Research On Sentiment Recognition Of Housing Price Twitter

Housing price satisfaction is caused extensive concern of society as a measureof an important indicator of social development. But because it’s hard to quantifyand collect the data,so relevant researches are unable to be further. Along with theadvance of Internet technology, online discussion platforms develop quickly such assina twitter, theme BBS, etc. People speak up sufficiently by these new channels.Some information of these speech associate with housing price high. Theseinformation is a kind of public housing price satisfaction fragment expression.Housing price twitter emotion classification, refers to information identification ofthe big orders of magnitude of house prices twitter emotion by the method of datamining, to provide support for the research on house price satisfaction.In this paper, we take housing twitter of Beijing for study. First, we collect allthe twitter messages by the keywords housing price of Beijing from January2011to January2014, a total of59,957. Then, we build a emotional inclinations classifierused by n-Gram language model. Make the accuracy of classifier reach95%throughcontinuous optimizing the training. Finally, on the premise of satisfactory accuracy,we analyze the emotional tendencies contained in twitter massages of housing price.On the basis of the results obtained in this paper, we take the relationshipbetween public satisfaction and housing price for empirical analysis. First, weidentify the emotional tendencies per month about the Beijing housing price basedon n-Gram language model and calculate the mood scores. Then, we do the statisticanalysis with three variables: new housing sales price index, average selling priceand the growth rate of sales price. In the end, the regression analysis results showthat the relationship of the public satisfaction and housing price growth trend or thegrowth rate of sale price is negative correlation, and the relationship of the publicsatisfaction and the average sale price is positive, but the second relationship isrelative weak.The empirical analysis sufficiently proves the meaning of this researchin chapter four. And the empirical analysis has carried on a new exporation forChinese text automatic identificatinon of emotional tendency in the field of realestate, provides data support and theoretical basis for the government to make publicpolicy, also provides a good idea for scholars to study text emotional tendency.