Although various methods for inputting Chinese characters exist, phonetic input is one of the simplest and easiest to use. We propose a new approach that combines a perfect hashing function, data compression, clustering analysis and self-organizing sequential search heuristics using Shuang Pinyin symbols. Experimental results show that our scheme is feasible. The memory space required by our scheme is almost identical to the size occupied by the existing phonetic symbols. Once a specified Chinese character is retrieved, we reorder the table of homophonetic characters by interchanging with the character that immediately precedes it; thus more frequently accessed characters are moved to the front whereas less frequently accessed characters are moved to the rear. In comparison with other Chinese character retrieving schemes proposed previously, our method is the most efficient in terms of time consumption and space occupancy.