MIDSI Project

Source code of the Midsi inference engine with case study and benchmark

Solution for data discovery in projects applicable to the Semantic Web, enabling the loading of ontologies and inference of results using the WSML language.

SableCC is used for translators for the WSML language.

Prolog is used as an engine to perform inferences.

Pyswip is used for connection between python and prolog resources.

Requirements:

• SwiProlog
• Python 3.*
  • pyswip

Installing project dependencies

Linux Debian-based systems:

sudo ./install.sh

Other linux systems install the following packages:

default_jre
python3
swi-prolog #swi-prolog ou pl
pip pyswip

MacOS (Intel or M1):

brew install swi-prolog
pip3 install pyswip

Benchmark

• To run and view the results:

  cd benchmark
  python3 benchmark.py
  

• Additional settings to the benchmark can be changed in benchmark.conf.

• A log folder will be created containing the result of the benchmark.

Using Midsi as a service

Just enter the ./service folder and start the midsi-service.py server. The configuration of ip and port is in the file config.py, to use another address just change it.

cd ./service
python3 midsi-service.py

To make it easier to use, the commands we can use on our server were centralized in midsi-client.py.

• Loading an ontology:

  python3 midsi-client.py --ontology ../wsmlcodes/ont1-ShipmentOntology/ShipmentOntology.wsml
  

• Running a query:

  python3 midsi-client.py --query "cityIsOnContinent(?x,?y)"
  

• Executing a query from a file:

  python --queryFile ../wsmlcodes/ont1-ShipmentOntology/query1-allPackageStatus.wsml
  

• Closing the server (can be done by closing midsi-service.py directly):

  python3 midsi-client.py --exit
  

WSML to VSCode

We created an extension for VSCode that provides syntax highlighting and bracket matching for WSML files. It also has integrated controls for the use of the Midsi client / server previously exposed.

It can be downloaded directly from the VSCode marktplace, searching for WSML. See more at https://github.com/devdaniellima/wsml-vscode.