jbovlaste
a lojban dictionary editing system
User:
Pass:

Home
Get A Printable Dictionary
Search Best Words
Recent Changes
How You Can Help
valsi - All
valsi - Preferred Only
natlang - All
natlang - Preferred Only
Languages
XML Export
user Listing
Report Bugs
Utilities
Status
Help
Admin Request
Create Account
Discussion of "cfilyfacki"
[parent] [root]
Comment #5: Re: Not lowest scoring lujvo
Wuzzy (Thu Feb 27 02:48:09 2014)

selpahi wrote:
> Wuzzy wrote:
> > Jbovlaste only should have the lowest scoring form of a lujvo.
Jbovlaste
> > tells you that every time you want to create a new word. The lowest
> > scoring form for this is not “cfilyfacki” but “cfifa'i”.
>
> So you also reject ci'omle because citmle has a better score?
Yes. I reject it being in the dictionary.
However, this does NOT mean that I reject “ci'omle” _in general_.

> What is
> the benefit of such a policy?
To avoid reduncancy in jbovlaste. Consider the opposite case: You include
every possible word form of a given lujvo. But everything is treated in
jbovlaste as if it were a completely different word. This would mean
you’d repeat the definition over and over again. And if you want to edit
something, you would have to edit the other definitions as well. It is
very easy to become inconsistent then. To make life easier, the dictionary
only includes the lowest-scoring form. It is assumed the user can infer
all the other valid forms of the lujvo. The definition applies to all
lujvo forms equally.

> If lots of people prefer the sound of
> ci'omle, why should the score matter?
The score only matters for dictionary purposes. In actual speech, the
score does not matter.

What about this compromise?: If you want a certain word form to be
included, mention it in the notes instead of creating an entire new entry.
Here’s an example: malgli (mentions “malglico”). There is no entry
for “malglico” because it is not needed.

> "jbovlaste says Y" so suggesting other word forms means
> we must downvote them all.
I have justified that. There is a very good reason for this policy.


> "The CLL says X so anything else is completely wrong and evil"
and
> I find this attitude very destructive.
I understand. This is a big topic and perhaps I will explain myself on the
mailing list in detail.

Comment #7: Re: Not lowest scoring lujvo
Lorenzo Von Matterhorn (Thu Feb 27 11:02:10 2014)

Wuzzy wrote:
> selpahi wrote:
> > Wuzzy wrote:
> > > Jbovlaste only should have the lowest scoring form of a lujvo.
> Jbovlaste
> > > tells you that every time you want to create a new word. The lowest
> > > scoring form for this is not “cfilyfacki” but “cfifa'i”.
> >
> > So you also reject ci'omle because citmle has a better score?
> Yes. I reject it being in the dictionary.
> However, this does NOT mean that I reject “ci'omle” _in general_.
>
> > What is
> > the benefit of such a policy?
> To avoid reduncancy in jbovlaste. Consider the opposite case: You
include
> every possible word form of a given lujvo. But everything is treated in
> jbovlaste as if it were a completely different word. This would mean
> you’d repeat the definition over and over again. And if you want to
edit
> something, you would have to edit the other definitions as well. It is
> very easy to become inconsistent then. To make life easier, the
dictionary
> only includes the lowest-scoring form. It is assumed the user can infer
> all the other valid forms of the lujvo. The definition applies to all
> lujvo forms equally.

Okay, this is a valid point. Under the current rules, every two lujvo that
contain rafsi of the same words in the same order are considered the same
word, so it would be easier to have them organized more automatically.
However, jbovlaste has no such feature, and it also doesn't seem to happen
very much that multiple forms are added.

> > If lots of people prefer the sound of
> > ci'omle, why should the score matter?
> The score only matters for dictionary purposes. In actual speech, the
> score does not matter.

Sure, but apparently the scoring algorithm can sometimes prefer words that
humans don't prefer. When that happens, maybe we (the speakers) should
have a bigger say than an artificial algorithm (whose initial purpose it
was to make the shortest and most pleasing forms get the best scores).
This sort of thing doesn't happen so often that adding an alternative form
causes a lot of extra effort. I can think of ci'omle and je'umlu,
which some people prefer because they get rid of the "tml" cluster, or
because they don't like cit in the word for "cute".


> What about this compromise?: If you want a certain word form to be
> included, mention it in the notes instead of creating an entire new
entry.
> Here’s an example: malgli (mentions “malglico”). There is no
entry
> for “malglico” because it is not needed.

That would be an option.

> > "The CLL says X so anything else is completely wrong and evil"
> and
> > I find this attitude very destructive.
> I understand. This is a big topic and perhaps I will explain myself on
the
> mailing list in detail.

Okay. It's not a very fun topic, though. :)

Comment #8: Re: Not lowest scoring lujvo
gleki (Thu Feb 27 11:10:36 2014)

li'a JyVySy enai le bangu cu fliba

Currently, jbovlaste will accept data for 70 languages.
You are not logged in.

  recent changes jbovlaste main
This is jbovlaste, the lojban dictionary system.
The main code was last changed on Wed 07 Oct 2020 05:54:55 PM PDT.
All content is public domain. By submitting content, you agree to place it in the public domain to the fullest extent allowed by local law.
jbovlaste is an official project of the logical language group, and is now headed by Robin Lee Powell.
E-mail him if you have any questions.
care to log in?