Smart identification : IF recorded description of object  
does't not identifies the object at run time ,if we enables
(Check the check box) smart identification thrhough object 
identification in tools, QTP identifies the object using 
smart identification

I think the above answer was close...

Smart Identification is a mechanism provided by QTP and 
used for identifying the objects smartly using Base filter 
properties or Optional filter properties when ever the 
normal identification fails due to the dynamic changes in 
the properties of objects.

i will give you a small example :

Suppose today u recorded two images of a different baby at 
8 yrs old. after 20 yrs, if u want to compare with the 
lastest images u need to be specify some basic filters like 
sex ......etc

Object identification:
QTP will identify the object in the following manner during 
the running.
It will under stand the script statement.
Then it releases that it needs to from some action on some 
object, for that it needs some information about that 
object.
For that information it will go to the object repository 
and get the information from their.
With that information it will try to identify that object. 
If it all the object is identified it will perform action 
on that object. Here the information means object 
properties.
And now the question is that how it is learning those 
properties, to identify the objects.
This is the point where every one thinking his one style, 
but the fact is that. That the process of learning is like 
this.
 There are two types of object identification that QTP 
normally uses apart from the ordinal identifiers.
In general there are four types of properties that QTP 
having.
1.	Mandatory properties.
2.	Assistive properties.
3.	Base filter properties.
4.	Optional filter properties.
And ordinal identifiers like location, index and creation 
time.
As I told that QTP will be using two types of object 
identification.
The first one is normal identification.
In this identification the learning of the properties will 
be in the following way.
First the QTP will learn all the mandatory properties at a 
time. And with these properties it will try to identify the 
object if at all it feels these properties are sufficient 
to identify the object uniquely, and then it will stop 
learning and use those properties. If at all it feels these 
properties are not sufficient to identify the object 
uniquely, then it will learn the first assistive property. 
And try to identify the object with those properties 
(mandatory and first assistive property), if at all it 
feels that these properties are sufficient to identify the 
object uniquely then it will stop learning and use those 
properties to identify the object. If it feels these 
properties are not sufficient to identify the object 
uniquely. Then it will learn the second assistive property 
and repeat the same process till it identify the object 
uniquely. If at all it fails to identify the object 
uniquely by using all these properties also. Then the 
ordinal identifiers come into picture. This is the process 
QTP identifies the objects when the smart identification is 
disable.
The second type of identification is smart identification.
In the process QTP will be learning all the mandatory, base 
filter properties and optional filter properties at time. 
But the identification process is as follows.
If you invoke the smart identification then also first QTP 
will try to identify the object by using normal 
identification process. That is the above process apart 
from the ordinal identifier. Now it will not use the 
ordinal identifiers. If the normal identification fails 
then it will come to the smart identification. And forget 
about the normal identification also all those properties 
learnt in the normal identification. And now it will using 
all the base filter properties at time and tries to 
identify the object uniquely. If at all it feels that these 
are sufficient then it will use these properties to 
identify the object uniquely. Otherwise take the first 
optional filter property and same process will continue as 
above till the object is identified uniquely. If at all the 
smart identification is also fails then it will go to use 
the ordinal identifiers. This the actual process of 
identification. If we have any questions regarding this 
please go head and send me the e-mail tome or directly you 
can contact me. only in the mentioned hours only.(8.00 PM 
TO 8.00AM)
                                                            
                       Parusuram challa,
                                                            
                     pramchalla@gmail.com
                                                            
                         (+91)9822769949

QTP uses SmartIdentification when QTP fails to recognise an 
object using Mandatory and Assretive properties. 

The secret behind the smart identification is: Even though 
the object properties have been changed during a period of 
time, there will be some common points that still exist in 
both the versions of the object. Smart Identification 
catches that common property and recognizes the object. 
Smart identification is an important feature in QTP.

The Smart Identification mechanism uses two types of 
properties:
· Base Filter Properties—The most fundamental properties of 
a particular test object class; those whose values cannot 
be changed without changing the essence of the original 
object. For example, if a Web link's tag was changed from 
{A} to any other value, you could no longer call it the 
same object.
· Optional Filter Properties—Other properties that can help 
identify objects of a particular class. These properties 
are unlikely to change on a regular basis, but can be 
ignored if they are no longer applicable