Mystery Number Four - Computational Linguistics and Talking Robots - page 76

The semiformal notation represents propositions by the core values of their
proplets, e.g., hungry ,andtheir prn value, e.g., 211 . Repeated core values
referring to the same item are shown as addresses, e.g., (food 212) , ensuring
coreference between contents throughout the chain.
The chain is activated by a trigger situation, represented by the content moi
be hungry 211 , which matches the antecedent of the chain-initial reactor
inference (shown in more detail in 5.2.3). The new content moi eat food 212 ,
derived by the consequent, serves as the trigger content for the antecedent of
the second inference (cf. step 2 in 5.2.1), and similarly for the remainder of
the chain. The E inferences with the connective exec derive a sequence of
blueprints for action, intended as a countermeasure.
The proplets derived by the completor consequent 211+7 , i.e., moi (eat
212) (food 212) 218 , are added to the Word Bank at the current end of the
respective token lines (i.e., the now front , cf. 4.1.1). Despite their prn value
218 , they are equivalent to the proplets derived earlier by the consequent of
the initial R inference, i.e., moi eat food 212 , due to their definition as the
indexical moi and the pointer proplets (eat 212) and (food 212) .

5.3 DBS Inferences for Meaning and Event Relations

The derivation of new content by means of DBS inferences may be used to for-
malize traditional relations of meaning. Standard examples from lexicography
are synonymy, antonymy, hypernymy, hyponymy, meronymy, and holonymy .
Consider the following example of a D inference implementing a synonymy:

5.3.1 I NFERENCE RULE IMPLEMENTING A SYNONYMY
⎡ ⎤ ⎡ ⎤

noun: abstract
fnc: α
prn: K

noun: summary
fnc: α
prn: K+M

⎣ ⎦ impl ⎣ ⎦ where

α {write, read, discuss, ...}

According to this rule, John wrote an abstract , for example, implies that
John wrote a summary . The restriction on the variable α specifies likely

verbs, obtained from a corpus.
Next consider an example of a D inference implementing an antonymy:

5.3.2 I NFERENCE RULE IMPLEMENTING AN ANTONYMY
⎡
⎣

⎤
⎦ impl

⎡
⎣

⎤
⎦

adj: good
mdd:

adj: not bad
mdd: α

prn: K

α
prn: K+M