In this paper we propose to use text chunking for controlling a bottom-up parser. As it is well known, during analysis such parsers produce many constituents not contributing to the correct final analysis, because they do not consider properly the text surrounding the portion under analysis in a given moment. This paper defines the following nested levels of chunks: words; non recursive NPs and verb groups (chunk1); non recursive PPs and DPs (chunck2); clauses and sentences. The paper compares the same CYK-like parser’s results during two blind tests (using or not a chunk-based strategy) on 51 Italian sentences (1379 words). The controlled parser was shown more efficient. Moreover, its (partial) results are qualitatively better, as here demonstrated
Controlling a Bottom-Up Parser through Text Chunking
1996-01-01
Abstract
In this paper we propose to use text chunking for controlling a bottom-up parser. As it is well known, during analysis such parsers produce many constituents not contributing to the correct final analysis, because they do not consider properly the text surrounding the portion under analysis in a given moment. This paper defines the following nested levels of chunks: words; non recursive NPs and verb groups (chunk1); non recursive PPs and DPs (chunck2); clauses and sentences. The paper compares the same CYK-like parser’s results during two blind tests (using or not a chunk-based strategy) on 51 Italian sentences (1379 words). The controlled parser was shown more efficient. Moreover, its (partial) results are qualitatively better, as here demonstratedI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.