The EKAW conference is a biannual conference gathering the knowledge engineering community, both from the more technical side, as well as from the side of knowledge management and conceptual modelling. The conference had around 140 attendees this year, and took place in Amsterdam, The Netherlands, from November 26 to 28. Riley Capshaw received the award at the closing ceremony on the last conference day.
Eva Blomqvist, you are the supervisor of Riley Capshaw, how would you describe his area of research?
"Riley Capshaw is a PhD student in computer science in the intersection between knowledge representation and modelling, machine learning, and language technologies. He is working on approaches for modelling and accessing the information encoded in natural language texts, either by extracting facts from the text, or simply querying the text as if it was a knowledge base. For this purpose, he uses language models for text understanding, but targets methods that are reliable, flexible and avoid fine tuning, in order to cater for a changing set of facts and a changing context."
What is the awarded article about?
"The article is about contextualisation of text representations in this context, and the experiments show that by using our method you can achieve results that are on par with fine tuned models for specific domains, but without the fine-tuning step."
Read more:
- The article entitled Contextualizing Entity Representations for Zero-Shot Relation Extraction with Masked Language Models
- The EKAW-24 Conference
