Systems-of-Systems (SoS) are composed of multiple independent systems called constituents that, together, achieve a set of goals by means of emergent behaviors. Those behaviors can be deliberately planned as a combination of the individual functionalities (herein named as features) provided by the constituents. Currently, SoS engineers heavily rely on their own creativity and prior experience to combine the features and design the behaviors. However, the limitation of human perception in complex scenarios can lead to engineering sub-optimized SoS arrangements, potentially causing waste of the resources, sub-optimal services and reduction in quality. To handle the aforementioned issues, this article presents a machine learning-based mechanism for inferring/suggesting emergent behaviors that could be designed over a given set of constituents. An initial dataset was elaborated from a systematic mapping to feed the mechanism and a web-application was developed as a means for experts to evaluate this mechanism. Results revealed that the algorithm developed is capable of predicting feasible emergent behaviors for different sets of constituents and the system can be useful in the sense of aiding SoS engineers and experts in the bottom-up design of these behaviors.