MOSAIC: Multi-strategy Ontology Alignment and Integration Suite at CSG

A system created as a combination of matching strategies developed by BSc and MSc students taking the module on Semantic Web Technologies and Knowledge Graphs at City St George's, University of London.

-- instructions --

dependencies

for windows NVIDIA users, download the repository and then create a virtual environment. Then use the terminal/CMD in the root folder to run pip install [paste the list in the requirements.txt file ignoring the commented out lines]

(or type pip install -r requirements.txt if that works for you)

for AMD users, use the amd_installation.md

formatting

tracks in DH/Bio-ML tested in rdf format - tracks with duplicate ontologies like pactols need to be named in the format pactols1 pactols2 etc.

reference files can be in .tsv format or .ttl

rename the source and target rdf files to their exact name: eg source.rdf -> idai.rdf used for the reference.rdf file idai.pactols1.rdf

the reference file should be the exact names of the 2 ontologies combined with a "-" in between: eg the format defc-pactols1.rdf or idai-pactols2.rdf

the rdf format was used because the track has different versions of the same ontology per benchmark

eval knowledge-graph track

use kg_eval.py after running amdCode.py to get the correct results saved to kg_eval_report.csv. mosaic_report.csv is the main csv file for the other scores.