MIND, Electronics, Prime Pantry
收藏资源简介:
本文使用的数据集包括MIND、Electronics和Prime Pantry,这些数据集由香港城市大学和腾讯公司共同创建,用于评估序列推荐模型的性能。数据集涵盖了大量用户与物品的交互记录,分别包含9,667,540、5,137,265和115,004条交互数据。数据集的创建过程涉及对用户历史行为的序列化处理,旨在捕捉用户兴趣的动态变化。这些数据集主要应用于推荐系统领域,特别是解决冷启动问题和提升推荐效果。
The datasets used in this paper include MIND, Electronics, and Prime Pantry. These datasets were jointly created by City University of Hong Kong and Tencent Inc., and are employed to evaluate the performance of sequential recommendation models. They cover a large number of user-item interaction records, with 9,667,540, 5,137,265, and 115,004 interaction entries respectively. The creation process of these datasets involves serializing users' historical behaviors, aiming to capture the dynamic changes of user interests. These datasets are mainly applied in the field of recommendation systems, particularly for addressing cold-start problems and enhancing recommendation performance.

- 1Pre-train, Align, and Disentangle: Empowering Sequential Recommendation with Large Language Models香港城市大学 · 2024年



