Infrastructure Tools for the Proto-OKN: Lessons from Extracting Scientific OKNs

Wednesday, August 21st, 2024
Dr. Doug Downey
Allen Institute for AI, Northwestern University

Proto-OKN PIs, Co-PIs, and collaborators were invited to join this virtual webinar on scientific OKN infrastructure on August 21, 2024. Dr. Doug Downey is a researcher at the Allen Institute for AI where he directs the Semantic Scholar research group. His work focuses on Natural Language Processing (NLP) and Human-Computer Interactions (HCI) research aimed at helping scholars discover, understand, and synthesize scientific literature.

Relevant resources noted in this presentation include the following:

• Papermage: An open-source Python toolkit for processing visually-rich scientific documents
• PaperMage project demo
• Semantic scholar topic page beta
• SciRepEval benchmark for training and evaluating document representations
• SPECTER2 code for document embedding models for scientific tasks
• SPECTER2 models on hugging face
• ForeCite repository containing data and code related to the paper “High-Precision Extraction of Emerging Concepts from Scientific Literature”

This webinar will be the first in a virtual series designed to advance scientific research and collaboration among Proto-OKN researchers.

Proto-OKN Workshop

Tuesday, May 7th, 2024
Knowledge Graph Conference 2024
Cornell Tech NYC

The goal for this 90 minute workshop was to enlighten the broader KGC community about the various avenues available for contributing to the Proto-OKN initiative and to explore the multitude of ways Proto-OKN can be leveraged for specific application scenarios. The initiative is interested in fostering a deep connection with companies specializing in knowledge graph platforms, initiating discussions on the integration of knowledge graphs developed through various techniques, and addressing the challenge of answering complex queries by navigating across multiple graphs.

Chaitan Baru & Jemin George
Directorate for Technology, Innovation and Partnerships (TIP)
U.S. National Science Foundation