Stats database for UKRM

I'm trying to knock up a few PHP pages so that posters can play with
personalised stats.

The table structure is as follows:

- Message-id
- Name
- Newsreader
- Cross-posted groups (up to 5 including UKRM)
- Day
- Month
- Year
- Hour
- Days since 2000
- Original size of message, including text from other posters
- Original number of lines from poster
- Signature size
- Characters in post
- OT or not?
- X-No-Archive or not

Any ideas as to what requests I can set up? Some stuff I was thinking
of:

- Number of times posted
- Posts per month for a specific year(s)
- Posts per day for a specific month(s)
- Largest post for day/month/year
- Preferred newsreaders

Oh, I've also seen that I've got 1,343,917 lines of data going back to
March 2003. I can't believe I've been running this script since then.
Bloody hell.

 

The stats package on my web account goes in to ridiculous detail right
down to hourly stats.

Modal posting hour ?
Most used cross-posted combinations ?

--
+-------------------------------------------------------------------+
| Pete Fisher at Home: |
| Voxan Roadster Yamaha WR250Z/Supermoto "Old Gimmer's Hillclimber" |
| Gilera GFR * 2 Moto Morini 2C/375 Morini 350 "Forgotten Error" |
+-------------------------------------------------------------------+

 

WHY

 

Borg wibbled forthrightly:

It's a hobby. HTH.

 

F*cking newbie ...

Paul.

 

Wee-el.

I think strip out the sigs into a separate DB table line per entry. Then
match these with make and model and year of bike. Following this we can
regex all the posts for addresses and mention of holidays "Play nicely"
for example would result in two of the following "ZX10R", "Hayabusa",
"GPz750 turbo" being available to "borrow".

We won't need to do this for Ivan of course, as we've all seen him mowing
his lawn in the back garden from satellite. <waves Hi Ivan we're up here>
Me too if you look back for an octagonal pool on a google earth link
somewhere.

Save all of the posters IPs that leave one in the posting host field,
this can be piped through some geoip lib for rough location or in the
case of some of our JANET users point right at their university. Well the
ones who don't bother to tell us they work at Bath in the Organisation
field anyway (Hi M J) . Whois servers will quite often spit out a netblock
which may equate with an employer which can be linked with the poster and
other information seeping slowly from us as we partake of our online
joys. This leakage slowly accumulating into a /distressingly/ large, pool
of information which as more arrives can be linked together to form a
picture of our poster.

So who was going to the EOSM? Can attendees confirm?

All of this now geo-linked(ish) info can be exported to a KML file for
viewing in google earth or a clickon link to google maps with bikes,
approximate ("second hand") values and holiday dates in a nice easy to
read map for........

By the way does anyone have a non-descript transit I can borrow by any
chance or a long wheelbase merc with soundproofing?

For a slightly less malign usage of the vast quantity of data you have
gathered.

You can extract urls and combine the gingometer with the overall
structure.
<link> was first posted by ????? in msgid ?????
This could be used to filter the internet of multiply posted videos of
the same name by having a report ginge button with the option to add the

link youtube.com/mongonbikecrashes the same video as
vimeo.com/mongonbikecrashes etc. these urls eh far from u.

You could also use it for a links of the day page, some people <waves>
like lots of clicky things recomended by the group. Update at 10.00pm
daily or some such arbitrary time or live if you are feeling adventurous.
Some links however, no matter how ginge they are, are often far more
brownish in colour. Veronica for example could be marked as
*******************************************
****** REALLY NSFW AT ALL!! EVER!!! ******
*******************************************
and left as a copy and paste rather than a directly clickable item.
Somebody (names not being mentioned here of course) /will/ post it again
they can't help themselves. Just as I always click it and go WTF.

Counting the swear words[1] and the per poster vocabulary count of
individual words was always my fave though. The oft loquatious and
garrulous postings found here from obviously well educated members of
society a few libations over the limit produced a word list worthy of
many a giggle. The compound nouns in particular of up to 30 letters were
a particularly funny list. However after going through my database and
seeing what I could find out about posters on UKRM I became very paranoid
about just how much there was. I peered out of my windows suspiciously
before closing the curtains and swiftly setting the X-No-Archive flag to
yes. I then deleted the DB and made a nice calming cup of tea.

There are limits to what man should know. With great power (or piles of
data) comes great responsibility etc.....[2]

[1] Not including Scunthorpe or Arsenal and some others of course.
cunning use of the perl word-boundary regex is required.
[2]Unless you work in government or hospitals apparently.

 

On 2010-10-01 23:00:08 +0200, davethedave <> said:

You Sir, have far too much time on your hands.

<wraps router in tinfoil>

 

<snip>

I echo DNC. Far, far too much time...

 

Largest number of individual replies to a single posting? Not 'longest
thread' style, as that'll activate all those 9-11 conspiracy ones, but
just the largest number of direct replies to a single posting.

 

Could do with some worthwhile employment.
Bit of a slack week.

 

des wibbled forthrightly:

What version of the script do you want to use? The new enhanced
version, requiring mySQL or the old, "trawl through the spool
directory" version?

 

The statistic I'd like to see is the difference between posting time for
the posting and the first-referenced post. So you could see who follows
most closely.

 

Jim wibbled forthrightly:

I don't drill down into the minute or second a post is made. I never
saw the point. I suppose I could do it now, but it won't be retroactive.

 

"Stalker rating"

Paul

 

<breathes heavily down neck>