Home

Awesome

FedBench Experiments

FedBench experiments for SemaGrow stack, FedX and SPLENDID. To run the experiment, follow the instructions below.

Datasets

In order to run the experiments you will need the following datasets. For each dataset you could use either the public endpoints, or you can set up your own (by using the pre populated virtuoso databases)

DatasetEndpoint
ChEBIhttp://143.233.226.25:8890/sparql
Dbpediahttp://143.233.226.25:8891/sparql
DrugBankhttp://143.233.226.25:8892/sparql
GeoNameshttp://143.233.226.25:8893/sparql
jamendohttp://143.233.226.25:8894/sparql
KEGGhttp://143.233.226.25:8895/sparql
LinkedMDBhttp://143.233.226.25:8896/sparql
NYThttp://143.233.226.25:8897/sparql
SP2Benchhttp://143.233.226.25:8898/sparql
SWDFhttp://143.233.226.25:8899/sparql

Configuration

Our current configuration uses the live endpoints, so if you want to use our public datasets skip this part.

If you want to use your own endpoints, you should rename the endpoints inside the configuration/metadata files, i.e.

Running the Experiment

In order to run one experiment you should execute:

sh runEval.sh ./suites/semagrow-reactive/lifeScience/lifeScience-config.prop

The ".prop" files are the entry points for the configuration for each experiment.

In order to switch between the three versions of SemaGrow you have to point the semagrow symbolic link to the corresponding semagrow directory.

Viewing the results

In order to view the results (total query processing time and number of results for each query), you should open the result/result.csv file in the directory of the .prop file. In order to view the optimization time you should check the .log files or the prints during the evaluation of the experiment.