Posting of sumo results (off topic)

[Stewart Nelson <sn@scgroup.com>]

I have been considering various means of providing this ML with sumo
results which are more accurate, timely, and/or comprehensive than
presently available.  The many Web sites which use this data would
also benefit.  This post is *not* a criticism of those who have been
typing in results manually - your efforts are greatly appreciated.

My thoughts have been in three areas: form-based software to ease the
input process, retrieving data from the NHK broadcast, and tapping the
official results.  If any of you have suggestions for other methods, I
would be glad to hear them.  (If Onaruto was right, some hacker could
penetrate the appropriate system and post "results" here even before
the matches take place :-)


FORM-BASED SOFTWARE

I am proposing to create a Web-based package for entering data.  Any
ML member wishing to post upcoming matches (torikumi-hyo) or results
would access the appropriate form.  The "upcoming" form for makuuchi
would display an array of 50 rikishi names, each with a checkbox.
(The top ten juryo rikishi are included, in case they fight in
makuuchi.)  The user clicks on the two participants of a match, and
hits "post".  The program then presents a new blank form for the next
match.  When all have been input, a formatted text file is returned,
to which the user may add commentary, before sending it out as e-mail,
posting to a forum, etc.

The server script would validate the match, checking that:
* both rikishi are not from the same heya
* they have not met on an earlier day this basho
* neither has been already scheduled for today
* the higher-ranked rikishi ranks above those already scheduled
* there are enough rikishi of higher ranks to fill matches remaining

I feel that these checks should catch most errors.  If an error is
detected, the user could correct his input, or override the warning
(in case the database at the server was wrong).  Who is east/west
would be autmatically decided, and the current record of each rikishi
in the basho would also be automatically supplied.  In most cases, the
server would know the record of prior meetings, but if these rikishi
had not met in the past eight basho for which I have data, the user
would be prompted for this information.

A sample form is at http://www.scgroup.com/sumo/upcoming.html
This is a non-working demo; code to process the data will be written
if I receive a positive response from those of you who would use it.
You will need a browser with forms and tables to see this
(e.g. Netscape).

Entering match results is easy, since the system already knows both
combatants for a given match.  The user simply checks the winner and
selects the kimarite.  The window shows the most common twelve
techniques, which cover 90% of all matches; the remainder are
available by scrolling.  See demo form at
http://www.scgroup.com/sumo/results.html

This system may be quite inconvenient for Japanese users, as it is all
in English and Romaji.  I could fairly easily display shikona in
Kanji, thanks to Kobayashi Toshiharu's excellent pages in both
languages ( http://www.bekkoame.or.jp/~tosiharu ), which can be
"joined" to make a translation table.  But I know almost no Japanese,
so I could not translate the instructions, error messages, etc.
However, the script would be freely available to anyone wishing to
make this enhancement (or other changes).

Please let me know what you think of this scheme, whether or not you
would use it, and what improvements might be added.


RETRIEVING DATA FROM THE NHK BROADCAST

The basic idea is to use a video capture card and optical character
recognition software to read the "electronic scoreboard" and related
information.  Unfortunately, this is a much bigger programming task
than I am prepared to tackle.  However, if anyone knows of OCR
software which handles Kanji and could be adapted to video input, that
would be a big start.


TAPPING OFFICIAL RESULTS

I have no idea how the Sumo Kyokai presently records and distributes
schedules and results, but I suspect it is largely manual.  I have
seen errors on both the Japanese and English schedules given to
spectators at the Kokugikan, and the crib sheets used by NHK cameramen
seem to be hand written.  I am hoping (dreaming?) that in exchange for
a good record keeping and match planning package, they would allow a
feature that tranmitted the data elctronically to us (as well as the
press).  This would permit us to follow lower-ranked rikishi, whose
scores are otherwise unavailable.  I don't know how to begin on this
project; does anyone have more details or have contacts with whom one
could begin a dialogue?