Enhancing Reciprocal Recommendation with Bidirectional Global-Local Insights

In the field of Reciprocal Recommender Systems (RRS), precisely aligning user preferences is a critical challenge, especially in application scenarios like online recruitment and social networking. Traditional recommender systems frequently underperform in these scenarios, primarily due to their failure to sufficiently address the bidirectional nature of interactions and the necessary fusion of global and local preference insights. To address this deficiency, our research introduces the Global-Local Bidirectional Enhancement (GLoBiE) framework. This innovative framework redefines the process of reciprocal recommendations by effectively merging global and local insights within a cohesive bidirectional model. The bidirectional model adeptly adapts to the changing nature of user preferences and intricately maps the dynamics of two-way interactions. Consequently, GLo-BiE attains greater precision and user satisfaction in its matches, thanks to its harmonized analysis of comprehensive interaction patterns and individual preferences. We conducted experiments on a real-world dataset, which provided concrete evidence of our approach's validity, illustrating its practical applicability and improved performance in RRS scenarios.