A method and apparatus to automatic curate the marketing creative/content employ the learning of internet contents. This method can be applied on all the contents/creatives curated based on predefined assets. One example of the content is image of women's apparel set which includes the dress, shoes and other assets like bag. The internet is full of the contents. There are high quality contents and low-quality ones. Examples of simple indicator are likes/views ratio or click through rate to distinguishing them. These indicators could be biased, however if there are enough viewers, it will still be statistically significant. With enough computing power, we can learn the patterns of high-quality content vs lower quality ones using these indicators. We created a process to automatically learn the assets groups and the arrangement of the assets from the internet contents and use the knowledge of the learning to create new content and creatives using a commercial product list or catalog for marketing purpose.
展开▼
