In Comperio we work on projects that aim to learn from the documents and social activity published on the web. The challenge is to summarize recent activity, drawing together current events with past activity and finding new sources to learn from. We base our on Elasticsearch and its significant terms technology. In this presentation we show how we expand on the base functionality provided in Elasticsearch to focus on areas such as immediate trends, entity identification and topic building using additional techniques from Information Retrieval (IR) and Natural Language Processing (NLP).