Subject:
|
Re: New 9V Digital Trains for Germany this Autumn
|
Newsgroups:
|
lugnet.admin.general
|
Date:
|
Wed, 5 Apr 2000 19:23:34 GMT
|
Highlighted:
|
(details)
|
Viewed:
|
804 times
|
| |
|
|
In lugnet.admin.general, Frank Filz writes:
> The problem I see with the rating system is that if it is something
> totally subjective and each person who rates an article can use whatever
> method and scale they feel like to assign ratings, then the rating
> system seems to me to be close to useless.
There are many different methods that people can use in their minds as they
give input, but all of the methods have the same net effect: They give a
number which gets averaged, and that average score is a reflection of a
collective "read recommendation" to other readers.
> Of course I tend to view any
> rating system as close to useless anywise. Unless you understand how the
> particular individual is making the rating, you have no idea of how
> usefull the thing being rated is to you.
>
> I'm not sure what the exact goal of the system is for Lugnet. [...]
Well, there isn't any one single exact goal, but a few are described here:
http://www.lugnet.com/admin/general/?n=5487
The main goal right now is to help casual readers identify stronger content
more quickly.
> If the
> goal is to allow someone who comes along two years from now to be able
> to easily find the posts and threads that have permanent value, then I
> think that should be stated. If that is the goal, then I would rate
> things like a post in .trains which lays out module standards as very
> high. I would rate a post in .shopping which tells about todays sale as
> a 0 (after today the post is useless). If I'm able to change the rating,
> I'd rate the .shopping post 100 today, and 0 tomorrow.
The elapsed time (article age) will be accounted for later in searches and
top-N lists by scaling the score according to some f(t). I recommend scoring
based on how useful (or fun or interesting, etc.) the article was to you at
the moment you read it (or when it was posted). Thus, S@H specials should
IMHO get 100 (or, say, 90 if there isn't anything good that week) even though
the information won't be very useful a week later.
In terms of auctions, I've been giving higher scores somewhat to auctions
which list older and harder-to-find items than those which list newer and
easier-to-find items (harder or easier to find at the time of posting, that
is). For example, I think I marked a 90 or 100 on an ultra-rare 1970's set
last week, and also some very high scores on some unique but brand-new
collections of parts that someone had parted out.
The scores should IMHO be your personal feeling about how you reacted to the
message and how strongly you would recommend it to someone stopping by that
very instant (or, say, within a few minutes). Me, personally, I would very
much highly recommend taking a look at auctions of rare things over auctions
of common things, but other people may have different reasons for recommending
or reacting to certain auctions and may give different scores, and that's OK,
it's all meant to be averaged.
> Unless it is clear what the intent of the rating system is, people are
> going to feel bad if their articles are consistently rated lower than
> other articles, and they are going to want to know why, and personally,
> I think in that case they would be deserving of hearing the reasoning.
> It may in the end be something totally impersonal (the person may just
> be talking about things that are of low interest to the group, perhaps
> an example might be someone who's primary interest was Duplo trains and
> was trying to strike up a conversation in .trains).
That's a good example. If many or several people all rate such an article
low, I think it's perfectly fair to wonder and ask why. But if one or two
people rate it low (especially one), it's probably not worth asking -- it's
not statistically significant yet until there are probably 4 or more people
all roughly agreeing.
> In another post, you talked about people being able to customize the
> system rating on a per group basis, what might actually be better, from
> the perspective of ordering searches, is for people to be able to scale
> ALL the ratings in a group (say I'm searching for stuff about trains,
> obviously I want to search .trains, but there might be some stuff in
> .town and .general, so I might chose to scale them down by a factor of 2
> so I only have to look at the train related posts there which really
> impressed people).
That would really rock for searches! Being able to tweak things on a high-
level per-group search-by-search basis... I suppose the ultimate tweaker
would do this...?
R = [ sum(d, v1, v2, v3, ... vn) / (n + 1) ] ^ a * b
where a>1 pushes medium and low scores lower and keeps high scores high, and
a>1 pushes low and medium scores up and keeps high scores high, and b defalates
or inflates things linearly (like b=.5 for .town and .general to scale 100's
in those groups down to 50's). And then d gives the default rating for the
group, which is currently always 50. [Assuming a range of 0 to 1 here rather
than 0 to 100 for simplicity of forumla.]
> Because of all of the above, I have to say that I am essentially
> ignoring the rating system. I just don't see how it is usefull to me
> right now.
--Todd
|
|
Message has 2 Replies:
 | | Re: New 9V Digital Trains for Germany this Autumn
|
| (...) One problem I could see with this is that lets say a given set shows up for 10 weeks in S@H, and all those S@H specials posts end up with a very high rating. Then, a review of the set, which is a set people generally don't care for gets a (...) (25 years ago, 5-Apr-00, to lugnet.admin.general)
| | | Top-N article list adjusted for article-age
|
| (...) Here's a crude example -- a static list built "by hand" from a few one-liners using current data: (URL) told it to examine all the articles which have been rated so far, and adjust their ratings downward the older the article, according to (...) (25 years ago, 7-Apr-00, to lugnet.admin.general) !
|
Message is in Reply To:
| | Re: New 9V Digital Trains for Germany this Autumn
|
| (...) The problem I see with the rating system is that if it is something totally subjective and each person who rates an article can use whatever method and scale they feel like to assign ratings, then the rating system seems to me to be close to (...) (25 years ago, 5-Apr-00, to lugnet.admin.general)
|
24 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
