Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!travelers.mail.cornell.edu!news.tc.cornell.edu!news.cac.psu.edu!news.pop.psu.edu!hudson.lm.com!newsfeed.pitt.edu!gatech!EU.net!sun4nl!sci.kun.nl!cs.kun.nl!markjan
From: markjan@cs.kun.nl (Mark-Jan Nederhof)
Subject: Re: Complexity of Chartparsing
Message-ID: <D9EJEK.3G9@sci.kun.nl>
Sender: news@sci.kun.nl (News owner)
Nntp-Posting-Host: zeus.cs.kun.nl
Organization: University of Nijmegen, The Netherlands
X-Newsreader: NN version 6.4.19
References: <3pv36l$n6u@cd4680fs.rrze.uni-erlangen.de> <HT.95May26163946@chestnut.ling.upenn.edu>
Date: Tue, 30 May 1995 17:32:43 GMT
Lines: 13

In <HT.95May26163946@chestnut.ling.upenn.edu> ht@chestnut.ling.upenn.edu (Henry S. Thompson) writes:

>The general result for worst-case asymptotic time-complexity for a
>chart parser for the recognition task given an unrestricted
>context-free phrase structure grammar and a string is the expected
>O(n3m2) where n is the number of symbols in the string and m the number
>of productions in the grammar.

Actually it is O(n^3 * m), not O(n^3 * m^2). E.g. Earley's algorithm can be easily
changed so that it becomes linear in the size of the grammar. For this I refer to my thesis.

Mark-Jan Nederhof
Univ. Groningen
