Processing Product Data For Marketplaces

This thesis describes problems of handling product data in e-commerce environment and focuses on their automatic batch processing, enhancing and cleaning, mainly for the purpose of selling on marketplaces. For this case, it is essential for sellers to be able to deliver properly structured, complete, up-to-date and correct product data in a short time. The author has chosen a startup company focusing on the subject to realistically describe the business needs. The thesis describes the data problem and presents, how it is currently being solved. As a practical example of how artificial intelligence algorithms deliver the business value in this industry, the author solves the task of finding missing values using text classification and compares the results to the current common approach. Finally, he introduces plenty of other relevant use cases where the same or similar state-of-the-art technologies could be very appealing from the business point of view.