REPRESENTING WORD MEANINGS
|Computer Science; Computer Science, Artificial Intelligence
|LECTURE NOTES IN ARTIFICIAL INTELLIGENCE
Starting with an approach to lexical semantics proposed by Bierwisch, a structure of semantic information in lexical entries is sketched which takes into account lexical ambiguities and polysemy. Some phenomena will be discussed which show that the strict discrimination between context-specific conceptual interpretations and context-invariant semantic forms may not be adequate. The proposed method for representing lexical meanings allows both, abstract context-invariant lexical entries which need specification during analysis by conceptual shifting and conceptual differentiation, and specific word meanings from which other meanings can be derived if needed. On the one hand, this integrated approach supplies a framework which is suitable for modelling processes of learning abstract meanings from typical uses of words. On the other hand, derivation operations may be sensitive to world knowledge, therefore the scope of possible context-specific meanings can change following respective variations in world knowledge without any need to revise lexical entries.
Show full item record
checked on Feb 27, 2024