Indexing intervals is a fundamental problem, finding a wide range of applications, most notably in temporal and uncertain databases. We propose HINT, a novel and efficient in-memory index for range selection queries over interval collections. HINT applies a hierarchical partitioning approach, which assigns each interval to at most two partitions per level and has controlled space requirements. We reduce the information stored at each partition to the absolutely necessary by dividing the intervals in it, based on whether they begin inside or before the partition boundaries. In addition, our index includes storage optimization techniques for the effective handling of data sparsity and skewness. We show how HINT can be used to efficiently process queries based on Allen’s relationships. Experiments on real and synthetic interval sets of different characteristics show that HINT is typically one order of magnitude faster than existing interval indexing methods.

Source code from the following publications:

• George Christodoulou, Panagiotis Bouros and Nikos Mamoulis, HINT: A Hierarchical Interval Index for Allen Relationships, https://doi.org/10.1007/s00778-023-00798-w, VLDB Journal 33(1), pp. 73-100 (2024)

• George Christodoulou, Panagiotis Bouros and Nikos Mamoulis, HINT: A Hierarchical Index for Intervals in Main Memory, https://doi.org/10.1145/3514221.3517873, Proceedings of the 2022 ACM International Conference on Management of Data (ACM SIGMOD), pp. 1257-1270 (2022)

• Panagiotis Bouros
• George Christodoulou
• Nikos Mamoulis

• g++/gcc
• Boost Library

Directory samples includes the BOOKS dataset used in the experiments and a query file containing 20k queries

• AARHUS-BOOKS_2013.dat
• AARHUS-BOOKS_2013_20k.qry

Compile using make all or make <option> where <option> can be one of the following:

• lscan
• 1dgrid
• hint
• hint_m

The code supports two types of workload:

• Counting the qualifying records, or
• XOR'ing between their ids

You can switch between the two by appropriately setting the WORKLOAD_COUNT flag in def_global.h; remember to use make clean after resetting the flag.

• main_lscan.cpp

• containers/relation.h

• containers/relation.cpp

• Examples

  $ ./query_lscan.exec -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  $ ./query_lscan.exec -q gOVERLAPS -v samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

• main_1dgrid.cpp
• containers/relation.h
• containers/relation.cpp
• indices/1dgrid.h
• indices/1dgrid.cpp

• Examples

  $ ./query_1dgrid.exec -p 500 -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  $ ./query_1dgrid.exec -p 500 -q gOVERLAPS -v samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

• main_hint.cpp
• containers/relation.h
• containers/relation.cpp
• indices/hierarchicalindex.h
• indices/hierarchicalindex.cpp
• indices/hint.h
• indices/hint.cpp

• Examples

  $ ./query_hint.exec -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  $ ./query_hint.exec -o SS -q gOVERLAPS -v samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

• main_hint_m.cpp
• containers/relation.h
• containers/relation.cpp
• containers/offsets.h
• containers/offsets.cpp
• containers/offsets_templates.h
• containers/offsets_templates.cpp
• indices/hierarchicalindex.h
• indices/hierarchicalindex.cpp
• indices/hint_m.h
• indices/hint_m.cpp
• indices/hint_m_subs+sort.cpp
• indices/hint_m_subs+sopt.cpp
• indices/hint_m_subs+sort+sopt.cpp
• indices/hint_m_subs+sort+sopt+ss.cpp
• indices/hint_m_subs+sort+cm.cpp
• indices/hint_m_subs+sort+sopt+cm.cpp
• indices/hint_m_subs+sort+ss+cm.cpp
• indices/hint_m_all.cpp

• Examples

  base with top-down

  $ ./query_hint_m.exec -m 10 -t -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  base with bottom-up

  $ ./query_hint_m.exec -m 10 -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sopt (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sopt -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort+sopt (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort+sopt -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort+sopt+ss (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort+sopt+ss -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort+cm (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort+cm -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort+sopt+cm (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort+sopt+cm -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  subs+sort+ss+cm optimizations (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o subs+sort+ss+cm -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

  all optimizations (only bottom-up)

  $ ./query_hint_m.exec -m 10 -o all -q gOVERLAPS -r 10 samples/AARHUS-BOOKS_2013.dat samples/AARHUS-BOOKS_2013_20k.qry

The following are missing from the current version of the code:

• HINT with SS optimization answering the basic predicates from Allen's algebra
• Updates