7 Hits in 1.4 sec

From MAXSCORE to Block-Max Wand: The Story of How Lucene Significantly Improved Query Evaluation Performance [chapter]

Adrien Grand, Robert Muir, Jim Ferenczi, Jimmy Lin
2020 Lecture Notes in Computer Science  
The latest major release of Lucene (version 8) in March 2019 incorporates block-max indexes and exploits the block-max variant of Wand for query evaluation, which are innovations that originated from academia  ...  thank Craig Macdonald, Joel Mackenzie, Antonio Mallia, and Nicola Tonellotto for helpful discussions on the intricacies of computing flexible per-block score bounds, and Torsten Suel for providing us with  ...  on Lucene 8.0 and uses blockmax Wand.  ... 
doi:10.1007/978-3-030-45442-5_3 fatcat:blgpxox6lbekbgcmvbfhs5z56e

Score-safe term-dependency processing with hybrid indexes

Matthias Petri, Alistair Moffat, J. Shane Culpepper
2014 Proceedings of the 37th international ACM SIGIR conference on Research & development in information retrieval - SIGIR '14  
Score-safe methods include MaxScore [17] , WAND [1] , and BlockMax WAND (BMW) [4, 5] .  ...  The sequence of improvements arising from score-safe heuristics -WAND and BlockMax WAND (BMW) -is clear.  ... 
doi:10.1145/2600428.2609469 dblp:conf/sigir/PetriMC14 fatcat:5gwuvn4xibh7bhzffk7omdczra

Faster top-k document retrieval using block-max indexes

Shuai Ding, Torsten Suel
2011 Proceedings of the 34th international ACM SIGIR conference on Research and development in Information - SIGIR '11  
We show how to integrate this structure into the WAND approach, leading to considerable performance gains.  ...  An important class of optimization techniques called early termination achieves faster query processing by avoiding the scoring of documents that are unlikely to be in the top results.  ...  Thus, WAND achieves early termination by enabling skips over postings that cannot make into the top results.  ... 
doi:10.1145/2009916.2010048 dblp:conf/sigir/DingS11 fatcat:enwbz3c7gjdjredqexnfwwzfgi

Does Selective Search Benefit from WAND Optimization? [chapter]

Yubin Kim, Jamie Callan, J. Shane Culpepper, Alistair Moffat
2016 Lecture Notes in Computer Science  
We also compare two methods for applying WAND to topical shards: searching each shard with a separate top-k heap and threshold; and sequentially passing a shared top-k heap and threshold from one shard  ...  We demonstrate that the WAND dynamic pruning algorithm is more effective on topical index shards than it is on randomly-organized index shards, and that the savings generated by selective search and WAND  ...  However, if the scoring regime is derived from a language model, then the distribution of scores is top-heavy, and BlockMax WAND should be used. Rojas et al.  ... 
doi:10.1007/978-3-319-30671-1_11 fatcat:jyjywxp2sfgczepq77pcjlxile

Efficient Query Processing for Scalable Web Search

Nicola Tonellotto, Craig Macdonald, Iadh Ounis
2018 Foundations and Trends in Information Retrieval  
Our explanations of query processing strategies, for instance the WAND and BMW dynamic pruning algorithms, are presented with illustrative figures showing how the processing state changes as the algorithms  ...  Meanwhile, search engines continue to rapidly evolve, with larger indexes, more complex retrieval strategies and growing query volumes.  ...  In particular, Kim et al. (2015) found that they could schedule queries with longer predicted execution times onto faster, more expensive CPU cores.  ... 
doi:10.1561/1500000057 fatcat:wx53qhvfhnfwfc4hgdva5ypw3u

Skewed partial bitvectors for list intersection

Andrew Kane, Frank Wm. Tompa
2014 Proceedings of the 37th international ACM SIGIR conference on Research & development in information retrieval - SIGIR '14  
This approach allows more granular calculations of maximum score, but taking into account the block sizes makes the blockmax WAND algorithm more complicated.  ...  Our skips implementation, however, is significantly faster than compressed intersection for any number of terms per query, and we find that combining skips with bitvectors is faster than bitvectors alone  ... 
doi:10.1145/2600428.2609609 dblp:conf/sigir/KaneT14 fatcat:fuwxdwf7x5cd7mie277owhkdt4

In Vacuo and In Situ Evaluation of SIMD Codecs

Andrew Trotman, Jimmy Lin
2016 Proceedings of the 21st Australasian Document Computing Symposium on ZZZ - ADCS '16  
For a document-ordered index document-at-a-time (DaaT) techniques such as WAND [4] and BlockMax [6] are typically used.  ...  With respect to efficiency, the PFor-based codecs take longer to decode than the others, which appear to behave similarly.  ... 
doi:10.1145/3015022.3015023 fatcat:2syrr7rj4fghdh5c5tmxby7cry