English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Learned indexes, which use machine learning models to replace traditionalindex structures, have shown promising results in recent studies. However,existing learned indexes exhibit a performance gap between synthetic andreal-world datasets, making them far from practical indexes.  In this paper, we identify that ignoring the importance of data partitioningduring model training is the main reason for this problem. Thus, we explicitlyapply data partitioning to index construction and propose a new efficient andupdatable cache-aware RMI framework, called CARMI. Specifically, we introduceentropy as a metric to quantify and characterize the effectiveness of datapartitioning of tree nodes in learned indexes and propose a novel cost model,laying a new theoretical foundation for future research. Then, based on ournovel cost model, CARMI can automatically determine tree structures and modeltypes under various datasets and workloads by a hybrid construction algorithmwithout any manual tuning. Furthermore, since memory accesses limit theperformance of RMIs, a new cache-aware design is also applied in CARMI, whichmakes full use of the characteristics of the CPU cache to effectively reducethe number of memory accesses. Our experimental study shows that CARMI performsbetter than baselines, achieving an average of 2.2x/1.9x speedup compared to B+Tree/ALEX, while using only about 0.77x memory space of B+ Tree. On the SOSDplatform, CARMI outperforms all baselines, with an average speedup of 1.2x overthe nearest competitor RMI, which has been carefully tuned for each dataset inadvance.

CARMI: A Cache-Aware Learned Index with a Cost-based Construction Algorithm