I40KG-Embeddings

by Author 1, Author 2, etc

This is a template for papers that use Python code and Jupyter notebooks to generate their results (though it can be adapted to use other technologies). The text is written in LaTex and tasks are automated using Makefiles. Ideally, all results, figures and the final paper PDF should be generated by running a single make command in the root of this repository.

Fill out the sections below with the information for your paper.

This paper has been submitted for publication in Some Journal.

Brief description of what this paper is about (2-3 sentences). Include a figure as well with the main result of your paper.

Caption for the example figure with the main results.

Abstract

Industry 4.0 (I4.0) standards and standardization frameworks have been proposed with the goal of empowering interoperability in smart factories. These standards enable the description and interaction of the main components, systems, and processes inside of a smart factory. Due to the increasing number of I4.0 frameworks and standards, there is an increasing need in the industry for approaches that automatically analyze the landscape of I4.0 standards. The fact that standardization frameworks classify standards according to their functions into layers and dimensions produces interoperability conflicts between standards. Describing the I4.0 landscape in knowledge graphs alleviates these interoperability problems. However, performing traditional graph analysis still requires significant domain knowledge. In this work, we study the relatedness among standards and frameworks based on community analysis to discover knowledge that helps us to cope with interoperability conflicts between standards. We use knowledge graph embeddings to create these communities automatically exploiting the meaning of the relationships among standards. In particular, we focus on the identification of groups of standards that are similar, i.e., communities of standards, and analyze their properties to uncover relations between them. We empirically evaluate our approach on a knowledge graph of I4.0 standards using the Trans family of embedding models for knowledge graph entities. Our results indicate the benefits of a knowledge-driven approach for unveiling novel properties of the I4.0 standards landscape.

Software implementation

Briefly describe the software that was written to produce the results of this paper.

The data used in this study are provided in embeddings and test_set. Results generated by the code are saved in accuracy, density_plot and evaluation_metric. See the README.md files in each directory for a full description.

Getting the code

You can download a copy of all the files in this repository by cloning the git repository:

git clone https://github.com/i40-Tools/I40KG-Embeddings

or download a zip archive.

A copy of the repository is also archived at insert DOI here

Dependencies

You'll need a working Python environment to run the code. The required dependencies are specified in the file requirements.txt. The libraries indicated in the above file must be installed using the pip or the package manager of the operating system.

Content

• experiment.py: This script load the input data and create the communities. 5-fold cross validation in the experiment is used. The input required are the embeddings training set. The output of each training set is in the folder clustering_measures. We combine three community detection algorithms with four embedding algorithms.

• density_plot_standard_similarity.py: Plot the probability density for each five-fold and embedding algorithm.

• clustering_measures/execute_metrics.py: Clustering Measures App (CMA). This script call CMA App, This program computes the clusterings measures for the communities obtained. See the Readme.txt.

• evaluation_metric.py: Plot the quality of the generated communities. A radar plot is computed with 5 metrics for the result of three community detection algorithms. For each embedding algorithms a radar plot is computed.

• community_analysis.py : Plot the quality of the generated communities. Percentage of the test set for the property \texttt{sto:relatedTo} is achieved in each cluster.

Reproducing the results

Running an example:

python3 experiment.py

python3 density_plot_standard_similarity.py

python3 clustering_measures/execute_metrics.py

python3 evaluation_metric.py

python3 community_analysis.py

License

All source code is made available under a BSD 3-clause license. You can freely use and modify the code, without warranty, so long as you provide attribution to the authors. See LICENSE.md for the full license text.

The manuscript text is not open source. The authors reserve the rights to the article content, which is currently submitted for publication in the JOURNAL NAME.