中文摘要：

本研究从在线评论的情感属性出发探索在线评论文本特征的动态变化走势，借鉴已有的情感分析框架，选取文本的主客观性、文本的情感极性和文本的情感强度三个维度，并从评论内容和标题文本两个角度提出研究假设。实证分析阶段，通过编写java程序采集京东网站上iPhone4手机的评论数据，利用逐步回归分析法对在线评论情感属性变量进行模型拟合，跟踪消费者在线评论内容的情感变化，结果显示评论内容的情感属性在三个维度上均存在动态变化特征，而标题文本的情感属性没有稳定的变化。研究结果丰富和完善了情感分析理论，对企业把握用户消费习惯以及有效管理在线评论提供了决策依据。

英文摘要：

This paper aims to explore the dynamic changes of online reviews＇ sentiment features. Through applying the existing sentiment analysis framework, we choose subjective classification, sentiment polarity, and sentiment intensity as the dimensions in this study. Then, we put forward the research hypotheses based on the reviews＇ content and title. Finally, in the part of data analysis, we code a specific java program to collect the online reviews data, which is about iPhone 4, from the www. jingdong, corn and to carry out stepwise regression analysis with the selected variables to track consumer＇s emotional change. The result suggests that sentiment features of reviews＇ content show a steady change amongst each of the three dimensions, however, unlike the reviews＇ content, sentiment features of reviews＇ title do not illustrate this kind of stable change. In conclusion, our research enriches and perfects sentiment analysis theory, and provides the basis for the enterprise to understand consumers＇ consumption habits and to effectively manage online reviews.