17.01.2014 Views

Propositional Argumentation Systems and Symbolic Evidence Theory

Propositional Argumentation Systems and Symbolic Evidence Theory

Propositional Argumentation Systems and Symbolic Evidence Theory

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

44 4 ASSUMPTION–BASED SYSTEMS<br />

the same, which in general is not the case.<br />

Example 4.4 (Sequence of Binary Inverters): Let us now study how symbolic<br />

arguments can be helpful in the simple diagnostics problem of Example 4.2. First,<br />

we may be interested in sets of components which, intact or faulty, explain the<br />

observed behaviour of the system. Such explanations are called diagnoses. They<br />

can be represented by conjunctions of literals of the assumptions ok 1 <strong>and</strong> ok 2 . Here,<br />

the diagnoses are given by the support of the tautology,<br />

sp[⊤] = [(ok 1 ∧ ∼ok 2 ) ∨ (∼ok 1 ∧ ok 2 )]. (4.18)<br />

In this example, it is easy to see to that ok 1 ∧ ∼ok 2 <strong>and</strong> ok 1 ∧ ∼ok 2 are the only<br />

possible diagnoses. If a system of two inverters connected in series produces 1 for<br />

an input 0, it is clear that exactly one component is faulty <strong>and</strong> the other is intact.<br />

It may also be interesting to know sets of possibly intact or faulty components,<br />

which are in contradiction to the system description <strong>and</strong> the given observations.<br />

Such sets are called conflicts, <strong>and</strong> they are given by the quasi–support of the<br />

contradiction:<br />

qs[⊥] = ∼sp[⊤] = [(ok 1 ∧ ok 2 ) ∨ (∼ok 1 ∧ ∼ok 2 )]. (4.19)<br />

Obviously, ok 1 <strong>and</strong> ok 2 can neither be true nor false at the same time. Therefore,<br />

it is clear that ok 1 ∧ ok 2 <strong>and</strong> ∼ok 1 ∧ ∼ok 2 are conflicts, <strong>and</strong> that in this example,<br />

these are the only conflicts.<br />

4.4 Representing <strong>and</strong> Computing <strong>Symbolic</strong> Arguments<br />

In the previous section we defined an evidence theory based on propositional<br />

logic. It assigns arguments (represented by formulae in L A ) to hypotheses<br />

(represented by formulae in L N ) in the sense of the general evidence theory.<br />

Although all results of Chapter 2 remain valid for assumption–based systems,<br />

there are still some important questions to be answered, for example:<br />

(1) How are arguments obtained from the knowledge base <strong>and</strong> the given<br />

hypothesis?<br />

(2) How can arguments be represented or stored efficiently?<br />

(3) Is there a way to store efficiently all arguments for all possible hypotheses?

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

Saved successfully!

Ooh no, something went wrong!