融合局部最优划分长短期兴趣的序列推荐
A Sequential Recommendation with Local Optimal Partition Long-Short Term Interest
孙克雷 1孙赛2
作者信息
- 1. 安徽理工大学计算机科学与工程学院,安徽淮南 232000
- 2. 安徽理工大学人工智能学院,安徽淮南 232000
- 折叠
摘要
序列推荐中划分用户的长期和短期兴趣非常重要.现有的序列推荐模型多简单预设短期兴趣长度,但性能提升不明显.为了更好地建模用户长短期兴趣,本文提出了一种融合局部最优划分长短期兴趣的序列推荐模型,采用了一种局部最优短期兴趣长度算法,自动和自适应搜索最优短期兴趣长度,并设计了 MLP层分别对长短期兴趣建模.在三个数据集上进行实验,结果表明运用该模型能够取得与最先进模型具有竞争力的性能.
Abstract
It is very important to identify both the user's long-term and short-term interest in sequential recommendation.Most of the existing models simply preset the short-term interest length,but the performance improvement is not obvious.For better model the user's long-term and short-term interest,this paper proposes a sequential reconmendation model with local optimal partition long-short term interest adopts a local optimal short-term interest length algorithm to automatically and adaptively search for the optimal short-term interest length,then designs an MLP layer to model long-term and short-term interests respectively.Through three experiments conduc-ted on a dataset,the model achieved competitive performance with the state-of-the-art models.
关键词
序列推荐/MLP/长短期兴趣/局部最优Key words
sequential recommendation/MLP/long-short term interest/local optimal引用本文复制引用
基金项目
国家自然科学基金项目(62076006)
安徽省重点研究与开发计划项目(2022AH050821)
出版年
2024
NETL
NSTL
万方数据