Abstract
Selective search is a distributed retrieval architecture that intentionally creates skewed postings and access patterns. This work shows that the well-known QtfDf inverted list caching algorithm is as effective with topically-partitioned indexes as it is with randomly-partitioned indexes. It also shows that a mixed global-local strategy reduces total I/O without harming query hit rates.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have