BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Interval Joins: Evaluation\, Parallelization\, Aggregation
DTSTART:20181126T140000
DTEND:20181126T150000
DTSTAMP:20260407T111851Z
UID:07b70f145b205cede653de55bbedfb360e964b9da416182ecfac138f
CATEGORIES:Conferences - Seminars
DESCRIPTION:Nikos Mamoulis joined the University of Ioannina in 2014. Befo
 re that he was a Professor of Computer Science at the University of Hong K
 ong. His main research interest is the management\, search and analysis of
  complex data types with a focus on spatially enriched data. He served as 
 PC co-chair of SSTD 2009\, COMAD 2013\, and HDMS 2014 and as a general cha
 ir of SSDBM 2008. He has been an Associate Editor of the VLDB Journal\, IE
 EE Transactions on Knowledge and Data Engineering\, Knowledge and Informat
 ion Systems\, and an Editorial Board Member of Geoinformatica Journal.\nAb
 stract:\nThe interval join is a basic operation that finds application in 
 temporal\, spatial\, and uncertain databases. In this talk\, I will presen
 t our recent research findings on the evaluation of interval joins. In our
  work\, we explore the applicability of a largely ignored forward scan (FS
 ) based plane sweep algorithm\, which is extremely simple to implement. We
  propose two optimizations of FS that greatly reduce its cost\, making it 
 competitive to the state-of-the-art single-threaded interval join algorith
 m while achieving a lower memory footprint. In addition\, for the parallel
  processing of interval joins\, we study a domain-based partitioning appro
 ach that does not produce duplicate results. Within our approach we propos
 e a novel breakdown of the partition-level join into a small number of ind
 ependent mini-join jobs with varying cost and manage to avoid redundant co
 mparisons. We show how these mini-joins can be scheduled in multiple CPU c
 ores and suggest an adaptive domain partitioning\, aiming at load balancin
 g. Finally\, we study the evaluation of a temporal aggregation operation\,
  which can be used for selecting or ranking intervals based on the number 
 of join pairs they appear in. For this problem\, we present an algorithm t
 hat generates the result at the cost of only scanning the sorted interval 
 endpoints.\n\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR