Adaptive score and highlighting?

Adaptive score and highlighting?

[Mats Lidell]

Hi,

I'm using the default adaption rules and I then get authors
highlighted if I just read them once (!?) Or they are "lowered" if just
a score of "-1" is attached to them.

I dont' like this. Is there some way to get "highlight" when above
some score (say 10) and "dimmed" when lower another score (say -10) so
that small random changes won't trigger highlighting.

I think there is a general problem also involved with this type of
mechanism. Authors that once get "highlighted" will get their score
increased because you must read him again because he is highlighted. I
guess there is no solution to this other than to remove his score when
you find out that this author doens't deserve that score. But then how
do you do that in a controlled maner. Lower on author?

%% Mats

Re: Adaptive score and highlighting?

[Steven L. Baur]

>>>>> "Mats" == Mats Lidell <konmsll@eua.ericsson.se> writes:

    Mats> Hi, I'm using the default adaption rules and I then get
    Mats> authors highlighted if I just read them once (!?) Or they
    Mats> are "lowered" if just a score of "-1" is attached to them.

    Mats> I dont' like this. Is there some way to get "highlight" when
    Mats> above some score (say 10) and "dimmed" when lower another
    Mats> score (say -10) so that small random changes won't trigger
    Mats> highlighting.

Nope.  You may set the default score, but you may not specify a
default score range.  That's a good idea though.

    Mats> I think there is a general problem also involved with this
    Mats> type of mechanism. Authors that once get "highlighted" will
    Mats> get their score increased because you must read him again
    Mats> because he is highlighted.

This is a problem.  The effect (which I've seen for as long as I've
been using Gnus) is that if you are unfortunate enough to read a
spammed article, subsequent copies you see (by default) will have
their scores raised.

I've found various workarounds using adaptive scoring.  First off,
when you leave a group, the scoring information is based on the mark
the article has.  Let's say you read an article, and are less than
pleased with the author or thread.  If you C-k or d after reading,
then the article (and author) will get a reduced score in the adaptive
scoring.  Note that the values by which the score gets reduced is
controlled by the gnus-default-adaptive-score-alist variable.

    Mats> I guess there is no solution to
    Mats> this other than to remove his score when you find out that
    Mats> this author doens't deserve that score. But then how do you
    Mats> do that in a controlled maner. Lower on author?

If you want to nuke the author, then lower on author will do nicely.
The worst offenders though, seem to have a propensity for varying
their from lines, so you will have to be clever about it.

-- 
steve@miranova.com baur

Re: Adaptive score and highlighting?

[Per Abrahamsen]

>>>>> "ML" == Mats Lidell <konmsll@eua.ericsson.se> writes:

ML> Is there some way to get "highlight" when above
ML> some score (say 10) and "dimmed" when lower another score (say -10) so
ML> that small random changes won't trigger highlighting.

Yes, by setting `gnus-summary-highlight', either directly or by typing
`M-x customize RET' and modifying the `Summary Line Highlighting'
field.

Be warned that it is rather complicated either way.  You need to
change `(> score default)' to `(> score (+ default 10))' each place it
occur in the definition of that value

ML> I think there is a general problem also involved with this type of
ML> mechanism. Authors that once get "highlighted" will get their score
ML> increased because you must read him again because he is highlighted. I
ML> guess there is no solution to this other than to remove his score when
ML> you find out that this author doens't deserve that score. But then how
ML> do you do that in a controlled maner. Lower on author?

Yep.  Type `L A' when you decide that it was a mistake to read that
author.