Using Twitter to explore consumers' sentiments and their social representations towards new food trends

摘要

Purpose This paper investigates the use of Twitter for studying the social representations of different regions across the world towards new food trends. Design/methodology/approach A density-based clustering algorithm was applied to 7,014 tweets to identify regions of consumers sharing content about food trends. The attitude of their social representations was addressed with the sentiment analysis, and grid maps were used to explore subregional differences. Findings Twitter users have a weak, positive attitude towards food trends, and significant differences were found across regions identified, which suggests that factors at the regional level such as cultural context determine users' attitude towards food innovations. The subregional analysis showed differences at the local level, which reinforces the evidence that context matters in consumers' attitude expressed in social media. Research limitations/implications The social media content is sensitive to spatio-temporal events. Therefore, research should take into account content, location and contextual information to understand consumers' perceptions. The methodology proposed here serves to identify consumers' regions and to characterize their attitude towards specific topics. It considers not only administrative but also cognitive boundaries in order to analyse subsequent contextual influences on consumers' social representations. Practical implications The approach presented allows marketers to identify regions of interest and localize consumers' attitudes towards their products using social media data, providing real-time information to contrast with their strategies in different areas and adapt them to consumers' feelings. Originality/value This study presents a research methodology to analyse food consumers' understanding and perceptions using not only content but also geographical information of social media data, which provides a means to extract more information than the content analysis applied in the literature.