In the first week of August, our group attended a Knowledge Prompting Hackathon organised by Kings College in London. The event started with a talk by Prof. Elena Simperl on knowledge prompting to set the stage of the research challenges to tackle for the next four days. After being assigned to a research challenge, each team explored the potential use of Large Language models for knowledge engineering in a specific area.
The hackathon was a collaborative, interdisciplinary sprint-style research endeavour. Researchers and practitioners from many different universities and backgrounds tackled the mission to prototype novel ideas, methods, tools, and evaluation frameworks. The innovative integration of LLMs in the Knowledge Engineering process to not only produce and access knowledge but also to ensure its authenticity and reliability was the common goal of all teams in the hackathon.
Two members of our group (Majlinda Llugiqi and Fajar Ekaputra) worked on an official challenge announced by ISWC (https://lm-kbc.github.io/challenge2023/), called Knowledge Base Construction from Pre-trained Language Models (LM-KBC). The goal of this challenge was to use any type of LLM to predict true triples for constructing a knowledge base. After only one week of work and dedication, they achieved second place in the overall ranking of the challenge.
Stefani Tsaneva and Katrin Schreiberhuber joined another challenge, where they explored the possibilities of LLMs to help in the collaborative aspect of Knowledge Engineering. During the hackathon, several options were explored, where the use of LLMs could assist the process and prototypes of such an assistant were developed. The result was a full framework of an LLM-assisted ontology engineering workflow, which could potentially save a lot of time for all stakeholders involved in the creation of a domain-specific ontology.
Overall, the event was a very engaging and interesting experience, where the participants could use their skills and curiosity to explore a new field of research as well as learn from each other and exchange ideas with like-minded individuals. We are all very happy about the opportunity to participate in this hackathon, returning home with new ideas and inspiration for our further research career.