Database Normalization Process to 3NF CustomerRental for CustNo, PropNo, OwnerNo, etc

问题

I am trying to normalize the following table. I want to go from the UNF form to 3NF form. I want to know, what do you do at the 1NF stage? It says it's where you remove the repetitive columns or groups (ex. ManagerID, ManagerName). This is considered repetitive because it's leads to the same data.

The Unnormalized data table has the following columns CustomerRental(CustNo,CustName,PropNo,PAddress,RentStart,RentFinish,Rent,OwnerNo,OName)

There are no repeating columns/fields and each cell has a single value, but there is not a primary key. The functional dependencies I see in the table are:

{CustNo}->{Cname}
{PropNo}->{Paddress,RentStart,RentFinish,Rent,OwnerNo,Oname}
{CustNo,PropNo}->
    {Paddress,RentStart,RentFinish,Rent,OwnerNo,OName,CustName}
{OwnerNo,PropNo}->{Rent,Paddress,Oname,RentInfo} 

The primary key I picked was a composite key, CustNo + PropNo. Since it has a primary key, the table is in 1NF form, correct? This is what I thought, but the answer excludes CustNo and CustName from the table. They are in their own table.

From the above, I normalized it 2NF. At this stage, you are supposed to ensure that all non-prime attributes are fully dependent on the primary key. This is not the case. These are the functional dependencies in the table:

{OwnerNo}->{Oname}
{CustNo}->{CustName}
{PropNo}->{Paddress,Rent,OwnerNo,Oname}

I moved these values out of the table to create three new tables in 2NF form:

Customers(CustNo(PK),CustName)
Property(PropNo(PK),Paddress,City,Rent,OwnerNo,OwnerName)
Rentals(RentalNo(PK),CustNo,OwnerNo,PropNo,RentStart,RentFinish)

Now the main table, Rentals, is in 2NF form. It has a primary key, RentalNo, and each of the non-prime attributes depends on it.

I think that there is a transitive dependency on it. You can find OwnerNo through the PropNo. So, to make it comply with 3NF rules, you have to move the OwnerNo to its own table to create these tables:

Customers(CustNo,CustName)
Property(PropNo,Paddress,City,Rent)
Owners(OwnerNo,OwnerName)
Rentals(RentalNo,CustNo,PropNo,RentStart,RentFinish)

Is this correct? I read that at the 1NF stage, you are supposed to remove repetitive columns (ex. OwnerNo,OwnerName). Is this true? Why or why not?

The picture showing my tables is here: Normalized Tables

回答1:

We don't normalize to a NF (normal form) by going through lower NFs between it and 1NF. We use a proven algorithm for the NF we want. Find one in a published academic textbook. (If that doesn't describe the reference(s) you were told to use, find one that it does & quote it.)

Pay close attention to the terms and steps. Details matter. Eg you will need to know all the FDs (functional dependencies) that hold, not just some of them. Eg whenever some FDs hold, all the ones generated by Armstrong's axioms hold. Eg PKs (primary keys) are irrelevant, CKs (candidate keys) matter. Eg every table has a CK. Eg normalization to higher NFs does not change column names. So already your question does not reflect a correct process.

You really need to read & quote the reference(s) you were told to use in order to get to "1NF", because "1NF" is in the eye of the beholder. Normalization to higher NFs works on any relation.

来源：https://stackoverflow.com/questions/45831937/database-normalization-process-to-3nf-customerrental-for-custno-propno-ownerno