Today, many international corporations have globally distributed supply chains. This exposes their operations to various local risks, e.g., natural disasters. To facilitate monitoring and assessment of these risks, corporations can identify their suppliers and their geographic locations. However, automated inferring of location information for supplier locations is problematic for areas where geocoding of addresses is not effective. In this paper, we present a method to infer location information from user-generated geographic information retrieved from Wikimapia and Foursquare. Using a sample of 139 Indonesian factories supplying four large international corporations, we compared results from our approach with locations inferred from four widely-used geocoding services. We found that best results could be achieved using user-generated information from Foursquare, where we could retrieve a location within 1km for 73\% of the factories. Given that coordinates are only an input for decision making, we showcase their importance using the example of volcanic activity. Therefore, we linked retrieved locations with semantic data in order to determine the risk exposure due to volcanic eruptions for each factory. Both steps combined present an approach for automated supplier risk assessment based on social and semantic data.