融合图表示学习和序列挖掘的景点推荐方法

Attractions recommendation method incorporating graph representation learning with sequence mining

针对旅游推荐系统中序列挖掘的推荐方法忽略了景点序列中复杂序列语义信息的问题,提出一种融合图表示学习和序列挖掘(graph representation learning and sequence mining,GRL-SM)的旅游景点推荐方法。利用图神经网络的方法,捕捉景点之间复杂的序列语义信息;考虑游客旅游过程中偏好会随时间变化的特点,利用注意力机制获取景点序列中蕴含的游客长短期偏好,实现个性化旅游景点推荐。真实数据集上的实验结果表明,该方法的推荐性能明显优于其它基线方法,验证了其有效性。

Aiming at the problem of ignoring the complex sequence semantic information in attractions sequence of recommendation method of sequence mining in tourism recommendation system,the travel attractions recommendation method incorporating graph representation learning with sequence mining(GRL-SM)was proposed.By using the method of graph neural network,the complex sequential semantic information between attractions in the sequence was captured.Considering the characteristics that tourists’preference might change over time in the process of tourism,the attention mechanism was used to learn the long-term and short-term preferences of tourists contained in the attractions sequence,and personalized attractions recommendations were provided.Experimental results on real data sets show that the proposed method is significantly better than other baseline met-hods,verifying its effectiveness.