Data Architecture Question - Different counts at depending on the hierarchical level being viewed

[mrichters]

The issue that I am currently working on involves a many to many relationship that connects a fact table to groups of objects which are arranged in a fixed level hierarchy. So to use foods as an example, my group of objects could include fruit, apple, and orange. Apple and Orange would be child objects in the fruit category. When looking at the individual object level(fruit and orange) I want to see the total number of references to apples and oranges, but when I rollup to the Fruit level I want to only see the count for the number of records that relate to fruit and its children or if the record relates to both fruit and orange or fruit and apple then I don't want to see the double count.

For my solution and I am trying to combine two solutions which I found in the Kimball Data Warehouse Toolkit (Fixed Depth Hierarchies and using Weighting Factors). In this case my fact table holds a Key to a bridge table which contains groups represented by the GroupKey, in the above case Food key, and the Weighting factor dependent on the number of objects that are in the group. The Food Key is in its own dimension with the attributes that describe it. This solution is currently not working and was hoping to get some suggestions to improve/fix it. Thanks.