W3C home > Mailing lists > Public > public-vocabs@w3.org > June 2014

Re: 2014 Sports Proposal - V3

From: Peter F. Patel-Schneider <pfpschneider@gmail.com>
Date: Wed, 18 Jun 2014 05:42:41 -0700
Message-ID: <53A18941.2040703@gmail.com>
To: "Jason Johnson (BING)" <jasjoh@microsoft.com>, PublicVocabs <public-vocabs@w3.org>
CC: Tom Grahame <tom.grahame@bbc.co.uk>, "paul@xmlteam.com" <paul@xmlteam.com>, Alice Swanberg <alices@yahoo-inc.com>
I have been asked to suggest answers to the questions and comments that I
raised yesterday, so here goes.

I think that the treatment of sports in OpenCyc is a very good start, and
this treatment answers most of the questions I asked and addresses most of
the concerns I raised.  There are a few missing bits in this treatment, but
I think that they can be easily added.

For those who do not want to delve through OpenCyc, here is a quick summary
of the general organization of sports in OpenCyc with some examples added.
Note: Camel-case strings are identifiers taken from some unidentified
namespace, i.e., not all of these identifiers exist in OpenCyc, although
most do.


The organization of sports in OpenCyc starts with SportsEvent, which is an
event involving a sport designed produce a winner or have a draw declared,
and thus competitive.  (This really needs to be expanded to have a full
definition of results of a sporting event.)  A SportsEvent, like any
HumanActivity, has participants, which are people or organizations over
people.

Then a Sport is defined as a collection of some sports events.  (Yes, this
is somewhat circular, but it still works out OK.)  A central part of a sport
is an AthleticActivity, which rules out non-athletic games, e.g., chess,
even though playing competitive chess better requires athletic training and
chess is recognized as a sport by certain sports governing bodies.  This
formulation of Sport is completely agnostic as to the granularity of a
sport, so Mens100mBreaststroke can a sport as well as AquaticSports.
However, and vitally, a Sport is not a string.  Sports can be identified by
well-known URLs, e.g., https://en.wikipedia.org/wiki/Hurling.  In schema.org
"strings as things" could be effectively used to guess as to which sport is
being identified, e.g., https://en.wikipedia.org/wiki/Ice_hockey for
"hockey".

ProfessionalSport is introduced primitively in OpenCyc.  This is acceptable,
because most professional sports are like ProAmericanFootball.  (However,
OpenCyc mistakenly states that the sport consisting of games played
according to the rules of FIFA is a professional sport.)  Better would be to
define a professional sport something like a sport where at least one of the
participants in each sporting event is a professional athlete or
professional sports organization.

TeamSport is introduced as a sport that is "usually" engaged in by teams.
Tennis then defined to be a team sport, so it would be better to change the
"usually" to something like "routinely".

Certain sports could be stated to be OlympicSports (e.g, OlympicAquatics, a
sub-class of AquaticSports), others could be OlympicDisciplines (e.g.,
OlympicAlpineSkiing, a sub-class of AlpineSkiing), and others could be
OlympicEvents (e.g., Olympic100mRunning, a sub-class of RacingSport).
Certain sports are those sanctioned by a governing organization, e.g., FIFA
Football.  Other sports are those played by sports leagues, e.g., NFL
football.  Yet other sports are those played according to a particular set
of rules, e.g., Association football.  Yet more other sports are just
convenient groupings, e.g., American football or rugby or shinny.

The rest of the treatment builds on this core.  A SportsCompetition consists
entirely or largely of one or more sports events in a single sport.
SportSeason is a specialization of SportsCompetition, as is
SportsFormalCompetition.  A SportsFormalCompetition produces a partial order
of its participants and can award prizes (e.g., olympic gold medals) or
trophies (e.g., the Stanley Cup).  Most professional sports seasons belong
to SportsFormalCompetition, but not all sports seasons do, e.g., American
college football seasons (because they include several divisions) or
Division 1 FBS American college football seasons (because there is as of yet
no formal winner determined in Division 1 FBS football).

A summer olympic games can then be a SportsCompetition with its sport being
the supersport containing all summer olympic sports and its teams being the
national olympic teams, as well as having SportsFormalCompetitions for the
various summer olympic events.

A SportsOrganization is an organization mainly involved with SportsEvents.
SportsTeam is a group of people and a SportsOrganization.  A SportsTeam has
sportsTeamPlayers, participates in a Sport, and can compete in a
SportsLeague.  Sports teams have coaches as well, introduced as those who
instruct or train the players.  (I don't think that this is a very good
notion for coach here.)  ProfessionalSportsTeam consists of those teams
whose players are paid a salary for being a member of the team.

A SportsLeague is a SportsOrganization.  An NFLTeam would then be a team
that competes in the NationalFootballLeague.  A SportsLeague has a
focalActivityType that is a sport.  As any organization, SportsLeagues can
have suborganizations, such as conferences and divisions.

Sporting events can have a homeSportsTeam, stated to be the geographically
local team, and a visitingSportsTeam.  As stated earlier OpenCyc does not
have a complete notion of the results of a sports event, so that needs to be
added as well.  OpenCyc also does not have much of a notion of a sports
governing body.


To make this all work out well requires both the higher-order capabilities
of CycL, as sports are both classes and values, and some reasoning engine,
to make the many inferences that should fall out of various statements,
e.g., that a SportsTeam that competes in a conference also competes in the
league.  Reducing the expressive level of the language and eliminating the
inferential support is possible, I think, but some compromises might have to
be made.
Received on Wednesday, 18 June 2014 12:43:14 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:42 UTC