Automatic combination of multiple ranked retrieval systems

By B. Bartell, Garrison W. Cottrell, Richard K. Belew. 1994

Subfields: Information Retrieval & Search

How researchers see this paper

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

In Information Retrieval & Search

Citations in the corpus

Pretrained Transformers for Text Ranking: BERT and Beyond (2020, background): “With respect to the second question, one obvious approach to combining evidence from initial BM25 retrieval scores and BERT is linear interpolation, whose usage in document ranking dates back to at least the 1990s [Bartell et al., 1994]:”
Condorcet fusion for improved retrieval (2002, methodology): “…model; Fox, Shaw et al. [12, 13, 27] who developed the CombSum and CombMNZ algorithms; Lee [18, 19] who performed more experiments with these Comb algorithms; and Bartell [3, 2] and Vogt [34, 33, 31, 32] who experiment with linearly combining the normalized relevance scores given to each document.”
Learning to Order Things (1997, methodology): “For instance, Boyan, Freitag, and Joachims [6] use simulated annealing to improve agreement with \click data," and Bartell, Cottrell and Belew [2] use conjugate gradient descent to choose parameters for a linear combination of scoring functions, each associated with a di erent search expert.”
Exploiting geographical influence for collaborative point-of-interest recommendation (2011, methodology): “We propose to use a linear fusion framework to integrate ranked lists provided by the three above-mentioned recommenders into the final ranked list [4, 21].”
A support vector method for optimizing average precision (2007, methodology): “3.2 Finding the Most Violated Constraint Using OP 1 and optimizing to ROCArea loss (.roc ), the documents ranked before it, as opposed to j -1.”
Optimizing search engines using clickthrough data (2002, unknown): “[2].”
Topic Detection and Tracking (2002, methodology): “The National Institute of Standards and Technology (NIST) has administered three open evaluations of the TOT tasks since 1998 [2,3,9].”
Query chains: learning to rank from implicit feedback (2005, methodology): “This approach of learning a combination of functions is also used by most other work in this area [1, 2, 4, 15, 21].”
Building efficient and effective metasearch engines (2002, background): “A number of functions have been proposed to combine individual ranking scores of the same document, including min, max, average, sum, weighted average, and other linear combination functions [Cottrell and Belew 1994; Fox and Shaw 1994; Lee 1997; Vogt and Cottrell 1999].”