Abstract: In order to improve the human-machine interaction performance of a home service robot and to deal with the drawback of neglecting user emotion during service recognition, a user emotion based autonomous service cognition method and a personalized service selection strategy for robot are presented. Firstly, an emotion-space-time ontology model is built based on the ontology technology of intelligent space and the information of user emotion and space-time, to eliminate the heterogeneity of information in intelligent space. Secondly, the emotion-space-time rule base is encoded and then used to train BP (backpropagation) neural network reasoner. The real-time updated information in intelligent space and the trained neural network are matched to automatically generate the robot services. The autonomous cognition of user emotion based robot service tasks is realized. Finally, the user emotion is taken as a reward feedback signal to dynamically adjust the user preference of each subclass service, and the personalized service selection is achieved. The simulation result shows that a robot based on the proposed method can autonomously achieve the user emotion based service recognition and provide a personalized service according to the variation of user preference. The proposed method effectively improves the intelligence and flexibility of the home service robot and enhances user experience.