Agent Relations

From DBpedia Mappings
Revision as of 15:17, 19 March 2016 by VladimirAlexiev (talk | contribs) (Created page with "Let's analyze all dbo:Agent relations in EN dbpedia, and see if any rationalization is warranted. = Finding Relations = The first task is to find all agent relations == Sch...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

Let's analyze all dbo:Agent relations in EN dbpedia, and see if any rationalization is warranted.

Finding Relations

The first task is to find all agent relations

Schema Exploration

I started doing this by querying the schema:

select ?d ?p ?r {
    ?p rdfs:domain ?d; rdfs:range ?r.
    ?d rdfs:subClassOf dbo:Agent.
    ?r rdfs:subClassOf dbo:Agent.
}

This returns a nice listing of domain-property-range:

d p r
dbo:Person dbo:school dbo:EducationalInstitution
dbo:Artist dbo:mentor dbo:Artist
dbo:SoccerPlayer dbo:trainerClub dbo:SportsTeam

However, it is incomplete: only 35 relations (at another query time 49), but there are a lot more.

Data Exploration

So then I turned to iterative data exploration: starting from the above list, getting a few sample triples, then adding to the list.

select * {
  ?s a dbo:Agent.
  ?s ?p ?o.
  ?o a dbo:Agent.
  filter (?p not in (
    # Already discovered relations
))} limit 10

dbo vs dbp

In the list of relations you'll see many raw props dbp: that duplicate the corresponding mapped property dbo:.

Aren't they redundant, since each dbp: is supposed to be present as dbo:? Yes, it is supposed to be mapped, but not all are. Let's check:

select * {
  ?s a dbo:Agent.
  ?s dbp:manager ?o.
  ?o a dbo:Agent.
  filter not exists {?s dbo:manager ?o}}

This returns many pairs, eg dbr:Chelsea_F.C. and dbr:Jos%C3%A9_Mourinho. Let's check Wikipedia:

{{Infobox football club
| chairman      = [[Bruce Buck]]
| owner         = [[Roman Abramovich]]
| manager       = [[Guus Hiddink]]
| mgrtitle      = Interim manager

(I guess Mourinho was replaced recently, and you can also see the difference between chairman, manager and owner). So let's check the mapping:

All seems to be in order, so let's do an extraction sample:

<http://en.dbpedia.org/resource/Guus_Hiddink> is in there: looks like the mapping works. Not sure why dbo:manager is missing then.

(Note: there is strange tabular data, see mappings-tracker#77

CareerStation

dbo:CareerStation is an unusual sort of Agent: it's an intermediate node with details of the position, eg:

dbo:Person dbo:careerStation dbo:CareerStation dbo:team dbo:SportsTeam.

List of Agent Relations

Org->Person: Leader

- dbo:manager, dbp:manager, dbo:chairman, dbp:chairman, dbo:president, dbp:president, dbo:ceo, dbo:chairperson, dbo:head, dbo:leader, dbp:leader, dbp:currentPresident, dbo:secretaryGeneral, dbo:superintendent, - dbo:trustee, - Academic: dbo:rector, dbp:rector, dbo:chancellor, dbp:chancellor, dbo:provost, dbp:provost, dbo:headteacher, dbo:actingHeadteacher, dbo:executiveHeadteacher, dbo:custodian, dbp:headOfSchool, dbo:dean, <http://dbpedia.org/property/president&Dean>, - Military: dbo:fourthCommander, dbp:currentCommander, dbo:officerInCharge, - Sports: dbp:chief1Name, dbo:trainer, dbo:trainerClub, dbo:managerClub, dbo:formerCoach, dbo:coachedTeam, - Religion: dbo:chaplain,

Org->Person: Founder, Owner

- dbo:foundedBy, dbo:founder, dbp:founder - dbo:owner, dbp:owner

Org->Person: Other

- Sports: dbo:club, dbp:club, dbo:team, dbp:team, dbo:clubsRecordGoalscorer - dbo:nobelLaureates, dbo:notableCommander, dbo:notableStudent, dbo:choreographer, dbo:formerChoreographer, dbo:dubber,

Person->Org

- dbo:employer, - Politics: dbo:party, dbp:party, dbo:federation, - Sports: dbp:clubs, dbp:youthclubs, dbp:currentclub, - Academic: dbo:affiliation, dbp:affiliation, dbp:affiliations, dbp:academicAffiliation, dbo:almaMater, dbp:almaMater, dbo:alumni, dbp:alumni, dbo:school, dbo:university, dbo:student, dbo:training, dbo:education,

Org->Org

- Part: dbo:subsidiary, dbo:childOrganisation, dbp:branch, dbo:militaryBranch, dbo:commandStructure, dbo:parentOrganisation, - Sports: dbo:league, dbp:league, dbo:athletics, dbp:athletics, dbp:conference, - Academic: dbo:accreditation, dbp:accreditation, dbp:mainFeederSchoolFor, - Politics: dbo:mergedIntoParty, dbo:house, dbp:house, dbp:houses, dbo:politicalPartyOfLeader, - Religion: dbo:denomination, dbp:denomination, dbo:religion, dbp:religion, - Various: dbo:distributingCompany, dbo:distributingLabel, dbp:oversight,

Person->Person: Familial

- dbo:parent, dbp:parents, dbo:sibling, dbo:spouse, schema:spouse, dbo:relative, dbo:relation, dbo:partner, dbo:currentPartner, dbo:formerPartner, dbo:friend,

Family->Person

- dbo:primogenitor, dbo:familyMember, dbo:lastFamilyMember, dbo:headOfFamily,

Person->Person: Professional

- dbo:influenced, dbp:influenced, dbo:influencedBy, dbo:colleague, dbo:academicAdvisor, dbp:doctoralAdvisor, dbo:doctoralAdvisor, dbo:student, dbo:doctoralStudent, dbo:mentor, - dbo:copilote, dbo:collaboration, dbo:seiyu, dbo:coemperor, dbo:usurper, dbo:cousurper,

Succession

- dbp:before, dbp:after, dbp:predecessor, dbp:predecessor, dbo:successor, dbp:successor,

Various

- dbo:beatifiedBy, dbo:canonizedBy, - dbo:opponent, dbo:detractor, dbo:rival, dbp:rival, dbp:rivals, dbp:sisterSchool, dbo:performer, dbo:associatedAct, dbp:mascot

Non-Relations

These are derived relations, or occur between Agents only coincidentally (eg when an Agent is also considered a Location)

Location

- schema:containedIn, gn:parentFeature, gn:parentCountry, - dbo:locatedInArea, dbo:location, dbo:county, dbo:province, dbo:district, dbp:district, dbo:settlement, dbo:city, dbo:country, dbp:country,

Generic/wrong

- owl:sameAs, rdfs:seeAlso, owl:differentFrom, - dbp:teamname, dbp:nickname, dbp:title, dbp:name, dbp:free, dbo:type,

Derived

- wd:P118, wd:P286, wd:P112, wd:P131, wd:P17, wd:P184, wd:P69, wd:P102, wd:P26, wd:P31, wd:P54, wd:P140, - dul:coparticipatesWith, dul:sameSettingAs, dul:hasLocation, dul:isClassifiedBy, dul:isSettingFor, dul:hasSetting, dul:isExpressedBy, dul:isPartOf, dul:hasPart, dul:isMemberOf,