Text-Generated Fashion Influence Model: An Empirical Study on Style.com Yusan Lin Penn State University yusan@psu.edu Yilu Zhou Fordham University yilu.zhou@gmail.com Heng Xu Penn State University hxu@ist.psu.edu Abstract Various fashion theories have been proposed to explain how fashion works and why it works that way. However, there is little research empirically examining fashion designers’ influences even though the benefit of understanding this field is significant. Unlike many other innovation domains such as patents where citations are explicit, a fashion designer hardly claims that s/he is influenced by others. To trace the hidden fashion influence network, we propose a novel approach to analyze the design influence in fashion industry by comparing similarity between designers in adopting same fashion symbols. We applied this approach to 6,180 runway reviews collected from Style.com between the year of 2000 to 2014 and constructed hidden influence links and the fashion influence network. Result of our approach is compared to 11 lists of “top fashion designers.” We believe this work is one of the first to empirically examine the design influence relationships among fashion designers and to visualize the design influence network. Also, this work explored the way of finding implicit links from textual review data, which may be applied to other fields as well. 1. Introduction Fashion is everywhere. It can be a way of wearing, a way of living, or even a way of thinking. But most of the time, when talking about fashion, the first thing most people think about is the way people dress in a given context. Outfits, items or persons considered as ‘fashionable’ are usually because the looks are trendy in that specific season. The notion of being fashionable changes from time to time, making fashion itself an extremely dynamic phenomenon. Designers in the fashion industry keep creating and updating new fashionable elements. But where do their inspirations come from? How do designers influence each other in such a way that they collectively drive the fast-paced fashion trends? Fashion is a highly subjective industry, where people influence and are influenced by each other because the urge of “following the trend setter” while there is no solid measurement of calculating how influential one is in the fashion industry. There are countless sources announcing the “top designers” without explaining how and why they determine those designers to be the ones. Is there any scientific way of measuring a fashion designer’s influence throughout the fashion industry quantitatively? To address this question, it is important to understand how fashion works and to reveal the insight under all fashion trends. Enough understanding in this domain can serve as a guide for fashion companies to make decisions; being able to foresee how fashion changes in the future can help fashion designer companies minimize the risk when deciding new designs for the next seasons [1]. Existing research has proposed various fashion theories, trying to explain how fashion works and why it works that way. Although conceptual and mathematical models have been proposed to conceptualize fashion trends, there has been limited empirical research conducted to validate these conceptual models with real data due to the reason that data in fashion industry is not highly accessible. Specifically, there is almost no research on examining or even defining fashion designers’ innovation and influence. “Fashion is one of the most important creative industries”, and yet the topic of examining design influences has not received enough attention in literature. We aim to fill in this gap by proposing a quantitative model of fashion influence network using fashion runway reviews from Style.com. Since “each new fashion is an outgrowth or elaboration of the previously existing fashion [2]”, we believe that we can trace the influence flow from earlier designs to later designs by using historical data to analyze silhouettes, shapes, colors, fabrics, and design details of specific objects. We construct a fashion taxonomy based on the domain knowledge and data collection, then derive implicit influence links from our design similarity model and construct a fashion influence network so as to understand the influence of fashion trend within the constructed network. 2015 48th Hawaii International Conference on System Sciences 1530-1605/15 $31.00 © 2015 IEEE DOI 10.1109/HICSS.2015.438 3642