Practical Natural Language Processing

Chapter 9. E-Commerce and Retail

Today’s new marketplaces must nurture and manage perfect competition to thrive.

Jeff Jordan, Andreessen Horowitz

In today’s world, e-commerce has become synonymous with shopping. An enriched customer experience compared to what a physical retail store offers has fueled this growth of e-commerce. Worldwide retail e-commerce sales in 2019 were $3.5 trillion and are projected to reach $6.5 trillion by 2022 [1]. Recent advancements in ML and NLP have played a major role in this rapid growth.

Visit the home page of any e-retailer, and you’ll find a lot of information in the form of text and images. A significant portion of this information consists of text in the form of product descriptions, reviews, etc. Retailers strive to utilize this information intelligently to deliver customer delight and build competitive advantage. An e-commerce portal faces a range of text-related problems that can be solved by NLP techniques. We saw different kinds of NLP problems and solutions in the previous section (Chapters 4 through 7). In this chapter, we’ll give an overview of how NLP problems in the e-commerce domain can be addressed using what we’ve learned in this book so far. We’ll discuss some of the key NLP tasks in this domain, including search, building a product catalog, collecting reviews, and providing recommendations.

Figure 9-1 shows some of these e-commerce tasks. Let’s start with an overview of them.

Figure 9-1. NLP applications in e-commerce