This paper proposes a method to help robots understand object semantics. The method presented in this paper can enhance robot’s performance and efficiency while working with ambiguous instructions to interact with unfamiliar objects. Specifically, the proposed method can reduce the complexity of assigning the functions, properties or other characteristics for each object which robot may interact within a social environment. The method assists the robot to comprehend the scene based on semantics analysis of the dictionary definition. The proposed semantics comprehension method includes the comprehension of dictionary definitions, the formulation of logic representation, and the generation of natural-language descriptions. The applicability of the approach has been demonstrated. The model performance has been evaluated based on precision, recall, and f-score. Both logic representation formulation results and natural language representation results have been displayed.