27.03.2014 Views

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Three rules match with the expression in working<br />

memory: Rule 1, Rule 2 and Rule 3. I f we resolve<br />

conflicts in favour of the lowest-numbered rule, then<br />

Rule 1 will fire. This causes X to be bound to the value<br />

‘Customer belongs to Bank1’ and the premise of Rule 1<br />

to be placed in working memory (Fig 3). The system has<br />

thus chosen to explore the possible hypothesis that the<br />

customer belongs to Bank 1. Another way to look at this<br />

is that the system has selected an ‘Or’ branch in the And<br />

/ Or graph in Fig 1.<br />

We also note that there are two premises to Rule 1, both<br />

of which must be satisfied to prove the conclusion true.<br />

These are ‘And’ branches of the search graph<br />

representing a d ecomposition of the problem (finding<br />

whether the customer belongs to Bank 1) into two sub<br />

problems (finding whether the PhoneNo matches with<br />

PhoneNo in Bank1 <strong>Knowledge</strong> Base, and Address<br />

Information matches with Address in Bank1 <strong>Knowledge</strong><br />

Base). We m ay then fire Rule 4, whose conclusion<br />

matches with ‘Address Information matches with<br />

Address Information in Bank1 <strong>Knowledge</strong> Base’<br />

causing its premises to be placed in working memory as<br />

in Fig 4.<br />

At this point, there are three entries in working memory<br />

that do not match with any rule conclusions. Our expert<br />

system will, in this situation, query the user customer<br />

directly about these sub goals. If the customer confirms<br />

all three of these as true, the expert system will have<br />

successfully determined that the customer belongs to<br />

Bank1. In this case, the use case will return the balance<br />

of the customer from Bank1. In our example, had we<br />

failed to determine that the customer belongs to Bank1,<br />

we would have needed to back up to the top level and try<br />

Rule 2 instead.<br />

Let us now consider a tr ace of the reasoning process.<br />

Suppose our expert system has an explanation facility,<br />

and let us assume that the consultation has reached the<br />

stage represented by the working memory configuration<br />

in Fig 4.<br />

The following dialogue begins with the computer asking<br />

the customer about the goals present in the working<br />

memory.<br />

Is your Phone Number 12345678?<br />

Yes<br />

Is your Address po_box_19_AB_Avenue_Kolkata?<br />

Yes<br />

Is your Name prakash_kumar?<br />

Why<br />

Here the system responds with a literal presentation of<br />

the rule under question.<br />

It has been established that:<br />

The Address Information matches with that in Bank1<br />

<strong>Knowledge</strong> base,<br />

Therefore if<br />

The Name matches with that in Bank1 <strong>Knowledge</strong> base<br />

Then you are a customer of Bank1.<br />

Following the above trace, the deposit of c ustomer<br />

would be displayed.<br />

Because the customer answered yes to the first two<br />

questions, Rule 4 was able to fire, verifying the second<br />

premise of Rule 1. When the ‘why’ query was made, the<br />

system was trying to solve the first premise of Rule 1. It<br />

is to be noted that the explanation for the systems<br />

behavior is just a r estatement of the current rule under<br />

consideration. However, this provides an adequate<br />

317

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!