-
Notifications
You must be signed in to change notification settings - Fork 14
Graphalytics
The following table contains the DATAGEN releases and their changes related to Graphalytics.
| release | date | changes |
|---|---|---|
| 0.2.1 | 2015-04-20 | scale factors for Graphalytics |
| 0.2.2 | 2015-10-27 |
graphalytics.CSVPersonSerializerWithWeights serializer |
| 0.2.3 | 2015-10-27 | new scale factors for Graphalytics and readjusted existing ones |
| 0.2.4 | 2015-11-02 | tunable clustering coefficients |
| 0.2.5 | 2015-11-03 | |
| 0.2.6 | 2016-06-20 |
graphalytics.CSVPersonSerializerExtended serializer |
| 0.2.7b | 2017-07-11 | gscale-based generation (initial implementation) |
| 0.2.7 | 2017-10-17 | gscale-based generation (stable implementation) |
| 0.2.8 | 2018-01-18 |
In the following table, we list the number of vertices/edges for Graphalytics graphs of scale factors 30, 100, 300, adhering to a "Facebook distribution". These graphs can be automatically generated with the graphalytics-generate-old.sh script.
| release | D30/V | D30/E | D100/V | D100/E | D300/V | D300/E |
|---|---|---|---|---|---|---|
| 0.2.1 | 650,000 | 30,845,741 | 1,800,000 | 99,689,585 | 4,800,000 | 299,646,264 |
| 0.2.2 | 592,500 | 29,970,158 | 1,670,000 | 100,291,564 | 4,350,000 | 299,446,981 |
| 0.2.3 | 592,500 | 29,970,158 | 1,670,000 | 100,291,564 | 4,350,000 | 299,446,981 |
| 0.2.4 | 592,500 | 30,361,660 | 1,670,000 | 101,716,179 | 4,350,000 | 304,036,710 |
| 0.2.5 | 592,500 | 30,361,660 | 1,670,000 | 101,716,179 | 4,350,000 | 304,036,710 |
| 0.2.6 | 592,500 | 30,338,404 | 1,670,000 | 101,747,778 | 4,350,000 | 304,035,394 |
| 0.2.7b | 593,750 | 30,374,544 | 1,562,500 | 94,415,527 | 4,281,250 | 299,248,729 |
| 0.2.7 | 593,750 | 30,402,030 | 1,562,500 | 94,488,768 | 4,281,250 | 299,298,877 |
| 0.2.8 | 593,750 | 30,402,030 | 1,562,500 | 94,488,768 | 4,281,250 | 299,298,877 |
Additionally, between versions 0.2.6 and 0.2.7, at commits e0a9f63 and later 02c322d the code contained a bug that caused it to generate an incorrect distribution (the change was introduced with the "gscale" option at commit c3da811, but is executable at the aforementioned commits).
An example for this bug is that the Graphalytics D30 dataset had 593,750 vertices (the same value as all versions from 0.2.7), but with 30,374,544 edges instead of 30,402,030. This issue was fixed in e54c8b6.
Scripts to convert the output files to Graphalytics-compatible vertex and edge files are available in the graph-format-conversion directory.