BlockMaxWand is a recent advance on theWand dynamic pruning technique, which allows ecient retrieval without any eectiveness degradation to rank K . However, while BMW uses docid-sorted indices, it relies on recording the upper bound of the term weighting model scores for each block of postings in the inverted index. Such a requirement can be disadvantageous in situations such as when an index must be updated. In this work, we examine the appropriateness of upper-bound approximation – which have previously been shown suitable for Wand– in providing ecient retrieval for BMW. Experiments on the ClueWeb12 category B13 corpus using 5000 queries from a real search engine’s query log demonstrate that BMW still provides benets w.r.t. Wand when approximate upper bounds are used, and that, if approximations on upper bounds are tight, BMW with approximate upper bounds can provide eciency gains w.r.t.Wandwith exact upper bounds, in particular for queries of short to medium length.
[1]
Andrew Trotman,et al.
A Comparison of Document-at-a-Time and Score-at-a-Time Query Evaluation
,
2017,
WSDM.
[2]
Torsten Suel,et al.
Faster top-k document retrieval using block-max indexes
,
2011,
SIGIR.
[3]
Torsten Suel,et al.
Optimizing top-k document retrieval strategies for block-max indexes
,
2013,
WSDM.
[4]
Jeffrey Dean,et al.
Challenges in building large-scale information retrieval systems: invited talk
,
2009,
WSDM '09.
[5]
Craig MacDonald,et al.
Upper-bound approximations for dynamic pruning
,
2011,
TOIS.
[6]
Andrei Z. Broder,et al.
Efficient query evaluation using a two-level retrieval process
,
2003,
CIKM '03.
[7]
J. Shane Culpepper,et al.
Exploring the magic of WAND
,
2013,
ADCS.
[8]
Tao Tao,et al.
A formal study of information retrieval heuristics
,
2004,
SIGIR '04.
[9]
Sebastiano Vigna,et al.
Quasi-succinct indices
,
2012,
WSDM.