An improvement was proposed by introducing a new integrated semantic representation that combines structural semantics and lexical semantics in a single representation. This new representation resulted in a much better predicate-based sense candidates selection process, which boosted up the accuracy of sentence sense

annotation task significantly. The designed framework is developed based on the theories of Head-driven Phrase Structure Grammar (Sag et al., 2003), Minimal Recursion Semantics (Copestake et al., 2005) and WordNet (Fellbaum, 1998).

The Integrated Semantic Framework has been implemented as a working system which produces computational semantic analysis and is evaluated against a human-annotated corpus. Although the final system only provides analyses for standard English text, the framework design is extensible to other languages and is already being developed for the Japanese language.

TL;DR