No, you're not overthinking it. What I may end up is the join table from Hell with a foreign key consisting of the 3 overriding categories, the class, and the function. The join table would hold the attributes common to only that combination.

>>category, a subcategory, a sub-sub category, then classed 4 ways.
>>attributes common to each level and, finally, attributes that will only be >needed with certain combinations of categories, subs, subs-subs and classes.
>
>Are the sub- and sub-sub- categories exclusive or common across parent categories? The answer will greatly affect how you model the data.
>
>Remember the Venn diagrams from math class? Conceptually, i'd think about each category as a set, and the shared attributes as a common subset. The problem is that you'll probably have so many categories and attributes that you'll need an incomprehensible hypercube to represent the whole mess.
>
>What kind of queries will they be attempting?
>
>One other point to ponder: Are we overthinking a simple solution?