- A data model is an abstraction that represents real-world objects and their relationships to help describe an organization's data requirements. It includes concepts for describing data, relationships between data, and constraints on the data. - Early data models included the hierarchical and network models, which used pointers to represent physical relationships between records. This led to issues like data redundancy and an inability to easily change relationships. - The relational model was developed to address limitations of earlier models by using logical relationships without pointers. It represented a significant improvement over previous approaches.

1. DATA MODELS
A model is a representation of "real world" objects and events, and
their associations. It concentrates on the essential, inherent aspects of an
organization and ignores the accidental properties. Actually, there isn't
really a data model "thing". Data models are abstractions, often times
mathematical algorithms and concepts. You cannot really touch a data
model. But nevertheless, they are very useful. A data model attempts to
represent the data requirements of the organization, or the part of the
organization that you wish to model. It should provide the basic concepts
and notations that will allow database designers and end-users to
communicate their understanding of the organizational data
unambiguously and accurately. The purpose of a data model is to represent
data and to make the data understandable.
A data model consists of a collection of tools for describing: data,
data relationships, data semantics and data constraints
Data model - an integrated collection of concepts for describing
data, relationships between data, and constraints on the data used by
an organization.
A data model can be thought of as comprising three components:
• a structural part, consisting of a set of rules that define how a
database is to be constructed;
• a manipulative part, defining the types of operations that are
allowed on the data (updating, retrieving data or changing the
structure of the database)
• possibly a set of integrity rules, which ensures that the data is
accurate
Thus, essentially a data model is a "description" of both a container
for data and a methodology for storing and retrieving data from that
container. The analysis and design of data models has been the
cornerstone of the evolution of databases. As models have advanced so
has database efficiency.
The main feature that differentiates a database from a collection of
traditional files is the existence of relationships between records regarding
objects or facts that had something in common. For instance, the record
that preserve data on a specific customer is related to records that store

2. data on the orders send by that customer and each order is related to
records that describe the products mentioned in order lines. On the other
side, more customers’ records may be related also to the record that holds
data on their sales agent. This complex set of relationships once frozen in
the database might be exploited to retrieve initial data in less time and
with considerable less programming effort.
The implementation of relationships is a technological matter
leading to the different database models emerged in the last 30 years. The
first attempt was to realize relationships between records at physical level.
The most known physical relationships are pointers - extra fields added to
the record and containing the address of the related record. The related
record could be accessed directly by making use of the pointer.
The pointer mechanism once set up, different data base models were
invented according to the relationships pattern. Among them, the
hierarchical data base model and the network database model, the two
most commonly used database models before the 1980's.
HIERARCHICAL DATABASE MODEL
As its name implies, the Hierarchical Database Model defines
hierarchically - arranged data. Perhaps the most intuitive way to visualize
this type of relationship is by visualizing an upside down tree of data. In
this tree, a single table acts as the "root" of the database from which other
tables "branch" out. The hierarchical database model use a tree pattern in
implementing relationships between records depicting different objects
Relationships in such a system are thought of in terms of children
and parents such that a child may only have one parent but a parent can
have multiple children. Parents and children are tied together by links
called "pointers". A parent will have a list of pointers to each of their
children.
This child/parent rule assures that data is systematically accessible.
To get to a low-level table, you start at the root and work your way down
through the tree until you reach your target. One serious problem is that
the user must know how the tree is structured in order to find anything.

3. Sales agent 1122
Customer 5543 Customer 6689 Customer 1122
order 123
order 145
product 144 product 553 product 337
Fig. 3.1. The hierarchical data model
The sales agent’s record at the root of the tree has pointers to
records of all customers he represents, each customer record has pointers
to all his orders records and each order record has pointers to all the
ordered products. The tree expands at lower levels with every new order
sent by a customer. The structure needs a lot of extra fields for each record
to accommodate the new emerging vertical relationships.
The hierarchical model however, is much more efficient than the
flat-file model because there is not as much need for redundant data. If a
change in the data is necessary, the change might only need to be
processed once. As we mentioned before, this flat file database would
store an excessive amount of redundant data. If we implemented this in a
hierarchical database model, we would get much less redundant data.
Consider the following hierarchical database scheme:
However, the hierarchical database model has some serious
problems. For one, you cannot add a record to a child table until it has
already been incorporated into the parent table (for instance, you can't add
a new customer if that customer is not represented by a sale agent). Also,
the hierarchical database model still creates repetition of data within the
database. Redundancy would occur because hierarchical databases handle
one-to-many relationships well but do not handle many-to-many
relationships well. This is because a child may only have one parent.
However, in many cases the child must be related to more than one parent.

4. Though this problem can be solved with multiple databases
creating logical links between children, the fix is very kludgy and
awkward.
NETWORK DATABASE MODEL
In many ways, the Network Database model was designed to solve
some of the more serious problems with the Hierarchical Database Model.
Specifically, the Network model solves the problem of data redundancy by
representing relationships in terms of sets rather than hierarchy. The
model had its origins in the Conference on Data Systems Languages
(CODASYL) which had created the Data Base Task Group to explore and
design a method to replace the hierarchical model.
The network model is very similar to the hierarchical model. In
fact, the hierarchical model is a subset of the network model. However,
instead of using a single-parent tree hierarchy, the network model uses set
theory to provide a tree-like hierarchy with the exception that child tables
were allowed to have more than one parent. This allowed the network
model to support many-to-many relationships.
Visually, a Network Database looks like a hierarchical Database in
that you can see it as a type of tree. However, in the case of a Network
Database, the look is more like several trees which share branches. Thus,
children can have multiple parents and parents can have multiple children.
The records at each tree level are related by horizontal links and form a
chained forward list that could be extended at the end of the chain with
new emerging records
Sales agent 1122
Sales agent 2233
Customer 5543 Customer 6689 Customer 1122
order 123 order 145
product 144 product 553 product 337
Fig. 3.2. The network data model

5. The vertical relationships (between records depicting different
entities) need only one pointer to reach the beginning of the chain of
related records. The horizontal relationships (between similar records)
need only one pointer to reach the next record in chain. An extra pointer
could be added to indicate the previous record, providing backward
chaining. The end of the record chain is indicated by a special stop value
for the pointer. The network model can be expanded easier with new
similar records at any level and the pointer number in each record remains
the same.
Nevertheless, though it was a dramatic improvement, the network
model was far from perfect. Most profoundly, the model was difficult to
implement and maintain. Most implementations of the network model
were used by computer programmers rather than real users. What was
needed was a simple model which could be used by real end users to solve
real problems.
Data accessing in data bases using physical pointers exploit the
chaining mechanism to retrieve related records. Special software support
must be provided for each database model to allow the user to extract data
without being very much aware of the internal organization of the
database.
A major inconvenient of physical relationships is that they depend on
the physical support of the database. Every time the database is
transported from one media to another, the pointers' values must be
updated. To overcome this inconvenient, a new technique in implementing
the relationships was invented: logical relationships.
The logical relationships are virtual relationships created between
records on the basis of a common field. The records are related at retrieval
time by matching records with the same value in the common field. At
storage time, the records are stored in separate files and checked to meet
relating criteria (values in the common fields to match existing values in
virtually related files).
Databases created with logical relationships store data easier but
require a lot of special software support to retrieve it. Also, the virtual
relationships lead to a lot of restrictions imposed to data at storage time to
ensure that the new entered record is truly related to the rest of the data
base.

6. THE RELATIONAL MODEL
The relational model - which implements logical relationships
between files in a database - was the first theoretically founded and well
thought out data model first proposed by E.F. Codd in 1970. The model is
based on branches of mathematics called set theory and predicate logic.
The basic idea behind the relational model is that a database consists of a
series of unordered tables (or relations) that can be manipulated using non-procedural
operations that return tables. This model was in vast contrast to
the more traditional database theories of the time that were much more
complicated, less flexible and dependent on the physical storage methods
of the data. It was the foundation of both database software and theoretical
database research ever since.
Relational data structure
The relational data model is based on the structures and mathematics
of relations. The term relation is a mathematical term which means a two-dimensional
table which is not homogeneous in its rows, i.e. , the number
of rows (unlike the number of columns) is not fixed. It is synonymous
with the term table, thus the table is not a fixed structure like a matrix or
an array which have fixed row and column dimensions, for the relation the
total number of rows can grow and shrink according to need.
In the relational model, we use relations to hold information about the
objects we want to represent in the database. We represent a relation as a
table in which the rows of the table correspond to individual records and
the table columns correspond to attributes. A row is also known as a tuple
(from quintuple, sextuple etc., a group of n elements is an n-tuple) and a
column an attribute. Each attribute has unique name and although it isn't
shown here the row order and column order are not significant. Each row
must also be unique.

7. Example: Table Customers
attributes (columns, fields)
Every value within a given attribute must be of the same type and the
collection of values for an attribute is known as a domain. A domain is
the set of allowable values for one or more attributes. The domain concept
is important because it allows us to define the meaning and source of
values that attributes can hold. As a result, more information is available
to the system and it can (theoretically) reject operations that don't make
sense.
Formally, given sets D1, D2, …. Dn a relation r is a subset of
D1 x D2 x … x Dn
Thus a relation is a set of n-tuples (a1, a2, …, an) where
each ai ∈ Di
Example: if
Customer_id={1111,1253,2121,1555}
Customer_name = {Jones, Smith, Curry, Lindsay}
customer_city = {London, London, Manchester, Reading}
balance = {500,200,600,300}
Then r = {(1111,Jones, London, 500),
(1253,Smith, London, 200),
(2121,Curry, Manchester, 600),
(1555,Lindsay, Reading, 300)}
is a relation over customer_id x customer_name x customer-city x
balance
The relation has the following properties:
tuples
(rows,records)
Customer_id Customer_name Customer_city Balance
1111 Jones London 500
1253 Smith London 200
2121 Curry Manchester 600
1555 Lindsay Reading 300

8. • Each entry in the table occurs only once (each row is unique).
• Each column is named
• All values of a given column are of the same type
• Column order is immaterial
• Row order is immaterial
A relational database consists of tables that are appropriately
structured. The appropriateness is obtained through the process of
normalization. So, we can define a relational database as being a
collection of normalized tables.
A relational table has the following properties:
• The table has a name that is distinct from all other tables in the
database.
• Each column has a distinct name.
• The values of a column are all from the same domain.
• The order of columns has no significance.
• Each record is distinct; there are no duplicate records.
• The order of records has no significance.
• Each cell of the table (field) contains exactly one value (first
normal form)
The terminology of the relational model can be quite confusing. You
can encounter terms like:
- for relation: table or file
- for tuple : row or record
- for attribute : column or field
Relational keys
Each record in a table must be unique; that means we need to be
able to identify a column (or combinations of columns) that provides
uniqueness.
Superkey - a column, or set of columns, that uniquely identifies a
record within a table.
Let K ⊆ R
K is a superkey of R if values for K are sufficient to identify a unique tuple
of each possible relation r(R)

9. by “possible r” we mean a relation r that could exist in the enterprise we
are modeling.
Example: {customer_id, customer_name} and
{customer_id}
are both superkeys of Customer, if no two customers can possibly have the
same identification number.
Since a superkey may contain additional columns that are not
necessary for unique identification, we're interested in identifying
superkeys that contain only the minimum number of columns necessary
for unique identification.
Candidate key - a superkey that contains only the minimum
number of columns necessary for unique identification.
K is a candidate key if K is minimal
Example: {customer_id} is a candidate key for Customer, since it is a
superkey (assuming no two customers can possibly have the same
identification number), and no subset of it is a superkey.
A candidate key has two properties:
1. Uniqueness : in each record, the values of the candidate key uniquely
identify the record
2. Irreductibility (non-redundancy): no proper subset of the candidate
key has the uniqueness property (no attribute in the key can be
removed without destroying property 1)
There may be more than one set of attributes which have both
properties, these are candidate keys, one of which will be the primary key
(the candidate key that is selected to identify uniquely records within the
table)
Thus, all columns (or combination of columns) in a table with unique
values are referred to as candidate keys, from which the primary key must
be drawn. All other candidate key columns are referred to as alternate
keys. Keys can be simple or composite. A simple key is a key made up of
one column, whereas a composite key is made up of two or more columns.
The decision as to which candidate key is the primary one rests in your
hands—there's no absolute rule as to which candidate key is best. Fabian
Pascal, in his book SQL and Relational Basics, notes that the decision
should be based upon the principles of minimality (choose the fewest
columns necessary), stability (choose a key that seldom changes), and

10. simplicity/familiarity (choose a key that is both simple and familiar to
users)
Usually the word key refers to the primary key which implies that
there are secondary keys. A secondary key is often used for speedy
retrieval of rows from a table.
There is another key called a foreign key - a column, or set of
columns, within one table that matches the candidate key of some table. In
other words, this is an attribute of a relation which identifies the primary
key of another relation. A foreign key is a column in a table used to
reference a primary key in another table.
It is important that both foreign keys and the primary keys that are
used to reference share a common meaning and draw their values from the
same domain. The foreign key permits the association of multiple
relations:
TableA (A1, A2, A3)
TableB (B1, B2, B3)
TableC (A1,B1,C1)
In TableC, attribute A1 is a foreign key of TableA and attribute B1
is a foreign key of TableB. Foreign keys make it possible to resolve many-to-
many associations between tables.
One of the advantages of the database approach was control of data
redundancy. This is an example of "controlled redundancy" -these
common columns in different relations play an important role in modeling
relationships. The foreign keys matching primary keys mechanism
implements relationships between tables that share common fields.

11. The example used to illustrate the hierarchical and network database
models is presented below in the relational approach:
CUSTOMERS
Customer_id
Customer_name
Customer_city
Balance
Creditlimit
Slsanumb
ORDERS
Order_nb
Order_date
Customer_id
PRODUCTS
Prnumber
Descrition
MU
Price
Status
Supply date
Figure 3.3. The relational data model
SALES
AGENTS
Slsanumb
Slsaname
Slasaaddr
Totcomm
Commrate
ORDER LINES
Order_nb
Prnumber
Quanyity
The common convention for representing a description of a
relational database is to give the name of each table, followed by the
column names in parentheses. Normally, the primary key is underlined
and foreign keys underlined with a dots line. In that example foreign keys
are italic.
SALES AGENTS(Slsanumb, Slsaname, Slasaaddr,Totcomm, Commrate)
CUSTOMERS(Customer_id, Customer_name, Customer_city, Balance,
Creditlimit, Slsanumb)
ORDERS (Order_nb, Order_date, Customer_id)
ORDER LINES (Order_nb, Prnumber, Quantity)
PRODUCTS (Prnumber, Description, MU, Price, Status, Supply date)
Besides the structure of data, the relational model also defines the
means for data manipulation (relational algebra and relational calculus)
and the means for specifying and enforcing data integrity (integrity
constraints).

12. Relational integrity
The relational model is very simple and efficient. Data are stored in
tables that emulate the well-known file concept and duplicated columns in
some tables that are to be related implement the virtual relationships. The
model simplicity is balanced by a lot of rules that must be imposed to table
structures and stored data to ensure the data precise retrieval. These rules
are known as integrity rules and normal forms.
Since every column (attribute) has an associated domain, there are
constraints (called domain constraints) in the form of restrictions on the
set of values allowed for the columns of tables. In addition, there are two
important integrity rules, which are constraints or restrictions that apply to
all instances of the database.
NULLS
Null represents a value for a column that is currently unknown or
is not applicable for this record. A null can be taken to mean "unknown".
It can also mean that a value is not applicable to a particular record, or it
could just mean that no value has yet been supplied (missing). Nulls are a
way to deal with incomplete or exceptional data. However, a null is not
the same as a zero numeric value or a text string filled with spaces; zeros
and spaces are values, but a null represents the absence of a value.
Therefore, nulls should be treated differently from other values.
INTEGRITY RULES
The relational model defines several integrity rules that, while not
part of the definition of the Normal Forms are nonetheless a necessary part
of any relational database. There are two types of integrity rules: general
and database-specific.
General Integrity Rules
The relational model specifies two general integrity rules. They are
referred to as general rules, because they apply to all databases. They are:
entity integrity and referential integrity.
Entity integrity
We know that a primary key is a minimal identifier that is used to
identify records uniquely. This means that no subset of the primary key is
sufficient to provide unique identification of records. If we allow a null

13. for any part of a primary key, we're implying that not all the columns are
needed to distinguish between records, which contradicts the definition of
the primary key.
The first integrity rule applies to the primary keys of base tables:
In a base table, no column of a primary key can be null
A base table is a named table whose records are physically stored
in the database (this in contrast to a view, a virtual table that does not
actually exist in the database but is generated by the DBMS from the
underlying tables whenever it's accessed).
The entity integrity rule is very simple. It says that primary keys
cannot contain null (missing) data. It's important to note that this rule
applies to both simple and composite keys. For composite keys, none of
the individual columns can be null.
Referential integrity
The second integrity rule applies to foreign keys.
If a foreign key exists in a table, either the foreign key value must
match a primary key value of some record in its home table or the
foreign key value must be wholly null.
The referential integrity rule says that the database must not contain
any unmatched foreign key values. This implies that:
• A row may not be added to a table with a foreign key unless the
referenced value exists in the referenced table.
• If the value in a table that's referenced by a foreign key is changed
(or the entire row is deleted), the rows in the table with the foreign
key must not be "orphaned."
In general, there are three options available when a referenced primary
key value changes or a row is deleted. The options are:
• Disallow. The change is completely disallowed.
• Cascade. For updates, the change is cascaded to all dependent
tables. For deletions, the rows in all dependent tables are deleted.
• Nullify. For deletions, the dependent foreign key values are set to
Null
Business rules
All integrity constraints that do not fall under entity integrity or
referential integrity are termed database-specific rules or business rules.
These type of rules are specific to each database and come from the rules

14. of the business being modeled by the database. It is important to note that
the enforcement of business rules is as important as the enforcement of the
general integrity rules. Without the specification and enforcement of
business rules, bad data will get in the database.
Business rules are rules that define or constrain some aspect of the
organization. Examples of business rules include domains, which
constrain the values that a particular column can have, and the relational
integrity rules. Another example is multiplicity, which defines the number
of occurrences of one entity that may relate to a single occurrence of an
associated entity. It's also possible for users to specify additional
constraints that the data must satisfy the user must be able to specify these
rules and expect the DBMS to enforce them. For example, in our example
database we have to model the following rules:
• Order date must always be between the date the business started
and the current date.
• Customer type field can take one of these values: new, regular,
preferential, doubtful
• For each product, status can be: available, in supply, finished.
• Credit limit value must be less then 1000000
• For preferential customers we apply a discount of 10% to ordered
value
• Orders from doubtful customers are not accepted
• The supply date will be specified only for products with status "in
supply".
The level of support for business rules varies from system to
system. We'll discuss the implementation of business rules in ACCESS
DBMS in chapter…
Operations with relations - relational algebra
In many respects a relation is like a set and many of the operations
that can be used with sets can also be used with relations. The relational
algebra is a mathematical language designed for specifying operations on
relations. The algebra is used to manipulate one or two relations as
operands to produce a third relation
The access to data stored in relational data base is done through a set
of elementary routines called relational operators acting like set operators
on the sets of records each table consists of.

15. The relational operators are basic data retrieval procedures that could
be applied to a file collection and produce a new file as result. It exists
eight relational operators:
UNION ∪
INTERSECTION ∩
DIFFERENCE −
CARTHESIAN PRODUCT X
SELECTION σ
PROJECTION π
JOIN
DIVISION ÷
The collection of tables and the relational operators form a
relational algebra (algebraic structure). The relational algebra provides a
collection of operations to manipulate relations (relational operators). It
supports the notion of a query, or request to retrieve information from a
database.
Relational operators
PROJECTION – a vertical subset of a relation. The resulting relation will
contain every tuple in the first table but only several columns.
Defined as
πA1, A2, …, Ak (r)
where A1, A2 are attribute names and r is a relation name.
Examples:
• relation r relation π X,Z (r)
X Y Z
a 15 10
a 25 10
b 30 10
b 50 25
X Z
A 10
B 10
B 25

16. The result is defined as the relation of k columns obtained by erasing the
columns that are not listed. Duplicate rows removed from result, since
relations are sets
• Customers
Cust_nb Cust_name Country City Bank_
acc
Credit
limit
111 England
222 Romania
333 USA
444 England
555 England
666 Romania
π (Customers) = Customers_finances
(cust_nb, Bank_acc, credit _limit)
π(Customers) = Delivery points
(Cust_nb, Country, City)
Generalized projection - extends the projection operation by allowing
arithmetic functions to be used in the projection list.
πF1, F2, …, Fn(E)
- E is any relational-algebra expression
- Each of F1, F2, …, Fn are arithmetic expressions involving constants
and attributes in the schema of E.
Example:
• Given relation credit_info(customer_name,limit,credit_balance), find
how much more each person can spend:
πcustomer_name, limit – credit_balance (credi_info)
SELECTION – a new relation is produced containing records of the first
relation that meet a given condition (selection criteria or selection
predicate)
Defined as:

17. σ p(r) = {t | t ∈ r and p(t)}
Where p is a formula in propositional calculus consisting of terms
connected by : ∧ (and), ∨ (or), ¬ (not)
Each term is one of:
<attribute> op <attribute> or <constant>
where op is one of: =, ≠, >, ≥. <. ≤
Examples:
• Relation r Relation σ X=Y∧Z>10(r)
X Y Z
a a 30
a b 20
d d 40
b a 15
• Customers
X Y Z
a a 30
d d 40
Cust_nb Cust_name Country City Street Credit
limit
111 - England
222 - Romania
333 - USA
444 - England
555 - England
666 - Romania
Selection is a horizontal subset of a relation (every column, but only
several rows).
σ (Customers) ⇒ English customers
Country = England
UNION – the basic process of concatenating two relations with the same
structure (the relations are compatible).
Defined as:
r ∪ s = {t | t ∈ r or t ∈ s}
For r ∪ s to be valid r and s must be compatible:
- r, s must have the same arity (same number of attributes)

18. - The attribute domains must be compatible (e.g., 2nd column
of r deals with the same type of values as does the 2nd column of s)
Examples:
• Relations r and s (compatible) relation r ∪ s
X Y
a 10
a 15
b 30
Last year customers ∪ This year customers = Customers
Last year customers
Customer no. Customer name Customer
address
Credit limit
111
222
713
514
This year customers
Customer no. Customer name Customer
address
Credit limit
213
555
777
222
713
Customers
Customer no. Customer name Customer
address
Credit limit
111
222
713
514
213
555
777
X Y
a 15
b 40
X Y
a 10
a 15
b 30
b 40

19. DIFFERENCE – records that belong to the first relation and not to the
second.
Defined as:
r – s = {t | t ∈ r and t ∉ s}
Set differences must be taken between compatible relations:
- r and s must have the same arity
- attribute domains of r and s must be compatible
Examples:
• Relations r and s Relation r - s
Relation s - r
X Y
a 10
a 25
b 30
• Last year customers - This year customers = Lost Customers
Customer no. Customer name Customer
address
Credit limit
111
514
• This year customers - Last year customers = New Customers
Customer no. Customer name Customer
address
Credit limit
213
555
777
X Y
a 25
b 15
X Y
a 10
b 30
X Y
B 15

20. INTERSECTION - the basic process of combining two compatible
relations and produce a new one containing common records to both initial
relations.
Defined as:
r ∩ s ={ t | t ∈ r and t ∈ s }
Assume:
- r, s have the same arity
- attributes of r and s are compatible
Note: r ∩ s = r - (r - s)
Examples:
• Relations r and s Relation r ∩ s
X Y
a 10
a 25
b 30
X Y
a 25
b 15
X Y
a 25
• Last year customers ∩ This year customers = Faithful customers
Faithful customers
Customer no. Customer name Customer
address
Credit limit
222
713
CARTESIAN PRODUCT of two relations – a new relation whose
records are every pair of the records of the first relation concatenated with
each record of the second relation. The new relation will have a number of
records equal to the first relation number of records multiplied by the
second relation number of records.
Defined as:
r x s = {t q | t ∈ r and q ∈ s}

21. - Assume that attributes of r(R) and s(S) are disjoint. (That is,
R ∩ S = ∅).
- If attributes of r(R) and s(S) are not disjoint, then renaming must be
used.
Examples:
Relations s and r Relation s x r
X Y
a 10
b 20
P Q R
a c 15
b d 30
c e 20
d c 18
X Y P Q R
a 10 a c 15
a 10 b d 30
a 10 c e 20
a 10 d c 18
b 20 a c 15
b 20 b d 30
b 20 c e 20
b 20 d c 18
• Faithful customers × Gifts = Gifts to customers
Cust_nb Gift_nb Description
222 × 1 x
713 2 y
Gifts to customers
Cust_n
b
Cust_name Address Gift_nb Description
222 - - 1 x
222 - - 2 y
713 - - 1 x
713 - - 2 y
The new table Gifts to customers has 2*2 = 4 records
DIVISION - the division is the reverse of Cartesian product when applied
on proper relations (the relation to be divided by another relation called

22. divisor is the Cartesian product of divisor and the quotient). The quotient
is the resulting relation of the division.
Let r and s be relations on schemas R and S respectively where
R = (A1, …, Am, B1, …, Bn) ; S = (B1, …, Bn)
The result of r ÷ s is a relation on schema R – S = (A1, …, Am)
r ÷ s = { t | t ∈ π R-S(r) ∧ ∀ u ∈ s ( tu ∈ r ) }
Example:
• Relations r and s Relation r ÷ s
X Y
a 10
a 20
a 30
b 30
b 10
c 10
c 15
b 20
d 25
e 10
Y
10
20
X
a
b
If r = s X q then q = r ÷ s or d = r ÷ q
If the dividing relation is not a complete Cartesian product, then
the result is the integer part of the quotient, meaning that the result of the
division is a set of records that may be encountered in the initial relation
fully concatenated with the divisor.
• Let’s suppose we have the table Cust_prod that presents all the pairs
cust_nb prod nb encountered in the orders lines ( every customer is
associated with all the products he ordered). We have also a customers
table and a products table. We are going to use projections on
important fields in every table.

23. CUSTPROD
Cust
nb
Cust
name
Prod
nb
Descript
1 C1 22 D2
2 C2 11 D1
3 C3 22 D2
2 C2 22 D2
2 C2 33 D3
CUSTOMERS
Cust
Cust
nb
name
1 C1
2 C2
3 C3
We want to find out which product was ordered by all the
customers stored in the customers table. This condition is met by each
prod_nb associated with all the cust_nb existing in the customers table..
The division between Cust_prod and Customers will give us the response.
CUSTPROD ÷ CUSTOMERS → Products ordered by all customers
Cust nb Cust name
2 C2
r ÷ s = πR-S (r) –πR-S ( (πR-S (r) x s) – πR-S,S(r))
- πR-S,S(r) simply reorders attributes of r
- πR-S(πR-S (r) x s) – πR-S,S(r)) gives those tuples t in πR-S (r) such
that for some tuple u ∈ s, tu ∉ r
JOIN – is applied on two relations that have similar attributes that could
be checked to have the same values. The resulting relation will contain
records of the first relation concatenated with records of the second
relation that meet a certain condition called join predicate expressed in
terms like:
value of a field of the first relation = value of a field of the second relation

24. In terms of relational algebra:
Let r and s be relations on schemas R and S respectively.
Then, r s is a relation on schema R ∪ S obtained as follows:
Consider each pair of tuples tr from r and ts from s.
If tr and ts have the same value on each of the attributes in R ∩ S, add a
tuple t to the result, where
- t has the same value as tr on r
- t has the same value as ts on s
Example:
R = (A, B, C, D)
S = (E, B, D)
Result schema = (A, B, C, D, E)
r s is defined as:
πr.A, r.B, r.C, r.D, s.E (σr.B = s.B ∧ r.D = s.D (r x s))
The join operator produces a larger record that could have fields
from both files. The number of records in the resulting file depends on
how many pairs could be made.

25. • Example
Relations r and s Relation r s
X Y W Z
a 10 e 13
b 20 f 16
c 15 g 20
d 18 h 18
• Customers Orders
Cu
st
nb
Cust
name
Y Q
10 p
25 m
15 n
20 p
Address Bank
account
111 C1 A1 Acc1
222 C2 A2 Acc2
333 C3 A3 Acc3
444 C4 A4 Acc4
Ord
nb
Ord date Cust
id
X Y W Z Q
a 10 e 13 p
b 20 f 16 p
c 15 g 20 n
Prod
nb
Q
1 111 457
2 222 890
3 111 123
4 222 457
5 222 890
6 333 234
7 555 890
Cust.
nb.
Cust.
Name
Addr
ess
Bank
acct
Ord.
nb.
Ord.
date
Cust.
id.
Prod.
nb.
Q
111 - - - 1 - 111 457 -
111 - 3 111 123
222 - 2 222 457
222 - 4 222 890
222 - 5 222 890

26. 333 6 333 890
According to the way the join predicate is formulated, there are several
kinds of JOINs :
EQUIJOIN - same value in fields with the same name in both tables
Customers.cust_nb=Orders.cust_nb
This join predicate is the logical expression of the
relationships between tables
foreign key = primary key
NATURAL JOIN – different fields names, the same value
Customers.Cust_nb=Orders.Customer_id
If the field Cust_id from orders would have been also Cust_nb then
the second column cust_nb from the new table disappears and the join is
called equi-join.
Cust
nb
Cust
name
Adr Bank
acc.
Ord.
nb.
Ord.
Date
Prod
nb.
Q
111 1 -
111 2 -
222 3 -
222 4 -
222 5 -
333 6 -
The equi or natural joins are called also INNER JOINS. They
present only records that meet the join condition.
OUTER JOIN - If the join condition is not compulsory, the records of
one relation may or may be not concatenated with a corresponding record
from the other relation. OUTER JOIN is an extension of the join
operation that avoids loss of information. It computes the join and then
adds records form one relation that do not match records in the other
relation to the result of the join. Records with no correspondent in the
other relation will be concatenated with a blank record (made of null
fields).

27. Nulls:
• It is possible for tuples to have a null value, denoted by null, for some
of their attributes. Null signifies an unknown value or that a value does
not exist. The result of any arithmetic expression involving null is null.
• All comparisons involving null are (roughly speaking) false by
definition.
• Comparisons with null values return the special truth value unknown
If false was used instead of unknown, then not (A < 5)
would not be equivalent to A >= 5
• Three-valued logic using the truth value unknown:
- OR: (unknown or true) = true,
(unknown or false) = unknown
(unknown or unknown) = unknown
- AND: (true and unknown) = unknown,
(false and unknown) = false,
(unknown and unknown) = unknown
- NOT: (not unknown) = unknown
This kind of outer join depends on which table is supposed to be taken
entirely: LEFT JOIN , RIGHT JOIN or FULL OUTER JOIN
LEFT JOIN – all the records of the left table concatenated with
corresponding records of the right table or with null fields
Customers Orders
Cust
nb
Cust
name
Adr Bank
acc.
Ord.
nb.
Ord.
Date
Cust.
id.
Prod
nb.
Q
111 1 - 111
111 2 - 111
222 3 - 222
222 4 - 222
222 5 - 222
333 6 - 333
444 null null null null null

28. RIGHT JOIN . All the records of the right table associated with
corresponding records of the left table or with null fields
Customers Orders
Cust
nb
Cust
name
Adr Bank
acc.
Ord.
nb.
Ord.
Date
Cust.
id.
Prod
nb.
Q
111 1 - 111
111 2 - 111
222 3 - 222
222 4 - 222
222 5 - 222
333 6 - 333
null null null null 7 - 555
FULL OUTER JOIN Customers Orders
Cust
nb
Cust
name
Adr Bank acc. Ord.
nb.
Ord.
Date
Cust.
id.
Prod
nb.
Q
111 1 - 111
111 2 - 111
222 3 - 222
222 4 - 222
222 5 - 222
333 6 - 333
444 null null null null null
null null null null 7 - 555
Relational calculus
The Relational Calculus is a formal query language. Instead of
having to write a sequence of relational algebra operations, we simply
write a single declarative expression, describing the results that we want.
to A specific relational query language is said to be relationally complete
if it can be used to express any query that the relational calculus supports.
There are two common ways of creating a relational calculus (both
are based on first order predicate calculus, or basic logical operators).
• In a Tuple Relational Calculus, variables range over tuples - i.e.,
variables can take on values of individual table rows. This is just what
we want to do a routine query, such as selecting all the customers

29. (tuples) from customers table where custmer_type (specific attribute)
is preferential (value).
• In a Domain Relational Calculus, variables range over domain values
of the attributes. This tends to be more complex, and variables are
required for each distinct attribute.
Both are nonprocedural query languages.
The relational operators may be used to form expressions to formulate
more complicated data processing. Even some relational operators might
be derived one from other using relational formula.
For instance, the result of the JOIN operator might be obtained if
we apply a selection with the join condition over the Cartesian product
between the two tables.
Customers Orders = σ (Customers × Orders)
(cust_nb=cust_id) (cust_nb=cust_id)
And the result of the CARTESIAN PRODUCT might be obtained
if we apply a join with a forever true condition on the tables.
Customers × Orders = Customers Orders
(cond)
The forever-true condition may be any condition met by all the records in
both tables.
Data Base Management Systems offers only some of the relational
operators (the easiest to implement essential operators) and the others
must be derived. There is however a minimal set of relational operators
from which all the others might be derived:
Selection, Projection and Join
The Join is the heart of relational algebra, the most important
relational operator. Given the fact that the join operator may be derived
from the Cartesian product, it exists an alternative set :
Selection, Projection and Cartesian product
We'll examine now the relational procedure used to derive the
other relational operators from the minimal set Selection, Projection and
Join
INTERSECTION – The set of common records of two tables with the
same structure is the same as the set of records produced by applying the
join operator with the condition that every field in the first table match the

30. value of the corresponding field in the second table. If the table has a
primary key, the join condition may be put only on that field only (equi-join).
Last year customers ∩ This year customers = Faithful customers
Last year customers This year customers = Faithful customers
Cust_nb
Or the intersection might be derived using a selection applied on
one table with a condition that the primary keys belong to a list of
primary keys belonging to the other table.( a projection of the second table
on the primary key)
σ (This year customers)
Cust_nb. in π (Last year customers)
Cust_nb
DIFFERENCE The difference between two tables might be obtained if we
apply a selection on an outer join of the two tables and exploit the null
fields
This year customers – Last year customers = New customers
Last year customers
Cust
nb
Cust
name
Address
111 C111 A111
222 C222 A222
713 C713 A713
514 C514 A514
This year customers
Cust
nb.
Cust
name
Address
213 C213 A213
555 C555 A555
777 C777 A777
222 C222 A222
713 C713 A713

31. RIGHT JOIN of Last year customers and This year customers:
Last Year
Customers
Cust nb.
This Year
Customers
Cust nb.
This Year
Customers
Cust name
This Year
Customers
Address
null 213 C213 A213
null 555 C555 A555
null 777 C777 A777
222 222 C222 A222
713 713 C713 A713
We select all the records with the Last year customers. Cust_nb = null
New customers
Last Year
Customers
Cust nb.
This Year
Customers
Cust nb.
This Year
Customers
Cust name
This Year
Customers
Address
null 213 C213 A213
null 555 C555 A555
null 777 C777 A777
The expression of the difference using an outer join:
New customers = σ ((Lastyear customers Right join This year customers))
Last year customers. Cust_nb = null
Using the same deductions:
Last year customers – This year customers = Lost customers
Lost customers = σ ( (Last year customers Left join This year customers))
This year customers. Cust_nb = null
LEFT JOIN of Last year customers and This year customers:
Last Year
Customers
Cust nb.
Last Year
Customers
Cust name
Last Year
Customers
Address
This Year
Customers
Cust nb.
111 C213 A213 null
222 C222 A222 222
713 C713 A713 713
514 C514 A514 null

32. We select all the records with the This year customers. Cust_nb = null
Lost customers
Last Year
Customers
Cust nb.
Last Year
Customers
Address
Last Year
Customers
Cust name
This Year
Customers
Cust nb
111 A111 C111 null
514 A514 C514 null
The Relational diagrams
The Relational Data Processing makes use of only the minimal set of
relational operators offered by the data base management system. To
express a complex task, a relational formula must be built up to reflect the
stream of relational operators that mimic the data flow that ultimately will
achieve the task. The formula is quite difficult to express, so a more
convenient layout is used, the data flow diagram.
As a general rule, we have to analyze the request for data made in
natural language and identify the relational operators or the stream of
relational operators we may apply on existing tables to produce the
required data. Most of them are joins, projections and selections. If a
condition is expressed using the prefix “un” (like unordered, unmentioned,
unsold products) then we’ll use the difference. If the condition is
expressed using the word “all”(like ordered all the products or ordered
by all the customers) then we’ll use the division. Attention must be paid
when in the condition is encountered the word “and” referring to different
entities
• Customers that ordered the product A and the product B (intersection)
• Customers that ordered product A and customers that ordered the
product B (union). We may reformulate: customers that ordered
product A or product B.
The best approach is to analyze the database and give a meaning to
every elementary relational operator applied on two tables. Not any two
tables may be united through an relational operator. Tables that do not
have any common field can be used only in Cartesian products. Take for
instance a collection of three tables:

33. CUSTOMERS (Cust_nb, Cust_name, Address, Bank_account)
ORDERS(Ord_nb, Ord_date, Cust_nb, Prod_nb, Quantity)
PRODUCTS(Prod_nb, Descript, Meas_unit, Price_unit)
All the requests one can formulate must contain words linked to
table names or field names. Apart from all kind of selections, the
following requests are the most likely to be made:
Ordered products ; Unordered products
Ordering customers; Un-ordering customers
Customers that order all the products
Products ordered by all the customers
CUSTOMERS ORDERS PRODUCTS
ORDERING
CUSTOMERS
ORDERD
PRODUCTS
_
UNORDERING
CUSTOMERS
_
UNORDERED
PRODUCTS
ORDERS EXTENDED WITH DATA
ON CUSTOMERS AND PRODUCTS
/
CUSTOMERS THAT
ORDERED ALL THE
PRODUCTS
/
PRODUCTS
ORDERED BY ALL
THE CUSTOMERS
The process diagram above is the basis for any complicated request
involving specific criteria like:
- Customers that ordered all the products in the category “xxx”
- Products ordered by all the customers from New York
- Unordered products in the current month

34. - Unordering customers in the current month.
(We add specific selections on the appropriate files from the diagram )
Relational languages
The two main languages that have emerged for relational DBMS are
SQL (Structured Query Language) and it's graphical front-end, QBE
(Query By Example).
SQL is both a Data Definition Language (DDL) and a Data
Manipulation Language (DML). As a DDL, it allows a database
administrator or database designer to define tables, create views, etc. As a
DML, it allows an end user to retrieve information from tables. SQL has
been standardized by the International Organization for Standardization
(ISO), making it both the formal and de facto standard language for
defining and manipulating relational databases.
QBE is an alternative, more intuitive to use, "point-and-click" way of
querying the database, which is particularly suited for queries that are not
to complex, and can be expressed in terms of a few tables.
The basic principle of the relational model is the Information
Principle: all information is represented by data values. Thus, the records
are not related to each other at design time: rather, designers use the same
domain in several field's descriptions, and if one attribute is dependent on
another, this dependency is enforced through referential integrity.
Advantages of the relational model:
• It is extensively studied, proven in practice, and based on a formal
theoretical model. Almost all of the things that are known about it
are actually proven as mathematical theorems. The data
manipulation paradigm is based on first order logic
• It offers an abstracted view of data. It was among the first major
application of abstraction as a way to manage software complexity.
It basically abstracts the physical structure of data storage, from
the logical structure of data.
• It offers a declarative interface (relational calculus) for the
specification of data manipulation, that is actually translated to an
efficient (sometimes the most efficient) implementation, given a
physical data layout and within reasonable heuristic limits.

35. The major disadvantage of the relational model: it's never been
fully, faithfully implemented. A relational database as implemented today
(with tables, rows, SQL as query language) is much more complicated and
less powerful than what a database should be in the relational model.
Tables and rows aren't equivalent to relations and tuples, because SQL
doesn't support user-defined data types and because tables are bags, not
sets. What is good enough varies with the complexity of the problem we
are facing, and for some problems, the miss implementation of the
relational model by current SQL DBMSes becomes really annoying