基于用户评论的深度学习模型预测旅游景点推荐

A Deep Learning Model Based on User Reviews to Predict Tourist Attractions Recommendations

随着在线旅游交易规模的增长,通过个性化推荐技术为游客提供更好的服务,吸引了业界的关注。本文主要探讨在线旅游平台上用户评论在景点推荐过程中的作用,首先利用用户评论构建景点间社会网络,并基于深度学习模型LSTM预测用户对景点的情感得分;然后按照时间序列对情感得分加权作为用户对景点的评分;最后基于评分、景点社会网络信息和协同过滤算法实现景点的推荐。采用携程网的用户评论数据对推荐效果进行检验,实验结果表明在线平台的用户评论在景点推荐的多样性和新颖性方面影响效果显著,但对推荐准确性改善并不显著。

With the growth of online tourism transaction scale, providing better services for tourists through personalized recommendation technology has attracted the attention of industry and academia. This paper mainly discusses the role of user comments in the process of scenic spot recommendation on the online tourism platform. Firstly, the social network between scenic spots is constructed by user comments, and the user’s emotional score on scenic spots is predicted based on the deep learning model LSTM;then, the emotion score is weighted according to the time series as the user’s score of the scenic spot;finally, the recommendation of scenic spots is realized based on scoring, scenic spot social network information and collaborative filtering algorithm. The data on CTRIP website is used to test the recommendation effect. The experimental results show that user comments of the online platform has a significant effect on the diversity and novelty of scenic spot recommendation, but the improvement of recommendation accuracy is not significant.