Streaming Partitioning of RDF Graphs for Datalog Reasoning

Paper

Click here to download the full version of the paper, which includes the proofs.

For any issues, please contact temitope.ajileye@cs.ox.ac.uk

Download settings.txt and edited following the comments in the file
Create a hosts.txt file in the same directory, with a line for every worker server, and an additional line for the coordinating server. In each line put the ip address or hostname of the server and a listening port, separated by white space
Download testgen.py into the same directory
Run 'python testgen.py'. The script will create a working folder for each server; the coordinating server folder is the last, numerically. The script will also create the auxillary shell script transfer.sh, used to copy the folders to the cluster.
Copy the DMAT executable into each folder
Copy the program file (.dlog) into the coordinating server folder. Furthermore, if it is a streaming test, copy the dataset file into the coordinating server, otherwise copy the pre-partitoned dataset files into each worker server folder.
Now copy each folder to the appropriate server. To speed this up, you can use transfer.sh, after checking that the assumptions made are suitable for your testing environment, and editing as appropriate.
Connect to each server. If in a linux environment, cd into the server folder and run ./test1. Otherwise edit test1 to suit your Windows environment