Urban Lifestyles Mining Based On Large Scale Sina Microblog Data

Half of the world’s population live in the urban area, and city is the center of modern society. Urban lifestyles contain many aspects, including food, clothing, shelter, transportation, study, entertainment and so on. Researching the status of urban lifestyles is meaningful both for government, enterprise, social organization and individuals. Traditional methods of researching urban lifestyles are mainly based on interviewing and questionnaire, requiring large human resources and working time. However in recent years, with the popularity of social network, huge, multi-level and fine grain user data are easier to be got. Researches based on social network also have become hot topics. Huge, public user data in social network can be good data resources for our research.In consideration of the weakness of traditional method and the advantages of the data in social network, our thesis first proposed to analyze urban lifestyles using large scale Sina Microblog user data. In addition, by solving two new problems-"Sleep quality evaluation of urban people" and "Analysis of subjective happiness index of major cities in China", we proved that our method is feasible by using the real big user data on Sina Microblog. Compared with traditional methods, the innovation points and major contributions of this thesis are as follows:1、In this thesis, we designed a distributed Sina Microblog crawling system. We can get the microblog data by our demand using the crawling system quickly. By the end of finishing this thesis, we successfully obtained 1.3 billion microblogs from 1.1 million urban Sina Microblog users, providing important data for our later researches.2、We first proposed to analyze urban lifestyles by using large scale user data on Sina Microblog. In addition, by using "Sleep quality evaluation of urban people" and "Analysis of subjective happiness index of major cities in China" as detail problems, we explained the feasibility of our method.3、Compared with traditional methods, our method can obtained the result based on large scale Sina Microblog data using less human resources and working time.4、In this thesis, we utilized both content info and time info in the same time. We designed detail math models and then gave accurate statistics analysis using real user data. Our method has good scalability and can be good reference for future studies.