OSIP-KB-ICE: System Engineering Models Meet Knowledge Graphs

The aim of this repository is to collect the source code of the activity "System Engineering Models meet Knowledge Graphs", OSIP ESA Contract No. 4000133311/20/NL/GLC, Strathclyde RKES REF 200980. The expected deliverables are:

• SW1: Schema of the Vaticle Knowledge Graph (KG), migrated from the ECSS-E-TM-10-25A Annex A UML model and enriched with metadata.
• SW2: Pipelines to automatically populate a Vaticle Knowledge Graph with Engineering Models based on the ECSS-E-TM-10-25A TM and rules to infer implicit knowledge, including a similarity analysis.

SW1: Generate a TypeDB Schema written in TypeQL from the ECSS-E-TM-10-25A Annex A UML model

The SW1 includes:

in schema_generation:

• xmi2typeql.py: the functions to map the UML classes, properties, relations to TypeQL entities, attributes and relations
• xmi2typeql_data_v2_0_1.py: the template to write TypeQL types called by xmi2typeql.py, compatible with TypeDB v2.0.1 up to v2.10.1
• TypeDBSchemaECSS.tql: the TypeDB schema compatible with TypeDB v2.10.1, generated with xmi2typeql.pyand xmi2typeql_data_v2_0_1.py
• defineMetadata.tql: an additional schema definition to enrich the KG with metadata

SW2: Population of TypeDB Graph and Inference

The SW2 includes:

in /knowledge_graph:

• migrate_em_json.py: Script to (i) automatically create a new TypeDB database, (ii) define the ECSS-based TypeDB schema and additional rules, (iii) populate it with the .json files of Engineering Models, and (iv) insert metadata.
• migrationTemplate.py: Templates to generate insert queries for each type of entities.
• similarityMetadata.py: Script to (i) extract metadata from TypeDB, (ii) compute the similarity factors between each iterations, and (iii) insert new relations with the factor values as attributes.
• similarityElements.py: Script to extract Elements and calculate Jaccard similarity between their containing parameters + cosine similarity between their names
• main.py : Main running all scripts all necessary steps

in recommendation

• Notebook_recommendation.ipynb: Notebook for running the similarity analysis with a new mission

For SW2 respective docker-files are available in /knowledge_graph to run the KG population automatically, as well as in /recommendation to provide a Jupyterlab-image to run the analysis. In order to run SW2, it is not necessary to run SW1 first. The modified output of SW1 is provided in SW2. Therefore, it is recommended to skip the steps of SW1 and start with the KG population in /knowledge_graph directly.

Pre installation steps for SW2:

• install docker-desktop on machine (tested on Docker Engine v20.10.13, Docker Compose version v2.3.3)
• git clone https://gitlab.esa.int/Luis.Mansilla/osip-kb-ice.git
• insert Engineering model (EM) files into app/datasets/
• insert metadata files into app/Metadata/

Licensing information

Information about the licenses of the used Python packages for SW2 can be found in the file package_licenses.txt.

Team

ESA T.O.: Serge Valera ESA Support: Luis Mansilla,Audrey Berquand, Alberto Gonzalez Fernandez
Contractor: University of Strathclyde
Contractor team: Paul Darm, Annalisa Riccardi, Audrey Berquand, Edmondo Minisci

Contact

[email protected] [email protected]