An Improved Human-Computer Interaction Content Recommendation Method Based on Knowledge Graph

Since human-to-human communication is contentful and coherent, during human-computer interaction (HCI), people naturally hope that the computer's reply will be contentful and coherent. How to enable computers to recognize, understand and generate quality content and coherent responses like humanshas drawn a great deal of interest from bothindustry and academia.In order to address the current problems of lack of background knowledge of robots and low consistency of responses in open-domain HCI systems, we present a content recommendation method on the basis of knowledge graph ripple network. In order to realize a human-computer interaction system with better content and stronger coherence, and this model simulates the real process of human-human communication. At first, the emotional friendliness of HCI is gotten through calculating the emotional confidence level together with emotional evaluation value of HCI. After that, the external knowledge graph is introduced as the robot's background knowledge, and the dialog entities are embedded in the ripple network of knowledge graphsto access the content of entities that may be of interest to participants. Lastly, the robot reply is given by comprehensively considering the content and emotional friendliness. The results of the experiments indicate a comparison with comparative modeling approaches like MECs models, robots with emotional measures and background knowledge can effectively enhance their emotional consistency and friendliness when performing HCI.