The article introduces PyGraft, a Python-based tool for generating highly customized, domain-agnostic schemas and knowledge graphs. It addresses the limitations of relying on a limited collection of datasets for model evaluation and proposes a solution to generate diverse datasets for benchmarking. PyGraft ensures logical consistency by utilizing a DL reasoner and provides a way to generate both schema and KG in a single pipeline. The article details the schema and KG generation processes, highlighting the importance of schema-driven generators and the need for more diverse benchmark datasets. It also discusses related work, efficiency, scalability, usage illustration, potential uses, limitations, sustainability, maintenance, and future work.
Til et annet språk
fra kildeinnhold
arxiv.org
Viktige innsikter hentet fra
by Nicolas Hube... klokken arxiv.org 03-07-2024
https://arxiv.org/pdf/2309.03685.pdfDypere Spørsmål