Cognitive Cosmetics Explorer
Publication Date: 2017-Jun-09
The IP.com Prior Art Database
The disclosed details a method of analyzing a user’s social media activity with image recognition and cognitive capabilities to make cosmetic product recommendations that are more meaningful than the processing performed by text analytics on captions, comments, and hashtags alone. In this method, the cognitive aspects of the cosmetic product recommendation are capable of revealing deeper insight into a user’s preferences. Beyond text, this type of classification allows for an understanding of color, style, genre, occasion, period/decade, and more. For example, an image classifier trained on beauty and cosmetic trends through the decades would be able to identify a bee-hive hairdo, thick winged eyeliner, and coral lipstick on a photo of a model as components of a 60s era look and recommend products that can be used to achieve such an aesthetic. This determination can be made regardless of the photo’s influence score, account, tagging, caption, or comment. Social metadata and text can be used to augment the visual recognition component of this method and will improve the identification of subtle style nuances and characteristics. Such a method would be of particular interest to consumers, fashion industry professionals, photographers, and designers of all kinds.
Cognitive Cosmetics Explorer
It is very popular to browse makeup and cosmetics on social photo sharing networks such as Pinterest and Instagram, but the number of posts related to makeup application, style, brands, and colors is overwhelming. One could spend hours browsing through photos and only find a few items to 'like' or 'pin.’ It is also difficult to pinpoint products and colors a user would need to purchase to achieve or replicate the style from a photo. The Cognitive Cosmetics Explorer helps Pinners and Instagrammers quickly find photos of makeup applications and cosmetic styles that match their taste. The CCE can also recommend and guide the purchase of products needed to achieve specific looks.
The Cognitive Cosmetics Explorer shows users pictures of makeup products and styles on Instagram and Pinterest that they may not have seen based on traditional recommender algorithms. Currently a user can see related, suggested, and similar pictures based on their network/community activity and simple text analytics. The CCE uses Visual Recognition to produce recommended pictures based on color, composition, and style that matches a user's pinned/liked activity. Text analytics are still used in CCE via Relationship Extraction to provide the most accurate results. The CCE app also integrates ecommerce into the social experience so that users can quickly shop for the products they have discovered.
The CCE works by first requesting the liked or pinned photos from a user....