News
CIRES Team at The Web Conference in Sydney
It’s been a big week for the CIRES Team from The University of Queensland at The Web Conference in Sydney! CIRES PhD Researcher Elyas Meguellati gave two presentations, including a workshop paper titled “Are Large Language Models Good Data Preprocessors?” on investigating whether Large Language Models (LLMs) can be good for cleaning text, and when […]
It’s been a big week for the CIRES Team from The University of Queensland at The Web Conference in Sydney!
CIRES PhD Researcher Elyas Meguellati gave two presentations, including a workshop paper titled “Are Large Language Models Good Data Preprocessors?” on investigating whether Large Language Models (LLMs) can be good for cleaning text, and when to strategically alternate between LLMs and rule-based methods. The second day he presented his PhD topic, journey, and progress as part of the PhD Symposium. Elyas had the opportunity to be mentored by world renowned scholar in machine learning, Professor Irwin King from The Chinese University of Hong Kong.
“I received valuable input and questions from the scholars and audience to consider for my research. It was a great opportunity to make new friends and connections with similar interests from all over the globe,” said Elyas.
Our PhD Researcher Hongliang Ni also attended and presented her work on operationalising harmlessness in online AI systems.
“It was a fantastic experience attending the WWW-25 PhD Symposium. I was inspired by the breadth of research presented, especially through the keynotes and paper sessions. One key takeaway for me was recognising both the potential and the vulnerability of multi-agent frameworks. Moving forward, I believe it’s crucial to focus on building systems that are not only effective but also robust. I’m grateful for the opportunity to learn, reflect, and connect with such a vibrant research community.”
CIRES Alum Dr Junliang Yu presented the Demo paper “BiasNavi: LLM-Empowered Data Bias Management.” BiasNavi is an LLM-powered toolkit developed by the Centre and designed to make bias management in data more accessible. It guides users through a structured, intelligent workflow – from bias detection to mitigation – leveraging the power of LLMs to streamline and simplify complex processes.
Congratulations to CIRES Chief Investigator Prof. Gianluca Demartini and PhD researchers Stefano Civelli & Pietro Bernardelle who won the best paper award at the Workshop on Multimodal Content Analysis for Social Good for their work “The Impact of Persona-based Political Perspectives on Hateful Content Detection“. Great work team!
Thank you to the ACM, Association for Computing Machinery and www25 organisers for an excellent conference.