my first post, my apologies if this does not belong here, please refer me to the/a right site. So I am looking for a general form for a primary key given a table containing non-atomic repeated entries in one-or-more fields. Let the fields with repeated entries be F1, F2, F3 . I am guessing that any primary key must contain an ordered triplet in F1, F2 and F3 respectively , maybe also from some other field. Is this correct? I am thinking when I expand the table along F1, F2, F3 to avoid repeated entries, fields other than these three will necessarily repeat.
I am thinking of a table with employee_num, dep_code , emp_lname, emp_dependants, emp_education and emp_training , with the last 3 the only ones containing repetition. I would then expand the table into a larger one to avoid repeated entries. Then each multiple entry of F1 (emp_dependents in this table) would combine with each of the entries of F2, and with each entry of F3 , for a total of |F1|x|F2|x|F3| new entries, where |F1| is the number of repeated entries of F1, etc. , i.e., I need a new row for each dependent, each combined with a new row for each entry in education, etc.
EDIT: I guess I am asking whether there is a general formula for " first normalizing" (finding a PK) for a table with repeated entries.
So, would a triple from F1, F2, F3, the 3 rows with repeated entries give me a primary key for my table? Sorry I don't know how to upload diagrams; I don't even own a scanner. Thanks.
