Unstructured Data Analytics

CHAPTER 3The Framework to Put UDA to Work

A woodsman was once asked, “Give me six hours to chop down a tree and I will spend the first four sharpening the axe.

—Abraham Lincoln

INTRODUCTION

Over the past decade, the explosion of digital information has provided an unprecedented opportunity for businesses and organizations to capture, store, and process different types of new data, both structured and unstructured. With the advent of computing power and the decreased cost of storing information, the real challenge that companies face today is the variety of the data they have at their disposal: e-mails, chats, tweets, audio, video images, and pictures.

Information such as customers' needs, feelings, and feedback, as well as employee narratives, stays buried in tweets, Facebook updates, and human resource information systems and files. These real-time insights encapsulating consumers' viewpoints about content, a product, a service, consumers' needs and preferences, or employee experience engagement or satisfaction remain untapped.

Enhancements in computing power originally propelled a variety of analyses of structured data, leveraging several statistical and mathematical techniques. However, due to computing limitations, it was only in the late 1990s that unstructured data began to be analyzed. Most recently, deep learning algorithms have been used to analyze videos and voice streams. Tech giants, such as Google, Facebook, and Amazon, and online hiring solutions, such as LinkedIn ...