It's never been easy to access - much less analyse - the vast amount of data available to, for example, determine changes in customer behavior or sentiment, optimize routing of telephone switches based on call patterns, or analyze financial portfolio pricing or risk. Traditional business intelligence systems rely on highly structured (and usually transactional) data stored in massive data cubes and data warehouses, which requires significant upfront work to decide what is being analyzed and to ensure all the data is consistent with that goal - that is, you know what you are looking for. But that approach isn't useful for exploring trends or patterns, especially in external data that wasn't formatted for your needs.