基于深度学习的游客餐饮评论情感分析——以秦皇岛市为例
Emotional Analysis of Tourists’ Catering Comments Based on Deep Learning—Taking Qinhuangdao as an Example

游客餐饮体验直接影响整个城市的旅游竞争力，对旅游业的影响不容小觑。游客餐饮评论体现了游客对旅行目的地餐饮体验的真实感受，具有很高的研究价值。本文对秦皇岛市游客餐饮评论进行情感分析，并基于负面评论探寻影响游客餐饮体验的因素。分别采用Word2vec和BERT-wwm-ext预训练语言模型作为词嵌入层，将词向量化结果输入TextCNN、GRU、BiLSTM深度学习模型进行对比分析。训练结果表明：BERT-wwm-ext-BiLSTM模型准确率达96.89%，模型效果优于其他对比模型。最后对该模型分类结果的负面评价进行主题分析，确定负面评价主要涉及服务、价格、环境、菜品以及味道五个方面。相关部门应及时做出整改，提高游客餐饮体验，增强秦皇岛旅游核心竞争力。
Tourists’ catering experience directly affects the tourism competitiveness of the whole city, and its impact on tourism should not be underestimated. Tourists’ catering comments reflect tourists’ real feelings about the catering experience of their travel destinations, which has high research value. This paper makes an emotional analysis of tourists’ catering comments in Qinhuangdao, and explores the factors that affect tourists’ catering experience based on negative comments. Word2vec and BERT-wwm-ext pre-training language models are used as word embedding layers respectively, and the results of word vectorization are input into TextCNN, GRU and BiLSTM deep learning models for comparative analysis. The training results show that the accuracy of BERT-wwm-ext-BiLSTM model is 96.89%, and the effect of the model is better than other comparative models. Finally, the negative evaluation of the classification results of this model is analyzed, and it is determined that the negative evaluation mainly involves five aspects: service, price, environment, dishes and taste. Relevant departments should make timely rectification to improve the catering experience of tourists and enhance the core competitiveness of Qinhuangdao tourism.