In this directory you can find samples of all the scenarios currently supported by KROWN's data generator.
- raw: scaling in number of rows, columns and cell size of the data.
- mappings: scaling in the number of Triples Maps, Predicate Object Maps, and Graph Maps.
- duplicates-empty: scaling the number of duplicates and empty values in the data.
- joins: scaling the number of join conditions, the data involved in joins, join relations, and duplicates generated through joins
You can generate these samples yourself with KROWN's data generator, the
scenario configuration is provided as sample-$SCENARIO-rmlmapper.json
within the directory of each sample. In the samples we use RMLMapper as
materialization engine, but any engine can be used by modifying the scenario
configuration file.
For example:
cd ../data-generator
./exgentool --root=../samples/raw --scenario=sample-raw-rmlmapper.json generate
Generates all the scenarios for scaling the raw data among number of rows, columns and cell size with RMLMapper as engine.
Executing can be done with the execution framework:
cd ../execution-framework
./exectool --root=../samples/raw --runs=1 run
This will invoke the RMLMapper as a Docker container to execute all sample scenarios for the raw data parameters.
Each sample directory is structured as followed:
RMLMapper/csv/raw_10_2_0
├── data
│ ├── rmlmapper
│ │ └── shared
│ └── shared
│ ├── data.csv
│ └── mapping.rml.ttl
├── metadata.json
└── results
└── run_1
├── case-info.txt
├── log.txt
├── metrics.csv
└── rmlmapper
└── out.nt
- The RML mapping is stored at
$SAMPLE/data/shared/mapping.rml.ttl - The input data is available at
$SAMPLE/data/shared/data.csv - The output of the RMLMapper at
$SAMPLE/results/run_1/rmlmapper/out.nt