%0 Journal Article
%T 基于方向性偏好的个性化序列推荐模型<br>Personalized Sequence Recommendation Model Based on Directional Preference
%A 胡雨萱
%J Computer Science and Application
%P 2932-2944
%@ 2161-881X
%D 2021
%I Hans Publishing
%R 10.12677/CSA.2021.1112297
%X <div style=\"text-align:justify;\">
	<span style=\"font-size:14px;\">目前推荐系统主要基于用户偏好和物品的相似度等标量进行推荐，忽略了用户的偏好方向，增加了不相关推荐的风险。基于此，本文提出基于方向性偏好的个性化序列推荐模型，通过推荐符合用户偏好方向的物品，综合用户的偏好和需求进行推荐从而提高了推荐的准确性。本文以电影评论数据集为背景，使用选择器多头自注意力机制和改进胶囊网络分别提取用户的长期偏好和短期需求，然后根据用户长期偏好和短期需求构成的偏好向量进行推荐。实验结果表明，模型的推荐准确性指标相比原模型提升了52%。<br />
The current recommendation system is mainly based on scalar variables such as similarity of user preferences and items, ignoring the user’s preference direction, and increasing the risk of irrelevant recommendations. Based on this, this paper proposes a personalized sequence recommendation model based on directional preference, which improves the accuracy of recommendation by recommending items that conform to the user’s preference direction and integrating the user’s preferences and needs. Based on the movie review data set, this paper uses the selector multi-head self-attention mechanism and the capsule network to extract the user’s long-term preferences and short-term needs, and then recommends based on the user’s long-term preferences and short-term needs. The experimental results show that the recommended accuracy index of the model is improved by 52% compared with the original model.</span> 
</div>
%K 序列推荐，偏好方向，胶囊网络，多头自注意力机制，短期需求<br>Sequence Recommendation
%K Preference Direction
%K Capsule Network
%K Multi-Head Self-Attention
%K Short-Term Demand
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=47267