Experiments in Graph Structure and Knowledge Graph Embeddings

Tracking #: 861-1868

Flag : New Paper Submitted

Authors: 

Antrea Christou
Brandon Dave
Cogan Shimizu

Submission Type: 

Article in Special Issue (note in cover letter)

Full PDF Version: 

Supplementary Files: 

Cover Letter: 

Dear Editors and Reviewers, Thank you for your thoughtful comments in response to our submission, "Experiments in Graph Structure and Knowledge Graph Embeddings" (Tracking Number: 806-1797), submitted as an Article in Special Issue to Neurosymbolic Artificial Intelligence. We appreciate the acknowledgment of the motivation behind our work and the interest in the new synthetic data assets we provide to enhance our understanding of knowledge graph link prediction. We have taken each comment or remark into consideration and provided a tabulated response in the following pages. Briefly, we have: -- Clarified the evaluation mismatch regarding relations P31 and P279 by discussing the limitations of the KGE methods used and the rationale behind our evaluation choices; -- Expanded the description of the synthetic knowledge graphs, especially SKG-237, to explain their properties and the aspects they replicate; and -- Provided a detailed explanation of our methodology for creating dataset splits for the knowledge graph embeddings, including motivations and procedural details. Changes in the manuscript have been highlighted for your convenience: deletions appear in red and are struck out, additions appear in blue, replacements appear in green (and are preceded by a deletion). We hope that these changes are sufficient and that the paper can be accepted in this form. We are poised to rapidly submit our camera ready version in that case. Sincerely, Cogan Shimizu On behalf of all authors

Previous Version: