Extended ER Features

Basic ER diagrams handle simple entities and relationships. But real-world systems often require more advanced modeling concepts.

The Enhanced Entity-Relationship (EER) model adds three important features: Generalization, Specialization, and Aggregation. These allow database designers to represent inheritance, classification, and abstraction.

---

Learning Objectives

After completing this chapter, you will be able to:

• Explain Generalization and Specialization.
• Differentiate between top-down and bottom-up approaches.
• Understand Disjoint vs Overlapping constraints.
• Explain Aggregation and when to use it.
• Represent ISA relationships in ER diagrams.
• Answer interview questions on EER concepts.

---

Specialization (Top-Down)

Specialization is the process of breaking a higher-level entity into lower-level entities based on distinguishing characteristics.

Approach: Top-down. Start with a general entity and split it into specific subtypes.

Example

           Account
          /       \
  Savings_Account  Current_Account

Account has common attributes: Account_No, Balance, Open_Date.

Savings_Account adds: Interest_Rate, Minimum_Balance.

Current_Account adds: Overdraft_Limit, Transaction_Fee.

When to use: When you start with a general concept and discover specific subtypes.

---

Generalization (Bottom-Up)

Generalization is the reverse — identifying common attributes in multiple entities and creating a higher-level entity.

Approach: Bottom-up. Start with specific entities, find commonality, and abstract upward.

Example

  Car         Truck        Motorcycle
    \           |           /
     \          |          /
        Vehicle (Generalized Entity)

Car, Truck, and Motorcycle all share: Vehicle_ID, Make, Model, Year.

These common attributes are generalized into a Vehicle entity.

When to use: When you have existing specific entities and want to extract commonality.

---

ISA Relationship

“ISA” stands for “is-a” — a Savings Account ISA Account. It is the database equivalent of Inheritance in Object-Oriented Programming.

Representation in ER diagrams:

       +-----------+
       | Account   | ← Rectangle
       +-----------+
          |
          ◊ ISA ◊ ← Triangle (ISA)
         / \
        /   \
+----------+ +-----------+
| Savings  | | Current   |
+----------+ +-----------+

---

Constraints on Specialization/Generalization

Disjoint vs Overlapping

Constraint	Meaning	Example
Disjoint	Entity belongs to at most ONE subtype (d)	A student is either Grad or Undergrad, not both
Overlapping	Entity can belong to MULTIPLE subtypes (o)	A person can be both Student AND Employee

In ER diagrams:

• Disjoint: d inside the ISA triangle
• Overlapping: o inside the ISA triangle

Total vs Partial

Constraint	Meaning	Example
Total (double line)	Every supertype entity MUST belong to a subtype	Every Employee must be either Manager or Worker
Partial (single line)	Some supertype entities may not belong to any subtype	Some Account may be neither Savings nor Current

---

Aggregation

Aggregation treats a relationship between two entities as a single entity so it can participate in another relationship.

The Problem

Standard ER doesn’t allow a relationship between two relationships. Sometimes you need to model that.

Example

Employee ─── Works_On ─── Project
                                │
                                │ Uses
                                │
                              Tool

A relationship between Works_On (relationship) and Tool (entity) is needed: “An employee working on a project uses a specific tool.”

Solution: Aggregation

┌──────────────────────────────┐
│ Employee ── Works_On ── Project │  ← Treated as ONE entity
└──────────────────────────────┘
                │
                │ Uses
                │
              Tool

The Works_On relationship is aggregated into a higher-level entity, which can then participate in the Uses relationship with Tool.

---

Mapping EER to Relations

EER Concept	Mapping Strategy
Specialization (Disjoint)	Create one table per subtype with PK from supertype
Specialization (Overlapping)	Create one table per subtype with PK from supertype
Generalization	Create supertype table + subtype tables (same as specialization)
Aggregation	Create a table for the relationship, then link the aggregated relationship to the third entity

---

Summary Table

Feature	Approach	Direction	OOP Equivalent
Specialization	Top-Down	General → Specific	Subclassing
Generalization	Bottom-Up	Specific → General	Superclass extraction
Aggregation	Abstraction	Relationship treated as entity	Composition

---

Interview Deep Dive

---

Key Takeaways

• Specialization breaks a general entity into specific subtypes (top-down).
• Generalization combines specific entities into a general type (bottom-up).
• ISA represents inheritance in ER diagrams, like OOP inheritance.
• Disjoint — entity belongs to one subtype; Overlapping — entity can belong to multiple.
• Total participation — every supertype entity must be in a subtype.
• Aggregation treats a relationship as an entity so it can participate in other relationships.
• EER features make ER diagrams powerful enough to model complex real-world scenarios.