A Project Recommender Based on Customized Graph Neural Networks in Online Labor Markets

[摘要] Project recommendation is a crucial task in the online labor markets, where freelancers seek projects that match their skills and preferences. The challenge in project recommendation is to effectively capture the complex relationships between projects, skills, and freelancers in a highdimensional space. Traditional recommender systems, such as matrix factorization, often suffer from sparsity, making it difficult to accurately recommend projects to freelancers. To address the challenges, this research proposes a customized graph neural network approach, called GCNRec, specifically using Light Graph Convolution Network (LightGCN) to effectively capture both explicit and implicit relationships between projects, skills, and freelancers, and make personalized project recommendations. The experiment results on real-world dataset demonstrate the effectiveness of the proposed method in solving the challenges in project recommendation in the online labor market.

[学科分类] 自动化工程

[关键词] project recommendation;online labor markets;graph neural networks;LightGCN