Jointly linking visual and textual entity mentions with background knowledge Invited Talk