The Future of NLP in e-Commerce: Generative Multimodal Language Models
Recent advances in Deep Learning have resulted in a new generation of natural language technologies. In the e-commerce setting, new transformer-based models have enabled enhancements across a vast array of product features and services. For instance, in online shopping product descriptions, call-to-actions, and blogs are all primary ways to inform and attract customers, playing crucial roles in conversion rates and SEO. At Einstein, we have developed new multimodal conditional natural language models trained to automatically craft unique, interesting, contextualized copy. We will present these new methods for generating new and enhancing existing text e-commerce, e.g. product catalogs, merchant sites, and other marketing channels.
Michael received a doctorate in mathematics from the University of Wyoming. Since 2012 he has led research and development teams at a number of successful Boston-based startups. Currently a lead data scientist on Salesforce's Einstein team, he enoys designing and building deep learning systems with applications to e-commerce and computer vision.