An intellisense algorithm for smart TV based on inference logic model of graph

In a smart TV interactive system with body-gesture control, body-gesture is the single input channel way of interaction, which leads to heavy cognitive load and operating load for users. So we propose a perception algorithm for the intentions of users (IPA). Firstly, we define interactive-context events, and build a perceptual model for associating events in interactive-context during the process of interaction. Secondly, we abstract the basic interactive-context events and analyze the scene context information at current moment and the context of user´s body-gesture behavior. Finally, IPA can help the computer speculate the interactive-context event which may occur in future time and perceive the users´ intent so as to achieve smart TV´s intelligent perception. It is innovative because: (1) it discloses the relationship between body-gesture behavior model and the functional state of smart TV; (2) it verifies that the users´ intentions based on interactive context can be perceived. Experiments show that IPA algorithm can accurately predict the users´ intentions based on interactive context, and help users manipulate smart TV with harmonious, natural, smooth style.