Detecting Users' Dietary Preferences and Their Evolutions via Chinese Social Media

Abstract

Dietary preferences are linked to human life and region culture. With the rapid development of the Internet, people are becoming frequently interested in sharing their opinions about dietary in social media. This article aims to mine social media users' dietary preferences and their evolutions with user generated content. The authors use microblogs from weibo.com to detect dietary preferences and their evolutions of social media users in China via sentiment analysis. First, the authors compare four aspect extraction methods to obtain dietary aspects. Second, sentiment polarities of aspects and dishes are identified. Finally, dietary preference evolutions are analyzed. Empirical analysis shows that social media users in weibo are not satisfied with status quo of a Chinese diet. Meanwhile, gender and region have significant effects on dietary preferences, and users' dietary preferences change over time. In addition, experimental results show that contextual information is useful in extracting dietary aspects.