It frequently happens that large amounts of data with various formats on computers are shared by multiple users. However, a method of accessing these data efficiently has not been established. This paper considers a browsing‐oriented digital library in which shared data are arranged linearly according to their relationships, and formulates a method of extracting the desired data efficiently. For the data extraction, a browsing method is first described in which relevance possibility is suggested on the basis of the content information and the position information in the directory structure. Then a method is described in which efficient browsing is realized in terms of the relevance judgment cost and the retrieval effectiveness by performing extraction and rearrangement of data with the directory as a unit. The effect of the proposed method was verified by an experiment using data in a university laboratory. A navigator for browsing the shared data has also been designed. © 2007 Wiley Periodicals, Inc. Syst Comp Jpn, 38(14): 79–89, 2007; Published online in Wiley InterScience (http://www.interscience.wiley.com). DOI 10.1002/scj.20265