<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-0708">
  <Title>I I I I I I I I I I I I I I I I I I I Linking WordNet Verb Classes to Semantic Interpretation</Title>
  <Section position="4" start_page="62" end_page="64" type="concl">
    <SectionTitle>
5 Conclusions and Related Research
</SectionTitle>
    <Paragraph position="0"> We have described a method for defining predicates for WordNet verb classes and illustrated it by ann- null lyzing some major WordNet verb classes. The thematic roles of the predicates are connected to Word-Net ontology for nouns and to the syntactic relations that realize them. Our approach to building verb meaning is based on the decomposition of a predicate into subpredicates. Subpredicates inherit not only thematic roles, but also inferences as explained in (Gomez, 1996). Inferences will be defined for generic predicates and subpredicates subsuming a large class of verb forms. The final product will be a network of predicates linked by subsumption, and inferences, and connected to the WorcbNet ontology for nouns and to grammatical relations. As of this writing we have defined predicates for 70% of WordNet verb forms.</Paragraph>
    <Paragraph position="1"> We are using the term &amp;quot;inference&amp;quot; to refer to both entailments and non-monotonic inferences, also called defeasible inferences. The term &amp;quot;entailment&amp;quot; is used in the sense of analytic implication (Quine, 1960). A sentence, say sl, entails sentence s2 if s2 is true in every world in which sl is true. An example of entailment is &amp;quot;if Peter killed the roach, the roach is dead.&amp;quot; An example of non-monotonic inference is &amp;quot;if Peter likes apples, he eats them.&amp;quot; Of course, it is very important to bear in mind that the difference between non-monotonic inferences and entailments is a question of degrees as Quine (Quine, 1953) has argued convincely. See (Gomez, 1996) for a discussion. null Our work differs from the semantic role list approaches (Fillmore, 1968) in several essential aspects. First of all in our method, the semantic roles are not defined independently of the meaning of the verb and are not semantically unanalyzable. In addition, the number of thematic roles depends on each predicate, and not on some general criteria saying which thematic roles there will be, irrespective of each predicate. Any thematic role in a predicate corresponds to a meaning relation from which inferences specific to that predicate or subpredicates must be established. Consider the sentence &amp;quot;These birds fly 11,000 miles from their breeding grounds to their winter home in the Antarctic.&amp;quot; What is the thematic role corresponding to the NP &amp;quot;11,000 miles?&amp;quot; Some semantic relation needs to be recognized so that inferences such as &amp;quot;there is a distance of 11,000 miles between the breeding grounds of these birds and their winter home&amp;quot; can be established. We have recognized that semantic relation by creating the role distance meaning the distance traveled by an animate agent in a change-oflocation predicate (Gomez, 1998). The inference is based on that role and on that predicate. Because the NP &amp;quot;11,000 miles&amp;quot; is not a subcategorized argument of &amp;quot;fly,&amp;quot; why call it a &amp;quot;role&amp;quot; and not an &amp;quot;adjunct&amp;quot;? From a semantic point of view, it makes no difference whether one calls it &amp;quot;adjunct&amp;quot; or &amp;quot;the- null if one assigns a thematic role to measures of distance, why not assign a role to &amp;quot;quickly&amp;quot; in &amp;quot;She walks quickly.&amp;quot; Our answer is that we should assign a role to &amp;quot;quickly.&amp;quot; But, that role, whatever its name, should not be placed in the predicate changeo\]-location but in the action node because it can be inherited by every subpredicate of action. It makes sense to say &amp;quot;She eats/studies/writes/.., quickly,&amp;quot; but not &amp;quot;She eats 20 miles.&amp;quot; Thus from our point of view, an &amp;quot;adjunct&amp;quot; is a role that is inherited by every subpredicnte of action.</Paragraph>
    <Paragraph position="2"> This approach does not lead to a propagation of roles since their number and nature depend on the generic predicate and its subpredicates. The critique of &amp;quot;role fragmentation - the subdivision of a single role into many subroles as result of subsequent analysis (Dowry, 1991) - is valid ff the entailments are based exclusively on the role, but not if they are anchored on the role and the predicate. The roles that we have used throughout our analysis have differences in meaning across diverse generic predicates, or verb classes. For instance, the meaning of the role theme in a change of state verb, say &amp;quot;break,&amp;quot; is different from its meaning in a transfer of information verb, say &amp;quot;tell.&amp;quot; Hence, if the entailments are based only on the role, one would be compelled to recognize several types of theme (Dowty, 1991), but because the entailments are based on the predicate and on the role, this is not necessary. Role entailments are shared by subpredicates of a generic predicate not across generic predicates.</Paragraph>
    <Paragraph position="3"> Our approach also differs from those analyses that attempt to reduce the verb semantic analysis to a small set of notions e.g., Jackendoff'localist hypothesis (Jackendoff, 1990), Dowty-Vendler's aspectual hypothesis (Vendler, 1967), Dowty (Dowry, 1979), or to a small set of primitives (Schank, 1975). Our major critique to reductionist analyses are Quinean (Quine, 1960) in nature, namely meaning is holistic. Trying to reduce verb meaning to a small set of notions is going to fail because verb meaning is intersentential. One may take some verbs whose meaning can be reduced to some few principles, or notions.</Paragraph>
    <Paragraph position="4"> Verbs of change of location fit very well within the localist hypothesis, while verbs of creation seem to support the aspectual hypothesis. But, if you consider a verb like &amp;quot;graduate&amp;quot; the situation is rather different because many sentences mediate to form its meaning. One may say that it is an accomplishment verb, but that is not saying much. Trying to provide a CD representation (Schank, 1975) for it seems hopeless. The key point here is that the representation needs to make possible the inferences normadly associated with such verb. For instance, that if one is asked which school Clinton attended, one should say &amp;quot;Yale&amp;quot; upon reading that Clinton grad- null uated from Yale. Hierarchical predicate decomposition and inferences shared by subpredicates of the same generic predicate is the solution that we offer.</Paragraph>
    <Paragraph position="5"> However, we always keep in mind that meaning is a question of degrees. It can be approximated, but not fully grasped. Fortunately, approximation is all we need for the task of natural language processing.</Paragraph>
  </Section>
class="xml-element"></Paper>