I05-2011 texts , is less reliable than the lexical method . However , it does for example
D08-1053 combines the length model with the lexical method . ( Simard and Plamondon 1996
J05-1005 Grefenstette 1994 ) . By contrast , lexical methods are able to acquire information
S01-1013 are fully unsuper - vised . The lexical method makes use of no external resources
J05-1005 . To solve this problem , many lexical methods estimate the probabilities of
S01-1013 windowed " the inputs for the lexical method , by allowing a maximum of 10
J05-3004 string-matching baselines and for the lexical methods are higher for definite coreferential
J05-1005 the main goal of semantic and lexical methods is precisely the acquisition
J05-1005 of the continuum , we find the lexical methods , that is , those strategies
J05-1005 . Special attention is paid to lexical methods . At the end , we situate our
J05-1005 co-occurrences . We thus follow a lexical method . However , selection restrictions
J05-1005 , in section 8.3.4 , we use a lexical method with similarity-based generalization
P06-2105 alignment . The performance of this lexical method ( LEX - ALIGN ) is shown in Tables
N06-1047 act length can outperform the lexical methods of text summarization ap - proaches
J05-3004 a small data set on which the lexical methods can differ . Thus , StrSetv2n
A00-1004 ; Gale and Church , 1991 ) to lexical methods ( Kay and Roscheisen , 1993 ;
J05-3004 can not be ruled out by purely lexical methods ( Example ( 10 ) ) . The integration
S01-1013 § 3 ) . Next , we outline the lexical method ( § 4 ) and structural method
P08-1117 avoids the locality problems of lexical methods . However , these approaches
J10-3003 lattices constructed via the purely lexical method . However , they present no analysis
hide detail