THE LOGICAL VIEW
Logicalization
Hazırlayan
Yard.Doç.Dr. Zehra KAMIŞLI ÖZTÜRK
Logicalization
• So far, we have been concerned with how the current system
works.
• Before we can design a new and better system, we need to be
clear exactly what the current system does (without physcial constraints),
not how it does it.
• the outcome will be a logical model of the current system.
We try to predict how things would work if no physical
constraints existed.
SSADM
Logicalization Process
Focus on the Data Flow Diagrams
especially for the lowest level of the DFD
• DFD: use terms and processes created by the users
– These may not be entirely logical
– They may be plain daft
At the Medical Centre, for example, patient records are made up of handwritten
cards stuffed into a cardboard folder. These are then stored in huge filing cabinets,
each the size of a small garage. These records will still have to be stored in our
new system, but probably not in this way
SSADM
Logicalization Process
Focus on the Data Flow Diagrams
especially for the lowest level of the DFD
• DFD: use terms and processes created by the users
– These may not be entirely logical
– They may be plain daft
• At the Medical Centre, for example, patient records are made up of handwritten
cards stuffed into a cardboard folder. These are then stored in huge filing
cabinets, each the size of a small garage. These records will still have to be stored
in our new system, but probably not in this way
SSADM
Logicalization Process
1. Logicalize the data stores:
logicalization uses the data store/entity cross-reference
Cross reference
Physical
• The cr is physical
Logical
• We need to produce a logical version
SSADM
Logicalization Process
1. Logicalize the data stores:
logicalization uses the data store/entity cross-reference
– Each data store will be an entity, or a group of related
entities
– their labels will change from M1, M2, ... to D1, D2, ... – this
indicates that the label no longer signifies a manual data
store, just a data store.
– The exact nature of the data store is not important
in a logical DFD.
SSADM
Logicalization Process
1. Logicalize the data stores:
Each data store
Entitty
• physical stores like ‘Student lists’, ‘Medical Records’
and ‘Patient Records’ will all fall into the logical
‘Patient’ data store.
SSADM
Logicalization Process
2. Remove physical time dependencies
• this involves removing any processes which exist purely as a
result of physical time constraints or from convenience.
• For example, we have a process at the Medical Centre: 4.3.3
– Post pay claim. This process is purely a physical restraint and involves
the receptionists waiting until all the timesheets are completed before
physically posting the pay claim. In the world of the computer we
probably won’t be posting much at all. So logically we can remove this
process.
SSADM
Logicalization Process
2. Remove physical time
dependencies
SSADM
Logicalization Process
3 Logicalize the processes: processes in the physical DFDs tell us
how
physically a process is done and who does the process,
but in the logical DFDs this isn’t necessary.
• Logically, processes can be done anyhow by anyone.
• Logical processes will input data, read it, change it, store it or
delete it.
Anything to do with physical tasks like Receptionist: ‘Receive
and distribute payslip’ can be removed.
SSADM
Logicalization Process
3 Logicalize the processes
Physical DFD: who? How?
Logically, anyhow by anyone.
SSADM
Logicalization Process
4. Logicalize the data flows: any reference to forms, reports,
sheets etc. must be removed and replaced with the actual data
which flows around the system.
The actual data is modelled in the entity model.
Thus the data flows will be labelled with the entity used by the
process.
For example, a patient’s medical card will become simply ‘patient’ or ‘patient
details’ in our logical model. This is imprecise because the medical card will
only contain some of the data stored in the PATIENT entity, but to specify
which attributes the medical card actually contains would make the diagram
too cluttered.
SSADM
Simple steps in logicalization
1 Consider the data stores.
We could simply take the entity model and make each
entity a data store.
it might be worth considering whether to combine some data stores !
This is just a convenient way of representing the data stores; we
are not actually merging the entities.
SSADM
Simple steps in logicalization
1 Consider the data stores.
SSADM
Simple steps in logicalization
1 Consider the data stores.
Similarly, ‘Doctor’ and ‘Nurse’ are closely related and could be
lumped together to form one data store. We would be in big
trouble with Nurse Payne if we called this store ‘Doctor’, so we
had better call it ‘Doctor/Nurse’.
We could also put ‘Work Rota’ and ‘Expenses’ in here, as they
too refer to the doctors and the nurse.
‘Order’ and ‘Order Line’ could sensibly be referred to as the
‘Order’ data store and finally it might be easier to put ‘Referral’
and ‘Treatment’ in with ‘Patient’ as they all relate to the
patient.We might call this store ‘Patient’.
SSADM
Simple steps in logicalization
2. Produce the logical data store/entity cross-reference.
This is simply amatter of listing the entities and corresponding data
stores.
Each entity must appear in only one data store. The data stores are
labelled
‘D1’, ‘D2’ etc. on the logical DFDs.
3. Remove time dependencies.
There are a few of these in our physical DFDs – 4.3.3 ‘Post pay claim’
has already been mentioned, and 4.3.4 ‘Receive and distribute
payslips’ is not a process we would want to keep if we could erase it.
This is not because we don’t want to pay the doctors (though the GPC
has considered this approach). It is rather that it serves no logical
purpose: it just wastes time
by passing on data which could go directly to the doctors. So we can
remove some processes in this way.
SSADM
Simple steps in logicalization
4 Remove names and departments from the process boxes.
Since it does not matter, logically, who does what, we can take out these
labels from the process boxes.
5 Change processes so that they describe what is done to the data.
Where a process refers to any particular ways of doing things instead of
what is actually done, we need to change the process to make it logical.
SSADM
Simple steps in logicalization
6 Remove any physical references from the process box.
Any process names which refer to physical objects or describe
physical actions must be changed. If possible, make the objects
entities and make the actions one of:
create, update, edit, delete or something similar.
For example, process 4.2 ‘Collate staff timesheets’ might
become something like ‘Update work rota’, since the timesheet
information (hours worked per day) is contained in theWork
Rota entity in our entity model.
SSADM
Simple steps in logicalization
7 Make sure that the data flows refer to logical items of data
These will normally be entities as items of data are contained within entities
and not things like forms, reports, handwritten notes, or chewed up
pieces of card.
8 Data flows between processes should be removed.
Instead, the data should be directed into and out of a data store.
SSADM
Simple steps in logicalization
start to consider what
requirements the user has for the new system