The results of research on string internationalization (abbreviated as i18n) in software are presented in this paper. It begins by defining the problem: strings in a graphical user interface (GUI) must be translated into many languages (along with standard usages for locales, such as the words for measures or date formats).

Today, professional programming addresses the problem from the beginning, by not hard-coding any string that is visible to the user. However, older programs and some open-source programs often contain such strings. Therefore, there is a need to locate all strings in a program that require i18n.

The proposed method retrieves such strings by collecting the methods from the GUI application programming interface (API) and finding the problematic strings that they use (hence the name “string-taint analysis”). Once a string is found, it is marked for i18n. The same method is also used for strings that are transmitted over a network.

Wang et al. give examples of such strings in Java, and the processing algorithm is described in pseudocode.

The method has been applied to several large systems, and the paper includes a section on evaluating the results, the analysis of false findings, and the efficiency of the algorithm. The paper concludes with a useful evaluation of the method and a comparison with other published techniques. The reference list contains 31 papers and books.

The paper covers the important problem of adapting software for use in different countries with different languages. The method is interesting and well described. As the authors note, there is still more to do to improve the validity of the method.