As all the society becomes computerized, there increases computerized data, and for digital forensic investigations, there is a great deal of unfixed-form data collected, whose exact forms are difficult to figure out, such as physical memory or page files. The most efficient method for investigating unfixed-form data is to extract strings. In case of document files, strings extracted from unfixed-form data come to include contents of the relevant documents, and in case of physical memory or page files, they can even include passwords that users have entered in addition to traces of users' using a messenger or a web page. Although extracting strings plays an important role in investigating unfixed-form data like this, the present method of extracting string includes a number of meaningless strings, while being carried out without considering the Unicode environment properly. Accordingly, this thesis intends to suggest a way of excluding meaningless strings effectively while considering the Unicode environment during the process of extracting strings.