RDF generator

The goal of this repository is to provide an easy way of generating triples from JSON/CSV sources and extend an extisting RDF triplestore. The input must be a CSV/JSON file, as well as a pickle file for the mapping of the variables to find in the CDV/JSON file, and an ontological scheme of tose variables, in RDF.

Installation

You will need to install RDFlib in order to use this code.

pip install rdflib
git clone https://github.com/klotzbenjamin/rdf-generator.git
cd rdf-generator

Usage

General example

This repository contains examples of input files, graph, mappings and a scheme.

'python rdf-generator.py -i input/input-flat.json -m mapping/mapping-flat.pkl -s scheme/SOSA-scheme.ttl'

This order will use data from input/input-flat.json to feed a scheme scheme/SOSA-scheme.ttl, mapped with mapping/mapping-flat.pkl.

Arguments

• -i: Input raw data. Currently it must be a CSV or JSON file.
• -m: Mapping file, serialized in pickle. This contains a python dictionary mapping the variables present in the scheme with a path to access the values in the raw data file
• -s: Ontological scheme file, serialized in turtle, containing triples of variables (written :var1, :var2...) and static constants (written: cst1, :cst2...)
• -g: (Optional) Input graph to extend. Currently it must be serialized in turtle.
• -o: (Optional) Output file to create. It will be serialized in turtle.

Available examples

This repository contains 3 input files:

• input-flat.csv
• input-flat.json
• input-complex.json

They all contain the same information, an observation of car sensors.

We use here the VSSo ontology to model car signals, and the SOSA pattern. The Scheme will be a SOSA observation and its neighbours regardless of the format of the raw data.

We need a mapping between raw data and the variable in the scheme. We have 2 mappings here:

• mapping-flat.pkl: for both input-flat files
• mapping-complex.pkl: for input-complex.json