Simple tags

The very simplest way to implement a metadata file system is to make it possible to provide a set of tags along with every file ⁷. The tag can be every thing, no limits or convention check are provided from the operating system. For instance if I was going to save this document in such a file system I would provide it with the following tags: {PDF, school work, nus, cs2106, metadata, report}. For the file system to be use full it should provide some kind of filtering/search facilities, so it would be possible to issue an enquiry on all files which have the tags, {report, nus} and a list containing all the reports written on NUS, which are stored in the file system should pop up. This functionality would be sufficient to provide the basic of a metadata file system. It would be simple to implement as all it require is some extra space in every file for the metadata and a set of hash tables, to speed up searching.

But this type of metadata file system is also limited in some ways, because it only operates using one data type (in this case string or blob⁸). This means that the search/filtering engine only are able to use the operators which are defined for this very data type. I will make my point clear in the following two examples: