Social media has been shown to have potential to predict various real world events, such as movements in the stock market and the outcomes of political elections. In this paper we present the Football Whispers (FW), a website dedicated to fans discussing transfer rumours. The unique selling point of the site is that it provides a crowdsourced assessment of those rumours, measuring the relative likelihood of a player’s movements from social media chatter. This talk will focus on the rumour identification process, highlighting the role of open knowledge graphs and linked data to augment a domain knowledge-based to enable effective Named Entity Linking in noisy, informal social media messages.