Visualization is great way to understand data, but it breaks down when the data gets big. Simply plotting everything to the screen won’t work, because there isn’t enough screen real estate, interactions slow to a crawl, and human working memory isn’t up to the task anyway. Big data requires specific interaction techniques for visual exploration, such as filtering, summarization, and context. We’ll go over some basic principles, and I’ll show examples of recent systems, including our work on the Overview Project, a system for visual exploration of huge unstructured document sets.