Social Media Analytics for Decision Support in Fashion Buying Processes

The Web 2.0 and the emergence of numerous social media services enable individual users to publish and share information on the one hand and to discuss diverse topics online on the other hand. Accordingly, different research streams have emerged in order to tackle the diverse phenomena related to social media. Social media analytics as an interdisciplinary research field has arisen and integrates the different approaches of structural attributes, opinion/sentiment-related as well as topic/trend-related approaches. This research follows topic- and trend-related approaches with the methods content and trend analysis on social media text data. These methods might be applied on different domains including the fashion industry. This research focusses on the fashion industry for three reasons. Firstly, this industry is a highly consumer-oriented industry, and these consumers themselves are the users of social media services. Secondly, the industry faces challenges in meeting the demand of the customer on time. Thirdly, in the last years, fashion blogs have gained increased relevance from the consumers and the industry. Accordingly, the fashion blogs may contain information for supporting decision maker in the industry, to perform their tasks such as meeting the demand with a lower degree of uncertainty. The objective of this chapter is to explore the potential added value of social media analytics for fashion buying processes, not only by presenting an abstract approach, but more by conducting experimental analyses on a fashion blog corpus covering a 5 year time period. Based on the topic detection and tracking research which origins from the intelligent information retrieval, a research approach is presented by integrating a text mining process, on the detecting and tracking of fashion features and topics in the blog corpus. A fashion topic may refer to different features such as a colour, silhouette or style. While for the topic detection task, the feature colour is focussed, the topic tracking includes topics on silhouette, style, colour and decorative applications. The analyses have shown that it is possible to detect single colour and co-occurred colour occurrences. In addition, it was demonstrated that it is possible to track fashion topics over a 5 year time period in a fashion post corpus. The fashion buyer might have an added value for his activities by quantifying the individual perceptions through the application of the presented approach.