太原理工大学学报2025，Vol.56Issue(5)：855-865,11.DOI:10.16355/j.tyut.1007-9432.20240910
频域协同增强的侧信息融合序列推荐模型
Frequency-Domain Cooperative Enhancement Model for Side Information Fusion in Sequential Recommendation
摘要
Abstract
[Purposes]Sequential recommendation with side information aims to predict users'preferred items according to user behavior sequences.However,existing methods integrate side infor-mation in the time domain,which often struggles to address high-frequency noise in user sequences.[Methods]To solve this issue,a frequency-domain cooperative enhancement model was proposed for side information fusion in sequential recommendation.The model applied adaptive filtering in the frequency domain to capture users'deep interest preferences and employed adaptive thresholds to re-duce noise effects.The Fourier transform was used to convert item and side information sequences into the frequency domain for feature fusion.The sparse nature of the spectrum helped to enhance im-portant features while suppress irrelevant noise.In addition,a frequency-domain prediction loss was introduced to further improve prediction accuracy.[Results]According to experiments are conducted on four public datasets,the proposed method outperforms the best baseline.The Recall@10 and Re-call@20 improved by 4.35%and 4.57%,respectively,while the average NDCG@10 and NDCG@20 increased by 2.45%and 3.02%,respectively.These results demonstrate the effectiveness and superi-ority of the proposed method.关键词
序列推荐/侧信息/自适应滤波/频域融合/噪声Key words
sequential recommendation/side information/adaptive filtering/frequency-domain fu-sion/noise分类
信息技术与安全科学引用本文复制引用
李泯沁,朱小飞,陈旭..频域协同增强的侧信息融合序列推荐模型[J].太原理工大学学报,2025,56(5):855-865,11.基金项目
国家自然科学基金资助项目(62472059) （62472059）
重庆市自然科学基金项目(CSTB2022NSCQ-MSX1672) （CSTB2022NSCQ-MSX1672）
重庆英才计划项目(CSTC2024YCJH-BGZXM0022) （CSTC2024YCJH-BGZXM0022）
重庆市教育委员会科学技术研究计划重大项目(KJZD-M202201102) （KJZD-M202201102）
