Tag, path and weighted tag
As mentioned in previous article, there are pros and cons for either the path or tag approaches in classifying information.
A problem with tags is that a single tag may associated with a lot of information that may or may not related to the specific interest of a particular search incidence. For example, a 'statistics' tag may associated with education, mathematics, physics, program ... etc. A single search of statistics becomes useless.
A possible approach is to make the search a tree/branch-liked search: Once a tag-of-interest is identified, the search program will create a list of data items that are tagged with the searching tag. The program will, then, collect all tags from these data items and try to present choices for the next tag selection and so on. By implement it right, tags can be considered a dynamic tree/path structure.