Upload
john-bradley
View
176
Download
1
Tags:
Embed Size (px)
Citation preview
A Semantic Web understanding
of the factoid prosopography
model John Bradley
Department Digital Humanities
King's College London
Ontologies for Prosopography:
Who's Who? or, Who was Who?
Workshop, July 8, 2014 at the
DH2014 conference, Lausanne
Switzerland
twitter: #KingsDH
@kingsdh
1
Tim Berners-Lee on Linked
DataAll kinds of conceptual things, they have names now that start
with HTTP.
I get important information back. I will get back some data in a
standard format which is kind of useful data that somebody
might like to know about that thing, about that event.
I get back that information it's not just got somebody's height
and weight and when they were born, it's got relationships.
And when it has relationships, whenever it expresses a
relationship then the other thing that it's related to is given
one of those names that starts with HTTP.
Tim Berners-Lee: Linked Data presentation at TED 2009
2
Tim Berners-Lee on Linked
DataAll kinds of conceptual things, they have names now that start
with HTTP.
I get important information back. I will get back some
data in a standard format which is kind of useful data that
somebody might like to know about that thing, about that
event.
I get back that information it's not just got somebody's height
and weight and when they were born, it's got relationships.
And when it has relationships, whenever it expresses a
relationship then the other thing that it's related to is given
one of those names that starts with HTTP.
Tim Berners-Lee: Linked Data presentation at TED 2009
3
Ontologies and a
“prosopographical” domainVast amount of data is related to
people to various degrees: what is
most properly prosopographical?
Good Ontology design practice: a kind
of “Occam's Razor”.
So what information related to persons
is IN prosopography and what is
OUT?
4
Traditional Prosopography
5
SourcesPeople
From J.R. Martindale, The
Prosopography of the
Later Roman Empire, 3:
A.D. 527-641. Cambridge:
Cambridge University
Press. 1992.
Places
Prosopography and Identity of
PersonsAn important part of prosopography is in the identifying of personsWhat's constitutes an historical person's identity?
Formal URIs provide a part of the linked data answer.◦ http://db.poms.ac.uk/record/person/749/
However, historical identity, by its nature of being contestable, requires more than this:◦ Abraham, bishop of Dunblane (fl.1210×14-
1220×25) (id 749)6
Prosopography: more than “just”
person identificationHistorical persons survive for us through their
appearance in sources, and historians identify
them not only by their name, but also by what they
did and by other ways that they are described.
7
Information about persons is a
part of prosopography.Assertions about a person have traditionally formed the basis of prosopography.
One could argue that historical people from before the immediate past only "survive" in our memory through their presence in sources: what sources assert about them.
Arguments about their identity flow from what these sources say to us.
8
Identity as a Contestable
thingHistorical Prosopography must be contestable, since it deals in information which is often uncertain.
◦ About Person (I think your A is my (slightly different) B)
◦ About statements about a person (I think this statement belong to a different person, or mis-interprets the sources)
◦ After date/timing (I think this event happened at a different time)
Thus, simply naming a person A, and then simply asserting that s/he is different from someone else's person B is not sufficient!!
The identification has to not only be made, it has to be testable and contestable.
To make such assertions testable, one needs data about the person.
This is Berners-Lee's 2nd point!
9
Data in a standard format for
prosopography?"I get important information back. I will get back
some data in a standard format which is kind of
useful data that somebody might like to know about
that thing, about that event.“ (Berners-Lee: TED
2009)
What is the data in a "standard format"?
Mainstream Linked Data, take this to mean data in
RDF
◦ highly structured.
What is a representation about a person then that
is highly structured?
A set of Assertions10
Core structure for DDH’s
Prosopographical databases: "factoid
model"
Person
Assertion
(factoid)
Authority Lists
Assertion Type
Source
Location Possession
11
Instance of
Typed by
Connected toConnected to
Appears in
Connected to
Role / Name
Date
Structuring Prosopography: the
factoid
Pasin, Bradley (2011). Factoid-
based Prosopography and
Computer Ontologies: towards
an integrated approach
12
Factoid Assertions are …
Time dependant◦ PoMS's Abraham 749's identity seems to be
wrapped up in his having been Bishop of Dunblane.
◦ Even so, an assertion such as "Abraham (ID 749) was Bishop of Dunblane" is actually time dependant
Material about him might well exist from times before or after he was Bishop of Dunblane.
Source Driven:◦ The factoid model aims to present what the
sources are saying, and downplays what the modern day prosopographers, as historians, believe.
13
Source Assertion: An Act of Document
Interpretation
14
Pasin, Bradley (2011). Factoid-
based Prosopography and
Computer Ontologies: towards
an integrated approach
Martindale asserts that... “Greg. Tur HF” asserts that... Victorius 4 imprisoned Eucherius 4
“Two levels” of assertion
Problems: Agent and Event
driven approachesHistorical persons are not (always)
agents
◦ Oxford English Dictionary: 2: A person or thing that takes an active role or produces a
specified effect.
2.1: (Grammar) The doer of an action, typically expressed a the
subject of an active verb of in a by phrase with a passive verb.
◦ Merriam-Webster:
1. one that acts or exerts power
Assertions are not always about
events15
TEI: Personography: “Basic
Principles”Information about people, places, and
organizations, of whatever type, essentially
comprises a series of statements or
assertions relating to: ◦ characteristics or traits which do not, by and large, change
over time
◦ characteristics or states which hold true only at a specific
time
◦ events or incidents which may lead to a change of state or,
less frequently, trait.
TEI, section 13.3.1 (http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ND.html)
16
Factoid types as traits/states and
eventsPoMS/PoNE:
* Titles/Occupation
(state)
* Relationship (trait?)
* Possession (state)
* Transaction (event)
PASE:
* Authorship (trait?)
* Education (trait?)
* Event (event)
* Relationship
(trait?)
* Occupation (state)
* Office (state)
* Status (trait)
* Personal Info (trait)
* Possession (state)
* Transaction
(event)
Charlemagne:
* Attribute/relationship (state)
* Place relationship (*)
* Transaction (event)
* Miscellaneous (event)
PBW(E):
* Family (trait)
* Ethnicity (trait)
* Activity (event)
* Dignity (state)
* Kinship (trait?)
* Possession (state)
Roman Republic:
* Family (trait)
* Relationship (trait)
* Office (state)
17
Trait/State factoid
Factoid
iii.6
"sanctus
anachorita
Dei"
Occupation
Type
"Anchorite"
Person
Cuthbert 1
"Saint; bishop
of Hexam ..."
Source
Anon.VitCuthberti
Location
Farne, island
Northumberland
19
Relationship Factoid
Factoid
1177 x 1185
Relationship
Type
"Father"
Person
Alexander Seton
(father of Philip)
Source
1/6/176
(RRS, ii, no. 200)
Person
Philip of Seton
Role
"father"
Role
"son"
21
Transaction: (Gift)
Factoid
"Gift of land
of Dodin"
1177 x 1185
Transaction
Type
"Gift"
Person
Henry, earl of
Northumberland
and Huntingdon
Source
1/5/2
(RRS, i, no. 106)
Person
Kelso Abbey
Role
Grantor
Person
Dodin of
Duddingston
Role
Previous
holder
Role
Beneficiary
Possession
Land
"land of Dodin"
BWK (Berwickshire)
23
Tim Berners-Lee on Linked
DataAll kinds of conceptual things, they have names now that start
with HTTP.
I get important information back. I will get back some data in a
standard format which is kind of useful data that somebody
might like to know about that thing, about that event.
I get back that information it's not just got somebody's
height and weight and when they were born, it's got
relationships. And when it has relationships, whenever it
expresses a relationship then the other thing that it's
related to is given one of those names that starts with
HTTP.
Tim Berners-Lee: Linked Data presentation at TED 2009
24
Relationships through
transactions
Factoid
"Gift of land
of Dodin"
1177 x 1185
Transaction
Type
"Gift"
Person
Henry, earl of
Northumberland
and Huntingdon
Source
1/5/2
(RRS, i, no. 106)
Person
Kelso Abbey
Grantor
Person
Dodin of
Duddingston
Previous
holder
Beneficiary
Possession
Land
"land of Dodin"
BWK (Berwickshire)
http://db.poms.ac.uk/record/person/90
http://db.poms.ac.uk/record/person/82
http://db.poms.ac.uk/record/person/108
25
A network of Witnesses derived
from PoMS data
Jackson, Cornell (2014). "Using Social Network Analysis to understand
the PoMS database. DH2014, 10 July, 201426
Factoid Data as Semantic Web
Data
D2R Server:
" D2R Server is
a tool for
publishing
relational
databases on
the Semantic
Web. "http://d2rq.org/d2r-server
27
Trait/State factoid as a triple
Factoid
iii.6
"sanctus
anachorita
Dei"
Occupation
Type
"Anchorite"
Person
Cuthbert 1
"Saint; bishop
of Hexam ..."
Source
Anon.VitCuthberti
Location
Farne, island
Northumberland
pasePerson:cuthbert_1 pase:hasOccupation
paseAuthority:Anchorite
28
A reified factoid
Factoid
iii.6
"sanctus
anachorita
Dei"
Occupation
Type
"Anchorite"
Person
Cuthbert 1
"Saint; bishop
of Hexam ..."
pasedata:factoid2568 rdf:type rdf:Statement ;
rdf:subject pasePerson:cuthbert_1 ;
rdf.predicate pase:hasOccupation ;
rdf.object paseAuthority:Anchorite ;
pase.fromSource paseSources:Anon.VitCuthberti ;
pase.associatedLocation paseLocation:farne.island […]
pasePerson:cuthbert_1 pase:hasOccupation
paseAuthority:Anchorite
Source
Anon.VitCuthberti
Location
Farne, island
Northumberland
29
Uncertainty
A data model needs to accommodate
statements of uncertainty. Where and
how?
The factoid model, by being "source
driven" deals with one aspect of this.
An factoid assertion is a claim about
what a source says, not what the
project believes (thus, a little less like
biography).30
Uncertainty
Dates: we use a TEI-like date model
for dates in our factoids
Qualification of statements, probably
at the factoid and factoid/person level:
Certain, Uncertain, Very Uncertain
31
Summary
Identity of an historical person is based on the assertions found in the sources.
The factoid model is designed to reflect this source-driven orientation.
Agent and Event orientation is not the full story for prosopography.
The factoid model can be mapped best to reified RDF statements.
Uncertainty is always an element in history, and needs to be accommodated.
32
... so, what's in
Identifiers for persons
names and forms of names for people
as they appear in sources
assertions about them, as derived
from the sources
models to deal with recording
uncertainty, and complex dating?
34
... what's near by?
Names for onomastic studies?
◦ I'm interested in seeing how name data
fits in in the SNAP-DRGN project!
Places (perhaps linked to, but the task
of formally organising places is a
project in its own right)
Sources (linked to, but not captured or
represented)
35