The Relational Model and Normalization Questions and Answers

Answer: B

Correct answer: relation.

Table, relation and file name the same structure in three vocabularies, just as row, tuple and record do, and column, attribute and field. The relational terms are the formal ones and the ones in which the normal forms are defined, which is why they are worth using precisely even when a product’s documentation does not.

Answer: A

Correct answer: normal forms.

The normal forms are the successive classes of relation defined by which dependencies they still permit, each one eliminating a further source of insertion, deletion and update anomaly. Functional dependencies are what the definitions are stated in terms of, and referential integrity is a separate constraint between relations rather than a class of relation.

Answer: C

Correct answer: row of a table.

A tuple is a row, the record of the older vocabulary. Since no two rows in a relation may be identical and the order of rows carries no meaning, a relation is formally a set of tuples - which is why a query result has no inherent ordering unless one is requested explicitly.

Answer: B

Correct answer: False

Referential integrity requires only that a non-null foreign key match an existing primary key value. A NULL is permitted and simply means the relationship is unknown or does not apply, unless the column is separately declared NOT NULL.

Answer: A

Correct answer: True

A candidate key is a minimal set of attributes that uniquely identifies a row, and identifying the row fixes the value of every attribute in it - so the key functionally determines every other column. This is why candidate keys are determinants; the point of Boyce-Codd Normal Form is to insist that every determinant be one of them.

Answer: A

Correct answer: split the relation into two relations, each with a single theme.

A multivalued dependency problem arises when one relation holds two independent multivalued facts about the same key, forcing every combination of the two to be stored. Separating them into two relations, each carrying a single theme, removes the spurious combinations entirely - and this is exactly what putting the relation into fourth normal form means.

Answer: B

Correct answer: False

Rows are tuples, or records; it is the columns that are called fields, along with attributes. Getting the correspondence the wrong way round is easy, so it is worth fixing the two triples in mind: relation, table, file for the whole structure; tuple, row, record across; and attribute, column, field down.

Answer: B

Correct answer: False

Boyce-Codd Normal Form requires that every determinant be a candidate key, and says nothing about whether that key is composite. A single-attribute key is perfectly acceptable. The condition exists because a determinant that is not a candidate key allows the same fact to be stored in many rows, which is the source of update anomalies.

Answer: C

Correct answer: fourth normal form.

Fourth normal form adds the elimination of multivalued dependencies to the requirements of Boyce-Codd Normal Form. A multivalued dependency arises when one attribute determines a set of values rather than a single value, and two such independent sets in one relation force every combination to be stored - which is what splitting into separate relations removes.

Answer: B

Correct answer: False

Row order carries no meaning in a relation: a relation is a set of tuples, and a set has no inherent sequence. A query that needs its output in a particular order must request it explicitly, and any apparent ordering in a result set without such a request is an accident of storage and must not be relied on.