setix 0.8.3

Fast data structures for finding intersecting sets and similar strings

At its core setix provides a “set intersection index”, an inverted index data structure designed for storing sets of symbols and fast querying of sets intersecting the given set, with sorting based on the number of intersections or a similarity measure.

Additionally, a wrapper for indexing strings is provided in setix.trgm, which implements a trigram index compatible with the PostgreSQL extension pg_trgm.

Examples

Using a set index:

import setix

ix = setix.SetIntersectionIndex ()
ix.add ((1, 2, 3))
ix.add ((1, 2, 4))
ix.add ((2, 3, 4))

ix.find ((1, 2), 1).get_list()
# returns [(2, [(1, 2, 3)]),
#          (2, [(1, 2, 4)]),
#          (1, [(2, 3, 4)])]
# (the order of the first two results can change as they have equal scores)

Using a trigram index:

import setix.trgm

ix = setix.trgm.TrigramIndex ()
ix.add ("strength")
ix.add ("strenght")
ix.add ("strength and honor")

ix.find ("stremgth", threshold=1).get_list()
# returns [(6, ["strength and honor"])
#          (6, ["strength"]),
#          (4, ["strenght"])]

ix.find_similar ("stremgth", threshold=0.1).get_list()
# returns [(0.5,  ["strength"]),           # 6 intersections / (9 total + 9 total - 6)
#          (0.29, ["strenght"]),           # 4 intersections / (9 total + 9 total - 4)
#          (0.27, ["strength and honor"])] # 6 intersections / (9 total + 19 total - 6)

In general, to search for phrases containing a misspelt word, a threshold of -3*N can be given where N is the number of misspellings.

ix.find ("stremgth", threshold=-3).get_list()
# returns [(6, ["strength and honor"]),
#          (6, ["strength"])]

Benchmarks

A benchmark is included in tests/dvd_db_test.py

Results from an Athlon II running at 2.6GHz:

Python 2.7

In [1]: import tests.dvd_db_test
Loading database...
Extracted 240577 titles
Memory used by data: 107.8MB
Building index...
CPU time used: 43.1s
Unique trigrams indexed: 11352
Unique phrases indexed: 228620
Memory used by index: 80.9MB

In [2]: %timeit list (tests.dvd_db_test.titles.find("daft punk", 8))
10 loops, best of 3: 27.8 ms per loop

In [3]: %timeit list (tests.dvd_db_test.titles.find("daft punk", 1))
10 loops, best of 3: 86.4 ms per loop

Python 3.2

In [1]: import tests.dvd_db_test
Loading database...
Extracted 240577 titles
Memory used by data: 108.8MB
Building index...
CPU time used: 45.8s
Unique trigrams indexed: 11352
Unique phrases indexed: 228620
Memory used by index: 86.2MB

In [2]: %timeit list (tests.dvd_db_test.titles.find("daft punk", 8))
10 loops, best of 3: 27.9 ms per loop

In [3]: %timeit list (tests.dvd_db_test.titles.find("daft punk", 1))
10 loops, best of 3: 86.3 ms per loop

DVD title list used in the benchmark was obtained from http://www.hometheaterinfo.com/dvdlist.htm Thanks for making it available.