房价微博情感分类研究

发布时间：2018-09-07 15:57

【摘要】：房价满意度作为衡量社会发展的一个重要指标，正在引起社会的广泛关注但是由于难以量化数据收集繁琐时效性弱等困难，相关研究无法深入伴随着互联网技术的不断进步，在线讨论平台的快速发展壮大，如新浪微博主题论坛等，民众利用这些新兴渠道畅所欲言，在其中就包括与房价高度相关的大量言论信息这些信息背后就是民众对于房价的情感态度，是民众对于房价满意度的一种碎片式表达，这些碎片化的信息中就包含着民众对于房价的满意程度房价微博情感分类，是指利用数据挖掘的方法，对大数量级的房价微博进行情感倾向信息识别，借此为房价满意度研究提供支持本文以北京房价微博作为直接研究对象首先，采集了以北京房价为关键字的，2011年1月到2014年1月这个时间段内的所有微博数据，其中有效数据共计59957条然后，基于N-Gram语言模型构建情感倾向分类器通过不断优化训练集使分类准确率达到95%以上最后，在准确率达到要求的前提下，，挖掘出蕴含在房价微博中民众对于房价的情感倾向依据本文前两章所取得的成果，对民众满意度与房价之间的关系进行实证分析首先，利用基于N-Gram语言模型的情感分类器对每月的北京房价微博数据进行情感倾向识别，计算情绪得分，借此量化民众对于房价的满意程度然后，联系北京市每月的新建住宅销售价格指数这一相对值住宅平均销售价格这一绝对值，以及推算出的每月住宅销售价格增长率这三个变量进行统计分析最终，统计分析结果表明民众对于房价的满意程度受到房价绝对值和相对值的显著影响，且房价相对值对其影响程度更强，相比于房价绝对值进而，联系所查阅文献与相关理论进行模型结果的解释最后，本研究利用所取得的成果，联系房地产实践领域，给予提高房地产领域民众满意度的建议本研究为中文文本情感倾向自动识别在房地产领域进行了新的探索，为政府制定公共政策提供数据支持和理论基础，也为学者继续研究文本情感倾向提供很好的思路
[Abstract]:As an important index to measure social development, house price satisfaction is attracting wide attention of the society. However, due to the difficulty of quantifying data collection, such difficulties as tedious and weak timeliness, the related research can not go deep with the continuous progress of Internet technology. With the rapid development of online discussion platforms, such as the Sina Weibo theme Forum, people use these new channels to speak freely. Among them is a large amount of speech information that is highly relevant to house prices. Behind this information is the public's emotional attitude towards housing prices, a fragmented expression of people's satisfaction with housing prices. These pieces of information contain the people's satisfaction with the housing prices, Weibo's emotional classification of housing prices, which refers to the use of data mining methods to identify the affective tendency information of the house prices in the order of magnitude, Weibo. In order to provide support for the research on the degree of house price satisfaction, this paper takes Weibo as the direct research object. Firstly, we collect all Weibo data in the period from January 2011 to January 2014, which is based on the key word of housing price in Beijing. There are 59957 valid data, and then, based on the N-Gram language model, the classification accuracy of emotion tendency classifier is over 95% by continuously optimizing the training set. Excavating the emotion tendency of the people to the house price in Weibo of housing price, according to the results obtained in the first two chapters of this paper, the relationship between the satisfaction of the people and the house price is analyzed empirically, first of all, The emotion classifier based on N-Gram language model is used to identify the emotion tendency of Weibo data of housing price in Beijing every month, to calculate the emotion score, so as to quantify the people's satisfaction with the house price, and then, Connecting with the absolute value of the monthly sales price index of newly built residential buildings in Beijing, the absolute value of the average residential sales price, and the calculated monthly residential sales price growth rate, the three variables are statistically analyzed. The results of statistical analysis show that the satisfaction of the public with the house price is significantly affected by the absolute and relative value of the house price, and the relative value of the house price has a stronger impact on it, compared with the absolute value of the house price, With reference to literature and related theories to explain the results of the model finally, this study uses the results obtained, the real estate practice field, Suggestions for improving the satisfaction of people in Real Estate this study provides a new exploration for automatic identification of emotional tendencies in Chinese texts and provides data support and theoretical basis for the government to formulate public policies. It also provides a good way for scholars to continue to study the emotional tendency of text.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：G206;F299.23

【参考文献】