graph-compress 1.1.0

Library designed to compress graphs

graphcompress

graphcompress is a Python library designed to compress graphs from NTRIPLES.

Formats

Compressed Graph

The resultant format writes the info of a node in a line. First the NODE ID, next, groups of connections, every connection has the EDGE ID and NODES CONNECTED USING THE EDGE. Considers a negative EDGE ID as a connection in opposite direction.

Example:

14 5.3.7 -3.2

The line says:

• Node 14 connects to Node 3 using Edge 5.
• Node 14 connects to Node 7 using Edge 5.
• Node 2 connects to Node 14 using Edge 3.

Indexed Graph

The resultant format writes the info of a node in a line. First the NODE ID, next, indexes of edges where the node participates

Example:

14 3 5 7

The line says:

• Node 14 appears in the edge with the index 3.
• Node 14 appears in the edge with the index 5.
• Node 14 appears in the edge with the index 7.

Install

pip install graph-compress

Usage

Input file: graph.nt.gz

<http://www.wikidata.org/entity/Q1> <http://www.wikidata.org/prop/direct/P5> <http://www.wikidata.org/entity/Q5> .
<http://www.wikidata.org/entity/Q1> <http://www.wikidata.org/prop/direct/P3> <http://www.wikidata.org/entity/Q6> .
<http://www.wikidata.org/entity/Q6> <http://www.wikidata.org/prop/direct/P5> <http://www.wikidata.org/entity/Q5> .
<http://www.wikidata.org/entity/Q6> <http://www.wikidata.org/prop/direct/P5> <http://www.wikidata.org/entity/Q2> .

(The lines above are the uncompressed file)

Compressed Graph

from graphcompress.readers import Reader_NT, Reader_NTgz, Reader_CGgz
from graphcompress.parsers import WikidataParser, CGParser
from graphcompress.builders.compress_graph import CGBuilder

r = Reader_NTgz('graph.nt.gz')
p = WikidataParser()
b = CGBuilder(r,p,'tmp','output.gz')

b.make_partitions()
b.merge_partitions()

Output file: output.gz

1 5.5 3.6
2 -5.6
5 -5.1.6
6 5.5.2 -3.1

(The lines above are the uncompressed file)

Indexed Graph

from graphcompress.readers import Reader_NT, Reader_NTgz, Reader_CGgz
from graphcompress.parsers import WikidataParser, CGParser
from graphcompress.builders.index_graph import IGBuilder

r = Reader_NTgz('graph.nt.gz')
p = WikidataParser()
b = IGBuilder(r,p,'tmp','output.gz')

b.make_partitions()
b.merge_partitions()

Output file: output.gz

1 0 1
2 3
5 0 2
6 1 2 3

(The lines above are the uncompressed file)