摘要
秦皇岛地理位置优越,旅游资源丰富,同时文化底蕴深厚,现代旅游业正在面临前所未有的大好发展机遇,秦皇岛如何抓住这一机遇,大力发展经济,实现旅游产业更好地发展,是一个现实课题。信息技术的飞速发展使得网络成为资讯流通的主要途径,而作为旅行者分享体验的重要媒介,旅游网站的评价内容对游客决策的影响显著。为此,从用户反馈中提取关键信息并精准判断其感情属性,对优化旅游服务品质具有重要意义,但以往的情感分析技术主要采用人工构建的情感词典和人工特征选择,这一过程既耗时又费力。为了优化模型性能并提高其工作效率,本文针对秦皇岛市五大著名景区的评论数据集,采用了基于深度学习的方法进行了情感分析。具体而言,我们使用了融合了注意力机制的双向长短期记忆(BiLSTM)模型来处理该数据集,旨在深入挖掘文本中的隐藏信息,并解决传统模型中存在的长期依赖问题。实验结果表明,与常用的机器学习模型,如卷积神经网络(CNN)相比,融合了注意力机制的BiLSTM模型在情感分析任务上取得了更为显著的效果提升。Qinhuangdao enjoys a superior geographical location, rich tourism resources and profound cultural heritage. The modern tourism industry is now facing unprecedented development opportunities. How Qinhuangdao can seize this opportunity to vigorously develop its economy and achieve better development of the tourism industry is a practical issue. The rapid development of information technology has made the Internet the main channel for information flow. As an important medium for travelers to share their experiences, the evaluation content on tourism websites has a significant impact on tourists’ decision-making. Therefore, extracting key information from user feedback and accurately judging its emotional attributes is of great significance for optimizing the quality of tourism services. However, previous sentiment analysis techniques mainly relied on manually constructed sentiment lexicons and manual feature selection, which was both time-consuming and labor-intensive. To optimize model performance and improve its efficiency, this study adopted a deep learning-based approach for sentiment analysis on the comment dataset of the five most famous scenic spots in Qinhuangdao. Specifically, we used a bidirectional long short-term memory (BiLSTM) model with an attention mechanism to process this dataset, aiming to deeply mine the implicit information in the text and solve the long-term dependency problem existing in traditional models. The experimental results show that compared with commonly used machine learning models such as convolutional neural networks (CNN), the BiLSTM model with an attention mechanism achieved more significant performance improvements in sentiment analysis tasks.
出处
《统计学与应用》
2025年第4期62-68,共7页
Statistical and Application
基金
秦皇岛市科学技术研究与发展计划(编号:202302B039)。
