new observation in search (above references, as well as [Bartell et al., 1994, Montague and Aslam, 2002]), seeing that these general findings are reproduced in a new context enhanc”
• Multi-Stage Document Ranking with BERT (2019, methodology): “The B I - NARY method is inspired by the Condorcet method (Montague and Aslam, 2002), which is a strong aggregation baseline (Cormack et al., 2009).”
• Unsupervised ensemble ranking of terms in electronic health record notes based on their importance to patients (2017, methodology): “CombSum 0.302 0.202 0.225 0.253 (P=.03) 0.335 0.266 (P=.01) 0.884 0.813 CondorcetFuse 0.264 0.168 0.191 0.218 (P .001) 0.277 (P .001) 0.225 (P .001) 0.819 (P .001) 0.753 (P .001) Reciprocal Rank Fusion 0.313 0.208 0.230 0.249 (P=.004) 0.322 (P=.02) 0.260 (P=.01) 0.878 (P=.04) 0.807 (P=.04) Figure 3 ”
• Reciprocal rank fusion outperforms condorcet and individual rank learning methods (2009, background): “RRF is simpler and more effective than Condorcet Fuse, while sharing the valuable property that it combines ranks without regard to the arbitrary scores returned by particular ranking methods [4].”
• Personalized Web search for improving retrieval effectiveness (2004, methodology): “It has been shown in [26] that this way of merging is both effective and efficient.”
• Voting for candidates: adapting data fusion techniques for an expert search task (2006, methodology): “Other data fusion techniques could also have been considered in this work, including one based on Condorcet votingtheory [24], a technique that models score distributions [19], and a logical regression model [32].”