Abstract: A file is separated into three sections: a root, one or more fragments, and a fragment index. The root, which is uncompressed, maintains relationship information regarding fragments within the file. The fragments contain the actual data and are compressed, and the index, which is uncompressed, provides a list of all the fragments and, for example, an identification of any attributes associated with those fragments. Using this nomenclature, selective compression and decompression documents is possible.