Cosine similarity

I’m trying to calculate cosine between two text lists, say list 1 and list 2. and it is a one to many relationship between two lists. Following your documentation, list 1 is my “category” and list 2 should be the “dimension”. I kept everything else as default to compute the cosime. The issue is, the text order of the lists completely shuffled after I run analytics. i.e. I’m expecting “list1 entry 1” compared to “list 2 entry 1” but that is not the case here.

Any advice on how to solve this issue?