From rspeer@MIT.EDU Thu Apr 08 18:51:40 2004 Received: with ECARTIS (v1.0.0; list lojban-list); Thu, 08 Apr 2004 18:51:40 -0700 (PDT) Received: from fort-point-station.mit.edu ([18.7.7.76]) by chain.digitalkingdom.org with esmtp (Exim 4.30) id 1BBlB8-0005QD-Gg for lojban-list@lojban.org; Thu, 08 Apr 2004 18:51:34 -0700 Received: from grand-central-station.mit.edu (GRAND-CENTRAL-STATION.MIT.EDU [18.7.21.82]) by fort-point-station.mit.edu (8.12.4/8.9.2) with ESMTP id i391pSpe017342 for ; Thu, 8 Apr 2004 21:51:32 -0400 (EDT) Received: from melbourne-city-street.mit.edu (MELBOURNE-CITY-STREET.MIT.EDU [18.7.21.86]) by grand-central-station.mit.edu (8.12.4/8.9.2) with ESMTP id i391pSL6011228 for ; Thu, 8 Apr 2004 21:51:28 -0400 (EDT) Received: from torg.mit.edu (TORG.MIT.EDU [18.208.0.57]) ) by melbourne-city-street.mit.edu (8.12.4/8.12.4) with ESMTP id i391pRUa022612 for ; Thu, 8 Apr 2004 21:51:27 -0400 (EDT) Received: from rob by torg.mit.edu with local (Exim 3.36 #1 (Debian)) id 1BBlB6-0007PA-00 for ; Thu, 08 Apr 2004 21:51:32 -0400 Date: Thu, 8 Apr 2004 21:51:32 -0400 From: Rob Speer To: lojban-list@lojban.org Subject: [lojban] Re: Beta Release of PEG-based Lojban parser. Message-ID: <20040409015132.GA28275@mit.edu> Mail-Followup-To: lojban-list@lojban.org References: <20040409002858.GH14789@digitalkingdom.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20040409002858.GH14789@digitalkingdom.org> X-Is-It-Not-Nifty: www.sluggy.com User-Agent: Mutt/1.5.5.1+cvs20040105i X-archive-position: 7507 X-ecartis-version: Ecartis v1.0.0 Sender: lojban-list-bounce@lojban.org Errors-to: lojban-list-bounce@lojban.org X-original-sender: rspeer@MIT.EDU Precedence: bulk Reply-to: lojban-list@lojban.org X-list: lojban-list On Thu, Apr 08, 2004 at 05:28:58PM -0700, Robin Lee Powell wrote: > I am considering an extension to allow 'si' or 'sa' at the beginning of > text (presumably to erase stuff from the proceeding utterance). The > morphology needs massive amounts of work, and ideally I'd like to get > Nora and Pierre's full algorithm encoded. I may also hack an extremely > minimal pre-processor to do zoi. At some point the parser needs to be > taught to output something more useful than just the text it succeeded > at parsing, but I'm really hoping someone with actual Java experience > will look at that. So I assume this means you can't currently get parse trees out of it. But it's not something about PEGs that you can't get parse trees, right? With the right code, the parse trees could be output? The reason I ask is that this summer, I've got a research job in semantic interpretation of natural languages. I've gotten permission from my advisor to apply my research to Lojban. This technique of semantic interpretation involves taking a CFG and assigning a lambda expression to every rule. You then apply these lambda expressions up a parse tree, and at the top you end up with an expression encoding the meaning of the sentence. So this should apply to Lojban very well. Now, I also think that the PEG is very cool and is fixing some of the most blatant things wrong with Lojban, and I'd like to be able to use it. So that requires being able to retrieve parse trees. -- mu'o mi'e rab.spir