Neural-symbolic and statistical relational artificial intelligence both integrate frameworks for learning with logical reasoning. This survey identifies several parallels across seven different dimensions between these two fields. These cannot only be used to characterize and position neural-symbolic artificial intelligence approaches but also to identify a number of directions for further research.