Principles and Practice of Big Data, 2nd Edition

2

Providing Structure to Unstructured Data

Abstract

Most of the data collected today is unstructured. For data managers, unstructured data is any stored information that comes in different sizes (e.g., a tweet, an email message, a book, a library corpus), that contains information that may express a single concept in many different ways (e.g., September 3, 2012; Sept 3, 2012, 09/03/12, 03/09/12, 3/9/12, Labor Day), that is not neatly packaged into spreadsheet cells (e.g., an audio file, a photograph of the Declaration of Independence, a tweet), that cannot be assigned a numeric value (such as yesterday, TRUE, nil), or that does not conform in any way to a specific data standard. Examples of unstructured data would include just about everything ...