The field of recommendation systems is moving towards more efficient and scalable models, with a focus on addressing issues such as popularity bias and cold-start problems. Researchers are exploring new approaches, including graph-based methods and mutual-influence-aware models, to improve the accuracy and diversity of recommendations. Additionally, there is a growing interest in post-hoc debiasing methods and persistent homology to investigate negative embedding space. Notable papers include: Lighter-X, which proposes an efficient and modular framework for graph-based recommendation, and Post-hoc Popularity Bias Correction in GNN-based Collaborative Filtering, which introduces a method to correct for popularity bias in pre-trained embeddings. MIARec is also noteworthy, as it employs a gravity-based approach to measure mutual academic influence between scholars and incorporates this influence into the feature aggregation process. SMILE is another significant contribution, as it proposes an item representation enhancement approach based on fused alignment of semantic IDs.