Social research data are often distributed with the research metadata such as research areas, definitions of research concepts, layouts of physical data, etc. The collection of these meta information for a research is called a codebook. Now that we know the usefulness of electric data of a research, why don't we computerize the codebook, too? In this paper, we discuss standards for describing a codebook, and apply DDI 3 , a new XML-based candidate of the standard, to the codebook of the Fishery Census of Japan. We argue what we need to put DDI 3 (or other candidates) to practical use in social sciences.