Hi KG community, I am Daniel Vila, Director of Recognai, a startup working on NLP and KGs. we are happy to share with you Rubrix (https://www.rubrix.ml/), an open-source human-in-the loop tool for iterating on data for AI, which can be used for node classification in KGs and more generally KG curation. The first time we unveiled the tool was during this past KG Conference, during the Graph-based Data Science workshop. https://towardsdatascience.com/bringing-more-humans-into-the-ai-lifecycle-ac5bed139314
thank you Daniel, welcome! what types of use cases might Rubrix be useful for?
Thank Ellie. There are several use cases related to knowledge graphs. One area would be related to knowledge graph construction and completion, where you have some sort of model or process for adding new entities, types or triples to a knowledge graph. In this area, Rubrix can be used for creating training data (in case you are using an ML approach) or curating the data to be added to the knowledge graph. During KGC we gave a tutorial related to this area, using kglab and Pytorch-geometric for adding types to an existing knowledge graph, here's the full tutorial and the use case description: https://docs.rubrix.ml/en/stable/tutorials/03-kglab_pytorch_geometric.html#Our-use-case-in-a-nutshell
Another area would be the intersection between Knowledge graphs and NLP, where you can use Rubrix for iterating on data for models such as relation classification, or named entity recognition and linking. All this, with a focus on involving subject matter experts, not only engineers or data scientists but also , which we believe is a key feature in the knowledge graph space.