Evaluating Prompt-based Question Answering for Object Prediction in the Open Research Knowledge Graph