From nobody@digitalkingdom.org Thu Nov 17 08:13:10 2005 Received: with ECARTIS (v1.0.0; list lojban-list); Thu, 17 Nov 2005 08:13:10 -0800 (PST) Received: from nobody by chain.digitalkingdom.org with local (Exim 4.54) id 1EcmNf-0005Te-Ab for lojban-list-real@lojban.org; Thu, 17 Nov 2005 08:12:59 -0800 Received: from mx2.mail.ru ([194.67.23.122]) by chain.digitalkingdom.org with esmtp (Exim 4.54) id 1EcmNa-0005TT-HB for lojban-list@lojban.org; Thu, 17 Nov 2005 08:12:59 -0800 Received: from [212.17.3.19] (port=1079 helo=yanis.vpn.plhs) by mx2.mail.ru with esmtp id 1EcmNY-0007PF-00 for lojban-list@lojban.org; Thu, 17 Nov 2005 19:12:52 +0300 Date: Thu, 17 Nov 2005 22:12:51 +0600 From: Yanis Batura X-Mailer: The Bat! (v3.5.30) Professional X-Priority: 3 (Normal) Message-ID: <688330814.20051117221251@mail.ru> To: lojban-list@lojban.org Subject: [lojban] Lojban Rating System In-Reply-To: <55b258c20511161440v4364d61tef8affec250d8937@mail.gmail.com> References: <20051116205055.GE23316@chain.digitalkingdom.org> <20051116214057.20856.qmail@web81301.mail.mud.yahoo.com> <55b258c20511161440v4364d61tef8affec250d8937@mail.gmail.com> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----------4B1B76EAD4943E" X-Spam-Score: -2.0 (--) X-archive-position: 10795 X-ecartis-version: Ecartis v1.0.0 Sender: lojban-list-bounce@lojban.org Errors-to: lojban-list-bounce@lojban.org X-original-sender: ybatura@mail.ru Precedence: bulk Reply-to: lojban-list@lojban.org X-list: lojban-list ------------4B1B76EAD4943E Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable I think Lojban at the current stage of its development does not yet have an= y sufficiently large =91classical=92 textual base which could allow judging= level of speaking. The language is still very young, and forming. For example, consider the word =91Friday=92 =96 la mudjed. Do you think thi= s word must be frequent enough in a language? If you search =93la mudjed.= =94 in Google, the only reference to this word will be the link to a LFB le= sson about dates. That would of course be nonsense for a really formed lang= uage. I am sure that there are plenty of words (=91zero-words=92), either g= ismu or higher, that have been used only in the canonical lists and a coupl= e of illustrating sentences, and nowhere else. At this stage, as the promotional materials do state, everyone writing in L= ojban becomes history. This is not compatible with competition and ratings = based on wins/draws/losses. Instead, I suggest to measure not speaking proficiency, but the level of co= ntribution to the language (in creating its textual base). Of course, the o= riginal team that has developed the language in 80-s and 90-s, has by defau= lt an infinite level of contribution, as do many other Lojbanists, authors = of translations and/or original works that we can all see on page http://ww= w.lojban.org/tiki/tiki-index.php?page=3DTexts+In+Lojban. An outline of such measuring system may be like this: 1. =91Portions=92, each consisting of one or more sentences in English or o= ther non-Lojban language, are posted and stored on a server, categorized by= topics.=20 2. The goal is to have a number of correct translations of the portions int= o Lojban. 3. Everyone is free to add new portions on the server. 4. Everyone is free to add his/her translation of the portions. 5. Everyone is free to mark any translation (by other people) of any portio= n as =93correct=94 or =93incorrect=94, with optional comments. 6. The weight of a translation varies between 0 and 1. It is 0, IF no one h= as approved it, and 1, IF (no one has disapproved it AND at least one perso= n with non-zero level of contribution has approved it), OTHERWISE it is the= special function, definition below. 7. The level of contribution to the language is the sum of weights of all t= ranslations made by that person. The definition of the special function: Suppose N persons have marked a translation of a portion. Let Fi be equal += 0.5, if the i-th person has approved the translation, and -0.5, if he/she h= as disapproved the translation. Let Ri be the rating (level of contribution) of the i-th person. Let T =3D SUM(i=3D1..N, Ri), i. e. the sum of ratings (levels of contributi= on) of all markers [those who mark, I mean]. Then the weight of the translation, x, equals x =3D SUM(i=3D1..N, Fi * Ri / T). It can be easily seen that =93the special function=94 works for 0- and 1- c= ases as well. :) I think this approach is much more efficient than the combat stylish approa= ch suggested earlier.=20 Combats leave only =93won / lost=94 info behind. Who will collect and judge= those dialogs? That is a difficult problem. I won=92t speak about ethical = side of the question, since that was done by the previous posters. Yanis Batura Every sentence you write in Lojban makes the world a little better ------------4B1B76EAD4943E Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable

I think Lojban at the current stage of its development does not yet have= any sufficiently large ‘classical’ textual base which could al= low judging level of speaking. The language is still very young, and formin= g.


For example, consider the word ‘Friday’ – la mudjed. D= o you think this word must be frequent enough in a language? If you search = “la mudjed.” in Google, the only reference to this word will be= the link to a LFB lesson about dates. That would of course be nonsense for= a really formed language. I am sure that there are plenty of words (‘= ;zero-words’), either gismu or higher, that have been used only in th= e canonical lists and a couple of illustrating sentences, and nowhere else.=


At this stage, as the promotional materials do state, everyone writing i= n Lojban becomes history. This is not compatible with competition and ratin= gs based on wins/draws/losses.

Instead, I suggest to measure not speaking proficiency, but the level of contribution to the language (in crea= ting its textual base). Of course, the original team that has developed the= language in 80-s and 90-s, has by default an infinite level of contributio= n, as do many other Lojbanists, authors of translations and/or original wor= ks that we can all see on page http://www.lojban.o= rg/tiki/tiki-index.php?page=3DTexts+In+Lojban.

An outline of such measuring system may be like this:


1. ‘Portions’, each consisting of one or more sentences in E= nglish or other non-Lojban language, are posted and stored on a server, cat= egorized by topics. 

2. The goal is to have a number of correct translations of the portions = into Lojban.

3. Everyone is free to add new portions on the server.

4. Everyone is free to add his/her translation of the portions.

5. Everyone is free to mark any translation (by other people) of any por= tion as “correct” or “incorrect”, with optional com= ments.

6. The weight of a translation varies between 0 and 1. It is 0, IF no on= e has approved it, and 1, IF (no one has disapproved it AND at least one pe= rson with non-zero level of contribution has approved it), OTHERWISE it is = the special function, definition below.

7. The level of contribution to the language is the sum of weights of al= l translations made by that person.


The definition of the special function:

Suppose N persons have marked a tra= nslation of a portion. Let Fi be equal= +0.5, if the i-th person has approved the = translation, and -0.5, if he/she has disapproved the translation.

Let Ri be the rating (level of cont= ribution) of the i-th person.

Let T =3D SUM(i=3D1..N, Ri), i. e. the s= um of ratings (levels of contribution) of all markers [those who mark, I me= an].

Then the weight of the translation, x, e= quals


x =3D SUM(i=3D1..N, Fi * Ri / T).


It can be easily seen that “the special function” works for = 0- and 1- cases as well. :)


I think this approach is much more efficient than the combat stylish app= roach suggested earlier. 


Combats leave only “won / lost” info behind. Who will collec= t and judge those dialogs? That is a difficult problem. I won’t speak= about ethical side of the question, since that was done by the previous po= sters.


Yanis Batura

Every sentence you write in Lojban makes the world a= little better

------------4B1B76EAD4943E-- To unsubscribe from this list, send mail to lojban-list-request@lojban.org with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if you're really stuck, send mail to secretary@lojban.org for help.