Abstract
Related topics of mobile phones have been widely studied by researchers of design sciences. The message is one of the most important functions for a mobile phone. In this paper, a visual perception study for design of message icons on mobile phones is given by questionnaire investigation, and neural-network modeling and predicting. This study includes two parts, i.e., (1) questionnaire investigation, and (2) neural-network modeling and prediction. In the first part, the participants' visual perception of message icons for mobile phones is investigated by experiments, questionnaires and statistical analyses. In general, such investigation work is very heavy and time-consuming. This then motivates us to develop efficient tools for modeling and predicting such a problem. In the second part, investigated results of the first part are modeled and predicted by neural networks. The investigation results of the first part are consistent with existing studies. The programming results of the second part show that the neural network can well model and accurately predict the investigated results in the first part. With the help of neural networks, one can well model and accurately predict the full investigation information from only part of the investigated results. Of great importance, the neural network can reduce the investigation efforts and still obtain almost the same results as the original full investigation.
