Information Technology Reference
In-Depth Information
(1) It is now summer.
(SL ( ) ( ))
(2) The weather is very humid.
(SL (1) ( ))
In this example, IN-list and OUT-list of the SL justification of node (1) are all
empty, it meas that the justification of node (1) is always valid and therefore the
node (1) will always be an IN-node. We call nodes of this type as premise. IN-list
of the SL justification of node (2) is composed of node (1), it means that node (2)
is believed if node (1) is a IN-node. According to this example, we can see that
the inference of TMS is in fact similar to the inference of predicate logic.
Difference between them is that premises in the TMS can be retracted and
correspondingly the knowledge base can be revised.
Based on the above example, we add an item to the OUT-list of node (2) and
get the following SL justifications:
(1) It is now summer.
(SL ( ) ( ))
(2) The weather is very humid.
(SL (1) (3))
(3) The weather is very dry.
In this case, the condition for node (2) to be believed is that node (1) is an
IN-node and node (3) is an OUT-node. All of these SL justifications state that “if
it is now summer and there is no evidence to prove that the weather is very dry,
then it can be derived that the weather is very humid”. We call nodes whose SL
justification has a nonempty OUT-list as assumptions.
Each CP justification is of the following form:
(CP <consequent> <IN-hypotheses> <OUT-hypotheses>)
(2.22)
A CP justification is valid if (1) the consequent node is an IN-node, (2) each
node of the IN-hypotheses is IN-node, and (3) each node of the OUT-hypotheses
is OUT-node.
The set of hypotheses must be divided into two disjoint subsets, since nodes
may be derived both from some IN-nodes and some OUT-nodes.
