BCube Semantics

The BCube Semantics project contains ontologies to describe web services and datasets found by the BCube Crawler. We try to reuse already existing ontologies and implement the remaining classes and attributes using the nsidc purl namespace. This project includes a python tool that will use these ontologies to generate triples to characterize these web services and datasets.

Overview

BCube Semantics works on top of the BCube crawler, the following stack represents how the information flows.

1. BCube Crawler Based on an initial set of url seeds the BCube crawler indexes documents that could potentially be geo-science web services and datasets.

2. Semantics The current project will characterize these indexed documents using formal ontologies being the input urls and the outputs triples in turtle, n3 or rdf formats.

3. Triplestores The final layer will perform meaningful semantic queries about the services found by the crawler.

Ontologies

Web Services Ontology The Web Services Ontology describes web services and the relationship between services and the data they serve.

Visualizing and editing the ontologies

The ontologies were created using Standford's Protege software and are represented in OWL.

BCube triple generator

btriple will transform web services and dataset documents into triple files. This tool uses rdflib to create the triples and bootstrap ontologies from the ontologies/ directory.

Installation

pip install -r requirements.txt

This may require sudo permissions if there is no VirutalENV in use.

Running the tests

nosetests

Usage

Once completed we intend to expose a python module to create the triples. For now the script creates triples and dumps them into a triple store or stdout.

python btriple.py -p PATH -f [n3|turtle|rdf] -s SPARQL_ENDPOINT

If the -s parameter is not provided the output will go to stdout. We can of course do something like

python btriple.py -f turtle -p ./opensearch/ > os.ttl

This will parse all the json files found in ./opensearch and will create the triples in turtle format, we are just piping the output to a ttl file. Also if there is no -f parameter the default format is turtle.

ttl file(s) can be uploaded manually to any SPARQL endpoint.

We can also send the triples directly to a SPARQL store with the -s parameter.

python btriple.py -p ../service_examples/ogc -s http://54.69.87.196:8080/parliament/sparql

The triples are sent to the default graph(for Parliament) urn:x-arq:DefaultGraph. This is for query simplicity but it will become a parameter as well.

TODO

• Modularize btriple and upload it to pip
• Create a use cases readme file to further expand on how to use the ontologies and triples.

License GPL v3