Identifying Fashion Accounts in Social Networks
Identifying Fashion Accounts in Social Networks
复制标题
识别社交网络中的时尚帐户
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Ranjitha Kumar
中科院分区：
文献类型：
--
作者：
D. Lee;Jinda Han;Dana Chambourova;Ranjitha Kumar
时尚业的特点是潮流的起伏。随着社交媒体、时尚博客和快时尚运动的兴起，自下而上的时尚潮流正在以越来越快的速度涌现。在新的影响因素和趋势发生时，识别它们对零售商来说是一项挑战。作为第一步，本文提出了一个用于识别社交媒体上与时尚相关的账户的分类器。为了开发这个分类器，我们使用基于内容的滚雪球抽样方法收集了10k个Twitter账户的数据集，并为这些账户众包了地面事实标签。我们训练一个分类器来识别Twitter账户是否与时尚相关，并评估我们方法的效果。我们希望利用这个分类器来识别关键的时尚影响者，并对时尚趋势进行大规模监测。
The fashion industry is characterized by the ebb and flow of trends. With the rise of social media, fashion blogs, and fast-fashion movement, bottom-up fashion trends are emerging at an ever-increasing rate. Identifying new influencers and trends as they happen is challenging for retailers. As a first step, this paper presents a classifier for identifying fashion-related accounts on social media. To develop this classifier, we collected a dataset of 10k Twitter accounts using a content-based snowball sampling approach, and crowdsourced ground-truth labels for these accounts. We train a classifier that identifies whether a Twitter account is fashion-related and evaluate the efficacy of our method. We hope to leverage this classifier to identify key fashion influencers and conduct large-scale monitoring of fashion trends.