Keyboard optimization using genetic techniques

Abstract

The problem of optimizing a keyboard for a particular finite dictionary, H, defined as a subset of the words over a finite alphabet, A is considered. The letters of A are assigned to elements of a set K. Thus, associated with every element i in K is a set C/sub i/ such that the (C/sub i/) partitions A. The aim of the optimization problem is to compute the set (C/sub i/) so that if every character in A is replaced by the index of the set in which it belongs, the transformed version of H has the minimum number of collisions. Initially, the problem is stated to be NP-hard. Later, the authors discuss the power of using genetic techniques to tackle the problem. The authors present the details of the only reported evolutionary method, and then a novel genetic solution is proposed. Experimental results demonstrating the power of this scheme are included. >