04.11.2014 Views

elektronická verzia publikácie - FIIT STU - Slovenská technická ...

elektronická verzia publikácie - FIIT STU - Slovenská technická ...

elektronická verzia publikácie - FIIT STU - Slovenská technická ...

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.

186 Selected Studies on Software and Information Systems<br />

P 0 A 0<br />

P 1<br />

A 1 P 2<br />

room(R1)<br />

room(R1)<br />

room(R1)<br />

at(Eva,R1)<br />

at(Eva,R1)<br />

at(Eva,R1)<br />

at(Adam,R1)<br />

at(Adam,R1)<br />

open(Adam,R1)<br />

at(Adam,R1)<br />

¬ opened(R1)<br />

open(Adam,R1)<br />

¬ opened(R1)<br />

move(Eva,R1,R2)<br />

¬ opened(R1)<br />

man(Adam)<br />

man(Adam)<br />

opened(R1)<br />

move(Adam,R1,R2)<br />

man(Adam)<br />

opened(R1)<br />

at(Eva,R2)<br />

at(Adam,R2)<br />

Figure 6-13. Planning graph example.<br />

The following actions are available:<br />

open() Parameters: ?r, ?who Preconditions: (room ?r), not(opened<br />

?r), (man ?who), (at ?who ?r) Effects: (opened ?r)<br />

move() Parameters: ?who, ?from ?to Preconditions: (at ?who ?from),<br />

(room ?from), (room ?to), (opened ?from) Effects: (at ?who ?to),<br />

not (at ?who ?from)<br />

In this example, the goal (Eva and Adam are in room R2) is achieved in the last, second layer.<br />

In [19] a mutex relation (Mutual Exclusions) between planning graph nodes is defined.<br />

The possible mutex relations between two action nodes are: inconsistent effects – one action<br />

negates the effect of the other, inference – one of the effect of one action negotiates a precondition<br />

of another, competing needs – one of the precondition of one action is mutually exclusive<br />

with the precondition of the other. In [19] also GRAPHPLAN is presented. It is an algorithm<br />

consisting of two stages altering in a loop: graph expansion and solution extraction. Graph<br />

expansion grows the graph until the reached proposition level includes goal propositions<br />

without mutex links. In solution extraction stage, backward-chaining strategy is used to look<br />

forward a potential plan.<br />

Backward Chaining<br />

One of the simplest methods used in web service composition is backward chaining. Its<br />

basic principle is to begin with the goal and search for services which produce its parts.<br />

This is used to reach the goal propositions and also the output. We take the set of goal<br />

propositions and look for services which produce them as effects. These services may have<br />

preconditions which are not satisfied. For this set of unsatisfied propositions we again<br />

look for services which produce them. This backward search is performed until we have<br />

service with unsatisfied preconditions. Similarly, this method is applied to get the required

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

Saved successfully!

Ooh no, something went wrong!