Abstract: With the widespread use of smartphones, vibration feedback has become an important approach to enhance interaction experience. However, current vibration feedback design mainly focuses on technical implementation by engineers, lacking systematic optimization based on user needs and perceptions in different scenarios. This can lead to a mismatch between functionality and user experience. This study proposes a haptic feedback optimization algorithm based on vibration properties and user-perceived semantic features. First, vibration acceleration data were collected from four smartphones across eight scenarios, and key parameters including frequency, amplitude, and duration were extracted. A semantic adjective system with thirteen adjective pairs is built to describe vibration perception, and user ratings and ideal values are collected for each scenario. Based on these data, a mapping between semantic perception and physical parameters is used to derive optimal vibration parameters. Finally, experiments across eight scenarios show that the optimized waveforms are preferred by users and provide a better haptic experience than the original smartphone vibration feedback.