ADM+S student paper accepted at prestigious Web Conference 2024
Author Natalie Campbell
Date 1 February 2024
ADM+S PhD Student Chenglong Ma from RMIT University will present his paper at the upcoming 2024 Web Conference in Singapore.
Chenglong’s PhD Supervisor and co-author Prof Mark Sanderson said, “this is a very prestigious conference, arguably the most important conference in database and information systems. Getting a paper into that conference is very difficult, only about 20% of the submissions are accepted.”
The paper titled ‘Temporal Conformity-aware Hawkes Graph Network for Recommendations’, acknowledges that traditional recommender systems often overlook the impact of peer influence and conformity, assuming user behaviour is solely driven by individual interests.
However, indiscriminate bias elimination may lead to depersonalized recommendations, neglecting valuable information. The proposed TCHN model addresses this by distinguishing between two types of conformity behaviour: informational and normative.
Chenglong explains, “leveraging attentional Hawkes processes and sequence graph attention networks, TCHN effectively models the interplay between user self-interest and conformity, providing personalized recommendations.
“Experiments on real-world datasets reveal TCHN’s superior performance in accuracy, diversity, and fairness across user groups, highlighting its potential in mitigating conformity biases in recommender systems.”
This research paper aligns with Chenglong’s PhD focus on enhancing recommender systems through the integration of social influence and conformity. It extends the the theoretical framework of his PhD research, providing insights into the dynamics of user behaviour influenced by both individual interests and conformity factors.
Chenglon’s thesis topic was inspired by the changes that occurred in consumer behaviour during the Covid-19 pandemic. He observed that people stopped thinking about their short term needs and instead adopted a herd like mentality, panic buying items such as toilet paper that they thought might run out.
Chenglong’s research examines whether automated systems such as recommender systems that drive shopping websites, need to adjust in line with changing behaviours.
“Peer recognition of the novelty and quality of the research provides me with great encouragement to continue contributing meaningfully to the academic discourse in my field.
Having my paper accepted at this conference is a proud moment marking a successful conclusion to my PhD studies and establishing a strong foundation for my future academic career post-graduation,” said Chenglong.