本文主要要解决的问题: 服装推荐。
本文关于服装推荐系统特别的地方:除了用户购买记录,还可以视觉图片,文字描述,社会关系,季节和流行趋势等角度进行用户偏好的挖掘。
具体做法:考虑Instagram和shopping web之间的信息迁移。
Instagram有社交关系(相似朋友,关注的服装模特),用户发布的有关服装的图片,文字描述等;Shopping Web有用户的购买记录,带有分类标记的服装图片和文字描述。
核心问题: domain adaptation strategy: two Convolutional neural sub-networks framework for integrating visual inputs with user feedback as an enhancing factor of image multi-label prediction.
关键技术:
(1) 通过讲deep CNN domain adaptation methodology 计算Instagram和Shopping website中图片的相似度,有了相似度之后,就可以把Shopping Website中的有关label信息同Instagram之间相互传递,如图片中服装的商标;这样一来,Instagram中有关生活场景的文本描述和Shopping website中本就有的文本描述就可以相互联系起来供进一步使用,如对图片的语义分析。
(2)有了图片之后,完全可以通过对图片的分析(deep pixel-wise),发现图片上的内容(如裙子的长度),再加上(1)中的文本分析,便形成了从图片到文本的关于用户的特征表述,结合用户的历史行为,就可以构造出跨领域的推荐系统。
启发:
以图片为桥梁,CNN作为基本技术,链接了两个domains,主要核心在于统一的feature space,这样就可以计算距离,进而可以迁移label等信息。
文献题目 | 去谷歌学术搜索 | ||||||||||
Deep Cross-Domain Fashion Recommendation | |||||||||||
文献作者 | Shatha Jaradat | ||||||||||
文献发表年限 | 2017 | ||||||||||
文献关键字 | |||||||||||
Fashion recommendation; deep learning; cross-domain knowledge transfer; transfer learning; domain adaptation; CNN | |||||||||||
摘要描述 | |||||||||||
With the increasing number of online shopping services, the number of users and the quantity of visual and textual information on the Internet, there is a pressing need for intelligent recommendation systems that analyze the user’s behavior amongst multiple domains and help them to find the desirable information without the burden of search. However, there is little research that has been done on complex recommendation scenarios that involve knowledge transfer across multiple domains. This problem is especially challenging when the involved data sources are complex in terms of the limitations on the quantity and quality of data that can be crawled. The goal of this paper is studying the connection between visual and textual inputs for better analysis of a certain domain, and to examine the possibility of knowledge transfer from complex domains for the purpose of efficient recommendations. The methods employed to achieve this study include both design of architecture and algorithms using deep learning technologies to analyze the effect of deep pixel-wise semantic segmentation and text integration on the quality of recommendations. We plan to develop a practical testing environment in a fashion domain. |