Selectional Preference and Sense Disambiguation

Selectional Preference and Sense Disambiguation Philip Resnik Department of Linguistics and Institute for Advanced Computer Studies University of Maryland College Park, MD 20742 USA re snik@um±acs, umd. edu Abstract The absence of training data is a real prob- lem for corpus-based approaches to sense disambiguation, one that is unlikely to be solved soon. Selectional preference is tra- ditionally connected with sense ambigu- ity; this paper explores how a statistical model of selectional preference, requiring neither manual annotation of selection re- strictions nor supervised training, can be used in sense disambiguation. 1 Introduction It has long been observed that selectional con- straints and word sense disambiguation are closely linked. Indeed, the exemplar for sense disambigua- tion in most computational settings (e.g., see Allen's (1995) discussion) is Katz and Fodor's (1964) use of Boolean selection restrictions to constrain semantic interpretation. For example, Mthough burgundy can be interpreted as either a color or a beverage, only the latter sense is available in the context of Mary drank burgundy, because the verb drink specifies the selection restriction +LIQUID for its direct objects. Problems with this approach arise, however, as soon as the domain of interest becomes too large or too rich to specify semantic features and selection re- strictions accurately by hand. This paper concerns the use of selectional constraints for automatic sense disambiguation in such broad-coverage settings. The approach combines statistical and knowledge-based methods, but unlike many recent corpus-based ap- proaches to sense disambiguation (¥arowsky, 1993; Bruce and Wiebe, 1994; Miller et al., 1994), it takes as its starting point the assumption that sense- annotated training text is not available. Motivat- ing this assumption is not only the limited avail- ability of such text at present, but skepticism that the situation will change any time soon. In marked contrast to annotated training material for part- of-speech tagging, (a) there is no coarse-level set of sense distinctions widely agreed upon (whereas part-of-speech tag sets tend to differ in the details); (b) sense annotation has a comparatively high er- ror rate (Miller, personal communication, reports an upper bound for human annotators of around 90% for ambiguous cases, using a non-blind evaluation method that may make even this estimate overly optimistic); and (c) no fully automatic method pro- vides high enough quality output to support the "an- notate automatically, correct manually" methodol- ogy used to provide high volume annotation by data providers like the Penn Treebank project (Marcus et al., 1993). 2 Selectional Preference as Statistical Association The treatment of selectional preference used here is that proposed by Resnik (1993a; 1996), combining statistical and knowledge-based methods. The basis of the approach is a probabilistic model capturing the co-occurrence behavior of predicates and con- ceptual classes in the taxonomy. The intuition is il- lustrated in Figure 1. The prior distribution PrR(c) captures the probability of a class occurring as the argument in predicate-argument relation R, regard- less of the identity of the predicate. For example, given the verb-subject relationship, the prior prob- ability for (person) tends to …

Content truncated to 500 words