基于深度学习的旅游景点推荐模型研究

Research on Tourist Attraction Recommendation Model Based on Deep Learning

针对大数据造成的信息冗余问题,个性化推荐系统的出现解决了这个困扰,它能够有效地挑选和剔除信息,让用户从众多数据中找出对自己有用的内容。文章聚焦于旅游景点推荐,提出了一种深度卷积融合位置信息模型DCNN-MPN。首先,借助CNN的水平与垂直滤波器,从用户的过往行为中提取出具有价值的特性。紧接着,将用多头自我注意的数据融入原记录中。最后,通过计算评估目标对象的相似度。实验数据证实,文章提出的DCNN-MPN模型是行之有效的,它能有效地从用户的历史兴趣序列中提取出每个旅游点的权重以及主要和次要的偏好度,并有助于与当前点击率预测模型进行对比,从而实现精准的预测。

The emergence of personalized recommendation systems has effectively addressed the issue of information redundancy caused by big data.These systems can selectively filter and eliminate information,thereby helping users find relevant content from a vast amount of data.This paper focuses on the recommendation of tourist attractions and proposes a deep convolutional fusion model called DCNN-MPN,which incorporates location information.Firstly,valuable features are extracted from users’past behaviors using CNN’s horizontal and vertical filters.Then,data is fused with multi-head self-attention and integrated into the original records.Lastly,the similarity of the target objects is evaluated through computation.Experimental data confirms the effectiveness of the proposed DCNN-MPN model.It successfully extracts the weights of each tourist spot,as well as the primary and secondary preferences,from users’historical interest sequences.Moreover,it facilitates precise predictions by comparing with the current click-through rate prediction model.