Data Management Class Notes for 24/2/2014

ER Diagrams

ER (entity relationship) diagrams, like UML, ODL

• Entities (Square) Ex. Lugano, Person, Robert
• Entities granularity: all entities of the same "type" form an entity set
• Attributes (Circle) Ex. for Lugano you can have population, position
• Attributes: integers, string, etc, NULL
• Attributes can be: Base(DoB) or derived(age), simple(DoB) or composite(Address) and single valued(DoB) or Multiple valued(0 or more values)
• Relationship (Diamond) Ex. Likes between Robert and Lugano

Sets and subset

• S = {1, 2, 3, 4}
  R = {1, 2}
  R is a subset of S

Keys

• SUPERKEY: subset of attributes that uniquely identify an entity
• A weak superkey is a subset that is not minimal
• A primary key is a minimal subset of attributes to identify an entity (in SQL, UNIQUE)

Relationship

• R is a relationship among entity sets E1, E2, ..., En iff R is a subset of E1xE2x...xEn
• E1 = {A, B} E2 = {C, D}
  E1xE2 = (A, C), (A, D), (B, C), (B, D) R = {(A, C), (B, C)}
• Example:
  • Person = {Rovet, Steven}
  • Computers = {Apple, Windows}
  • Relationship Buys{Robert, Apple}

More on relationships

• Likes: Bob likes Alice
• Likes is a subset of cartesian product of all pairs of the form (person, person)
• Likes is an arbitrary directed graph in which person serves as the vertices and the edge are the relation
• Buys(Person, Computer, Company)
• Buys(Vendor, Product, Vendor)
• Ex. Dell buys a printer from HP
• Relationship can be one to one, many to one, one to many
• You can add an attribute to an entity when the relationship is one to one. Many to one you can go to the many

Strong VS Weak Entities

• A strong entity set is a set of elements that can be identified entirely by their attributes
• A weak entity set can not be identify by their attributes

Discriminant

• An attribute becomes the discriminant when a weak entity needs that attribute to be identify by a strong entity
• The thick line represent the path to identify a weak entity from a strong entity

ISA

• ISA means "it's a" relation
• Different types of ISA relations
• disjoint: no entity can be in more than one subclass
• overlapping: an entity can be in more than one subclass
• total: every entity has to be in a subclass
• partial: an entity does not have to be in a subclass
• Instead of ISA you can put the first letter (Ex:O, P for overlapping and partial)

Cardinality Constrains

The numbers on the connections represents the cardinality in the relation.