From taral@taral.net Tue May 09 07:48:00 2000 Return-Path: Received: (qmail 9637 invoked from network); 9 May 2000 14:48:00 -0000 Received: from unknown (10.1.10.26) by m1.onelist.org with QMQP; 9 May 2000 14:48:00 -0000 Received: from unknown (HELO mail.taral.net) (209.217.149.127) by mta1 with SMTP; 9 May 2000 14:47:59 -0000 Received: by mail.taral.net (Postfix, from userid 500) id 14A2B26332; Tue, 9 May 2000 09:47:58 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by mail.taral.net (Postfix) with ESMTP id 0DB2A24B62; Tue, 9 May 2000 09:47:58 -0500 (CDT) Date: Tue, 9 May 2000 09:47:58 -0500 (CDT) X-Sender: taral@r149127-2815.dobiecenter.com To: "Bob LeChevalier (lojbab)" Cc: lojban@egroups.com Subject: Re: OT - programming logflash Re: [lojban] Logflash In-Reply-To: <4.2.2.20000508234405.00aaf760@127.0.0.1> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII From: Taral On Tue, 9 May 2000, Bob LeChevalier (lojbab) wrote: > The ultimate test is in the results. People have found LogFlash boring or > frustrating, but so far as I know, no one who has completed the program has > had cause to complain that they hadn't thoroughly learned the words. I was thinking of making a couple modifications to the algorithm once a complete replica was made: 1) Using word frequencies to alter how frequently a word is tested. 2) Tracking error rates on both L->E and E->L directions. (Possibly using an aging algorithm... it will need tuning.) 3) Providing multiple correct answers for translation. (I'm not so sure about this one.) Any comments? Taral