Abstract: To improve the accuracy and efficiency of eye tracking on mobile device,and to reduce the cost of hardware,an eye movement data perception and computing method based on gaze recall was proposed.Firstly,setup the mechanism of gaze recall and self reporting based on short-term memory of human being,and then asked users to tap the screen of mobile device,and submit the gaze position data.Proposed the model of gaze data error compensation based on support vector regression,and used this model to correct the gaze data users recalled and submitted,so as to improve the data accuracy.To validate the gaze data error compensation model,user study was designed and conducted,and the results showed that after using the gaze data error compensation model,for different test tasks and images,accuracy of the gaze recalled data was improved by 15 % to 40 %.