[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Pan-users] Re: Scoring update

From: mlist . pan . users
Subject: [Pan-users] Re: Scoring update
Date: 21 Feb 2003 10:12:35 -0800
User-agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Honest Recruiter)

==> "ck" == Charles Kerr <address@hidden> writes:

    ck> Since Pan 0.13.4 is in string freeze, and I'm home from work
    ck> between doctors' appointments, I spent a little time this
    ck> morning working on scoring for Pan 0.14.0.  Things are looking
    ck> really good, so I made up a screenshot:


I have a feature suggestion for the scoring interface in PAN.
This is from my experience using GNUS, which has a full-featured
scoring engine.  Depending on how good PAN gets in handling
scores, I may drop GNUS entirely...

I find that sorting by score doesn't really work very well.  If you
sort strictly by the numeric value, the thread ordering is more or
less arbitrary.  What I found from using GNUS was that some sort of
'banding' was necessary, to quantize the articles by score ranges.
You do something like this with the 'really high score' and 'really
low score' stuff, but it's really broad.

When I use GNUS, and have the ability to write my own primary-key
article sort functions in elisp, I find that a good 'banding' function
is something like int(ln(SCORE)), or int(log(SCORE)).  This results
in a good thread presentation, because threads with the same rough
score are then sorted alphabetically.  Something to think about, in any

Oh, and scoring is pretty addictive.  Think carefully about what
keyboard accelerators you want to provide, to make it as quick as
possible to score-up or score-down by subject substrings.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]