=Paper=
{{Paper
|id=Vol-2532/paper1
|storemode=property
|title=Digital Edition Publishing Cooperative for Historical Accounts and the Bookkeeping Ontology
|pdfUrl=https://ceur-ws.org/Vol-2532/paper1.pdf
|volume=Vol-2532
|authors=Christopher Pollin
|dblpUrl=https://dblp.org/rec/conf/rodbh/Pollin19
}}
==Digital Edition Publishing Cooperative for Historical Accounts and the Bookkeeping Ontology ==
https://ceur-ws.org/Vol-2532/paper1.pdf
T. Riechert, F. Beretta, G. Bruseker (Ed.) RODBH 2019,
Proceedings of the Doctoral Symposium on Research on Online Databases in History 2019 7
Digital Edition Publishing Cooperative for Historical
Accounts and the Bookkeeping Ontology
Christopher Pollin1
Abstract: The Project ”Digital Edition Publishing Cooperative for Historical Accounts”, a Andrew
W. Mellon funded cooperation of five US partners and the Centre for Information Modelling at Graz
University, aims to link the knowledge domain of economic activities to historical accounting records.
For this purpose the so-called Bookkeeping Ontology is developed. DEPCHA creates a publication
hub for digital editions on the web. It converts multiple formats into RDF and publishes these in
combination with the associated transcriptions. DEPCHA also allows the usage of retrieval and
visualization functionalities, as well as interoperability and reuse of information in the sense of Linked
Open Data.
Keywords: Web of Data; GAMS; Historical Financial Records; Bookkeeping Ontology; Knowledge
Domain; History; Digital Humanities; Semantic Web; Linked Open Data
1 Introduction
On the first of August 1808 James Haley purchased ¼ lb of powder, 1 lb of shot and 1 lb
of sugar for the price of 2 shilling and 6 pence each from Stagville Plantation in North
Carolina (USA). We can find information like that, and numerous similar ones, in historical
financial records. In the 1980s, two groups emerged which applied different approaches
to tackle such historical sources: the ”traditionalists” and the ”quantifiers”. While the
”traditionalists” used a hermeneutic approach to historical sources, the ”quantifiers” tried to
formally describe and evaluate the historical dimensions, in order to support an intuitive
process of understanding with empirically identifiable facts [JAT85]. This divide tells us
that, when using formal methods on historical data, research should distinguish between the
representation of the original source and its interpretation. The latter is the core knowledge
domain of historical research. It is advised to share the basic assumptions and definitions in
a knowledge domain in a formal way [Th17]. In this context, the Web of Data (aka Semantic
Web) and Linked Open Data are central concepts that offer technologies going hand in
hand with that new understanding of historical research. The hermeneutic method and the
method of transforming historical phenomena into formal models, as well as its connections
to other domains and its reuse by the scientific community, makes the work of historians
more dynamic and comprehensible.
1 University of Graz, Centre for Information Modelling - ACDH, Elisabethstraße 59/III, 8010 Graz, Austria
christopher.pollin@uni-graz.at
Copyright © 2019 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). c b
8 Christopher Pollin
The Project ”Digital Edition Publishing Cooperative for Historical Accounts” (DEPCHA),
a Mellon funded cooperation of five US partners and the Centre for Information Modelling
at Graz University, aims to link the knowledge domain of economic activities to historical
accounting records. After the discussion about common entities in historical financial
records in the beginning of this paper, the second part focuses on the formalization of these
entities within the Bookkeeping Ontology. The third section defines a workflow to publish
RDF data, as part of the digital editions of historical financial records, as Linked Open Data.
In conclusion, future challenges and results concerning ontology engineering, retrieval and
visualization functionalities of the web prototype2 are discussed briefly.
2 Historical Financial Records and Relevance
Historical financial records provide rich and highly structured data sets over long periods,
containing substantial amounts of individual information. This individual information is
often not in the core of research interest. Instead, the records acquire their significance in
aggregation of the single entries. Pure transcription does not cover the full range of dimen-
sions of such a source: the linguistic/textual, the quantifiable and the semantic dimension
[Vo15], [Vo16]. For research purpose, historical sources are subject to a transformation
process towards (linked) information sources that can be used in various research scenarios.
In order to illustrate this we will discuss three case studies of project partners and their
respective research interests, which go far beyond economic and administrative aspects.
The George Washington Financial Papers (1748-1799)3 gives insight into the life of George
Washington and other topics such as the material culture, social history, manufacturing
and agriculture. The financial papers exist as digital edition, created and published via a
Drupal4 based editorial platform, and aim to make Washington’s records freely accessible.
The platform allows editing and publishing financial documents and gives the users the
possibility to perform simple analytical functionalities. Samples of research questions that
could be of interest to historians are: How much money did Washington spent annually
and for which specific commodities? Which role slave trade plays in his business? How
did the price of certain commodities fluctuate? What did the network of partners look like
and who did business with him? How was the value of tobacco calculated through different
currencies [St14]?
The Wheaton Accounts (1828-1859) contain a daybook5 of a store selling commodities of
daily life. The digital edition follows a TEI/XML approach. It extends the range of questions
to historical narratives and geographical information. It is interesting to follow an individual
or a family as they appear in the daybook over time and reconstruct their social background
2 DEPCHA Prototype, https://gams.uni-graz.at/depcha
3 The George Washington Financial Papers Project, www.financial.gwpapers.org
4 Open-source content management framework, www.drupal.org
5 Daybook of L.M. Wheaton’s Store, expenses of building houses and barns, and expenses of constructing
Wheaton Female Seminary buildings, http://hdl.handle.net/11040/17982
DEPCHA and the Bookkeeping Ontology 9
for a historical narrative. The same applies to geographic information allowing to track
geographic relationships of people or the origin of commodities [TB13].
The digitization project of the Stagville Financial Papers (1767-1892), including daybooks
and ledgers from the Stagville plantation store in North Carolina, follows an open science
and crowdsourcing approach using From the Page6 to transcribe and encode the material.
Research questions in this context include the numbers and connections of customers, as
well as commodities, which ”go together”. Furthermore, economic dependencies (who is in
debt of whom) as well as the social status of customers (e.g. free/enslaved) are of interest
[BA15].
3 Methods
Common structures can be drawn from the research questions mentioned above. To do so,
data must be prepared and structured according to a formal set of rules. The Bookkeeping
Ontology7, a conceptual data model based on the REA [Mc82] model and CIDOC CRM8,
is developed in an ontology engineering process, involving historians, software developers
and digital humanists. The ontology is published in a stable version in GAMS9 [SS18] and
in OntoMe10 and can be further discussed by the scientific community.
The Bookkeeping Ontology formalizes the interpretation of a transaction (bk:Transaction)
as combination of transfers (bk:Transfer) of measurable objects (bk:Measurable) from one
accounting object (bk:Between) to another. bk:Between defines an abstract class, which
unites bookkeeping categories (bk:Accounts, e.g. a cash account) and actors (individual
bk:Party e.g. Washington or an unknown group of individuals bk:Group e.g. four farmers).
Its physical representation in a historical source is an entry in a written accounting record
(bk:Entry). The bk:Entry is an information fragment of a bk:Transaction often naming
only one party, while the other party is implicit in the textual context of the entry. Further
information on the temporal (bk:when), spatial (bk:where) dimension of a bk:Transaction
as, well as the status (bk:status) of it (”partly paid”), can be expressed optionally. In regard
to one of our research questions named above a bk:Transaction can be assigned to a specific
context. Every transaction consists (bk:consistsOf ) at least one transfer (bk:Transfer). A
single bk:Transfer describes the action of transferring a bk:Measurable in one direction
(bk:from or bk:to). bk:Measurable is defined as everything that can be quantified. It has
subclasses for economic goods (bk:EconomicGood, as labor: bk:Service or as physical
things: bk:Commodity) and money (bk:MonetaryValues). bk:Measurable is describe by
its quantity (bk:quantity) and the unit of calculation (bk:unit). The bk:Entry is described
by the transcription fragment of the original source (bk:text). bk:EconomicGoods can be
6 Crowdsourcing manuscript transcription plattform, www.fromthepage.com
7 Bookkeeping Ontology in DEPCHA, https://gams.uni-graz.at/o:depcha.bookkeeping
8 Semantic framework for mapping cultural heritage information, www.cidoc-crm.org
9 Humanties Asset Management System at Graz University, https://gams.uni-graz.at
10 Ontology Management Environment, ontologies.dataforhistory.org
10 Christopher Pollin
categorized (what is measured) and can be assigned a price. A bk:Transfer can be carried
out by (bk:by) someone who conducts the transfer process in place of the business partner
(bk:Agent). When writing it down into the ledger, accounting categories (bk:debit and
bk:credit) are coded optionally.
The DEPCHA web prototype is realized in the GAMS infrastructure11, an open source,
FEDORA12 based digital repository for storing and publishing data in the humanities.
Digital objects, containing multiple data streams and methods, allow disseminating data via
HTML, as archival data in XML, and via various APIs. Furthermore, GAMS implements a
disseminator for RDF data via the triplestore Blazegraph.13 An encapsulated query object
stores predefined SPARQL queries including fulltext search over the RDF data sets. This
provides data for retrieval, analysis and visualization [St18].
4 From Digital Edition to Linked Open Data
TEI/XML serves as an interchange format for the data from different systems (Drupal, From
the Page, or generic CSV). Historical financial records are transcribed and annotated in these
respective formats or directly in TEI/XML. TEI allows structuring the textual dimension of
a source, marks up text-specific phenomena and normalizes places, dates or persons. The
semantic relations covered by the Bookkeeping Ontology are inserted into the XML/TEI
via attribute @ana. It allows a global, multiple and outgoing annotation of any structure
in the TEI markup. During ingest into GAMS, XSLT extracts the annotated structures in
the XML/TEI and transforms it into RDF. The repository stores this data in the triple store.
This approach has already been successfully applied in other projects regarding historical
sources, like the Municipal accounts of the city of Basel 1535-161114, the Urfehdebücher of
the city of Basel - digital edition15 [PV17]. The following example illustrates the workflow
from the historical source to RDF data. Following figure represents the origin of an entry in
the Stagville Accounts.
Fig. 1: Entry in the Stagville Accounts
11 GAMS and Cirilo Client: Policies, documentation and tutorial, https://gams.uni-graz.at/o:gams.doku
12 Flexible Extensible Digital Object and Repository Architecture, https://duraspace.org/fedora
13 Triplestore and graph database, https://www.blazegraph.com
14 Municipal accounts of the city of Basel 1535-1611, https://gams.uni-graz.at/srbas
15 Urfehdebücher of the city of Basel - digital edition, https://gams.uni-graz.at/ufbas
DEPCHA and the Bookkeeping Ontology 11
In addition to the table structure of this example, historical financial records can also
be represented in other textual structures, such as continuous text or lists. To make the
connection to the Bookkeeping Ontology comprehensible, referencing concepts are identified
by the bk-prefix in the XML annotation. The entry in the sample above can be interpreted
as follows: The person James Haley (bk:Party) buys ¼ lb of powder, 1 lb of shot and 1 lb of
sugar (bk:EconomicGood) from Stagville (bk:Party) and transfers in return the monetary
value of 7 shilling and 6 pence to Stagville (bk:MonetaryValue). Some information is not
explicitly mentioned in a bk:Entry and could be found in the header or is known by the editor
of the source. As semantics are defined through the attribute @ana, the textual structure of
the TEI document is not relevant for further processing and the full expressiveness of TEI
can be used.
The starting point is to define a container for a bk:Transaction by annotating the textual
representation of an bk:Entry by ana=”bk:entry”. The following simplified XML/TEI
snippet illustrates this very entry and its conceptual counterpart bk:Transaction, which
consists of two bk:Transfers. One of them transfers powder, sugar and shot, the other
bk:Transfers transfers 7 shilling and 6 pence.
List. 1: Simplified XML/TEI-Snippet
Stagville
August 1st 1808
69 James Haley
Self Dr.
|
1/4 lb Powder 2/6
|
7
|
6
|
12 Christopher Pollin
To elaborate the given code further, I am going to explain the entries in detail. Besides
trivial markup, e.g. the date of the entry using , the flow of money
is expressed through the values ana=”bk:from” and ana=”bk:to”. Since in most sources
of this kind a purchase of a commodity is documented, this is an assumption. The flow
of economic goods - a commodity or service - can then be deduced from this conclusion.
The single entry usually names only one party in the transaction, while the shopkeeper
Stagville is already identified in the heading of the account or as external archival knowledge.
Furthermore, the text ”Self Dr.” in the header of the entry can be interpreted as transaction
and booking status and expresses that the entry was made by that person on the debit side of
a double entry style ledger.
Another important aspect is the normalization of entities such as persons, places, currencies,
commodities or services, which can become very complex in the context of such a source. For
example, certain weight units or currencies are affected by regional and temporal differences.
Due to the large variety of context sensitive entities, normalization of the data is achieved by
linking to Wikidata.org or by extending workflows for other LOD vocabularies, which seems
useful. The use of Wikidata, however, allows to create independent new concepts, which
are not represented in any vocabulary yet. The following example illustrates this: another
dataset ingested in DEPCHA contains lists of goods from the city of Regensburg in the 13th
century.16 In this data set the term ”Scheffen Hafer” is used and refers to a quite specific
temporal and regional historical unit, a bushel of oat. The workflow described above allows
to directly connect such terms to Wikidata.org concepts by adding the Q-number directly
to an attribute. This pragmatically approach is supported by tools like OpenRefine17 for
semi-automatic identification. As a result of this workflow an XSLT extracts the annotated
structures in the TEI to RDF - the following RDF/Turtle snippet shows the processed
XML/TEI:
List. 2: Simplified RDF/Turtle
@ p r e f i x bk : < h t t p s : / / gams . u n i − g r a z . a t / o : d e p c h a . b o o k k e e p i n g #> .
@ p r e f i x d e p c h a : < h t t p s : / / gams . u n i − g r a z . a t / d e p c h a #> .
@ p r e f i x wd : < h t t p s : / / www. w i k i d a t a . o r g / w i k i / > .
< d e p c h a : T r a n s a c t i o n −0> a bk : T r a n s a c t i o n ;
bk : c o n s i s t s O f < d e p c h a : T r a n s f e r −1 > , < d e p c h a : T r a n s f e r −2> ;
bk : when "1808 −08 −01" ;
bk : t e x t " 1 / 4 l b Powder 2 / 6 1 l b S h o t 2 / 6 7 6" .
< d e p c h a : T r a n s f e r −1> a bk : T r a n s f e r ;
bk : t r a n s f e r s < d e p c h a : M e a s u r a b l e −1> ;
bk : from < d e p c h a : s t a g v i l l e > ;
bk : t o < d e p c h a : p e r s .2 > .
16 StadtA Regensburg, Cameralia 3
17 Open source desktop application for data cleanup and transformation, https://openrefine.org
DEPCHA and the Bookkeeping Ontology 13
< d e p c h a : T r a n s f e r −2> a bk : T r a n s f e r ;
bk : t r a n s f e r s < d e p c h a : M e a s u r a b l e −2−1> ;
bk : from < d e p c h a : p e r s .2 > ;
bk : t o < d e p c h a : s t a g v i l l e > .
< d e p c h a : M e a s u r a b l e −1> a bk : Commodity ;
bk : u n i t ;
bk : q u a n t i t y " 0 . 2 5 " ;
bk : commodity ;
bk : p r i c e < d e p c h a : P r i c e −1> ;
bk : t e x t " 1 / 4 l b Powder " .
< d e p c h a : M e a s u r a b l e −2−1> a bk : Money ;
bk : u n i t ;
bk : q u a n t i t y " 7 " .
< d e p c h a : p e r s .2 > a bk : Between ;
bk : name " James H a l e y " .
5 Conclusion
This project work has shown that workflows can be defined transferring historical financial
records, and its annotations according the Bookkeeping Ontology, from different data
sources, via a standardized XML/TEI, to RDF. The outcome of this process enables us
to implement basic functionalities when dealing with entities in the sources and make
digital editions on a shared platform - the DEPCHA prototype - available. In addition to
very general structures within transaction processes in historical financial documents, other
source-specific phenomena have been identified, which the ontology has not yet been able
to deal with. An example for this is the annotation of unspecified groups of participants,
such as four farmers and a baker, who equally pay a certain tax. In an iterative ontology
engineering process, comparable to a hermeneutic circle, the project team discussed such
phenomena during so-called ”deep-dives”, trying to understand the sources together to
further improve our knowledge domain. In this particular case bk:Group and bk:Tax were
added to the ontology.
The advantages in summary: a common vocabulary makes communication in the project
team easier and it can be reused by other projects. Through the reuse of the model
the development process is promoted, in which existing concepts can be criticized and
missing concepts can be added. This leads to a more widely accepted and comprehensible
conceptual model. At the data level in the sense of LOD, references to Wikidata.org support
interoperability and further use of the data in different scientific disciplines. Synergies arise
14 Christopher Pollin
in particular when this data can be used for the implementation of generic information
retrieval, analyzation and visualization functionalities, facing project-specific use cases.
Such functionalities allow us to use larger amounts of data more efficiently. The prototype
has shown that the JavaScript library d3.js can be used to implement such functionalities
based on a SPARQL result for the above-mentioned use cases. A future challenge in this
endeavor comes with the difficulty to normalize classifications of currencies, goods and
services. The assignment of these classifications to superordinate unified platforms, such
as Wikidata.org, or independent taxonomies in the teiHeader may allow a higher level of
normalization, even though this may still be work in the future.
References
[BA15] Brumfield, Ben; Anna, Agbe-Davies: , Encoding Account Books Relating to Slavery in the
U.S. South. https://medea.hypotheses.org/182, 2015. Accessed: 2019-03-20.
[JAT85] Jarausch, Konrad Hugo; Arminger, Gerhard; Thaller, Manfred: Quantitative Methoden in
der Geschichtswissenschaft. Wissenschaftliche Buchgesellschaft, 1985.
[Mc82] Mccarth, W.: The REA accounting model: A generalized framework for accounting systems
in a shared data environment. Accounting Review, pp. 554–578, 1982.
[PV17] Pollin, Christopher; Vogeler, Georg: Semantically Enriched Historical Data. Drawing on the
Example of the Digital Edition of theÜrfehdebucher der Stadt Basel". In: WHiSe@ ISWC.
pp. 27–32, 2017.
[SS18] Stigler, Johannes; Steiner, Elisabeth: GAMS–Eine Infrastruktur zur Langzeitarchivierung
und Publikation geisteswissenschaftlicher Forschungsdaten. Mitteilungen der Vereinigung
Österreichischer Bibliothekarinnen und Bibliothekare, 71(1):207–216, 2018.
[St14] Sterzer, Jennifer: Working with the Financial Records of George Washington: Document vs.
Data. Digital Studies/Le champ numérique, pp. 554–578, 2014.
[TB13] Tomasek, Kathryn; Bauman, Syd: Encoding financial records for historical research. Journal
of the Text Encoding Initiative, (6), 2013.
[Th17] Thaller, Manfred: Historical Information Science: Is there such a Thing? New Comments
on an old Idea [1993]. Historical Social Research/Historische Sozialforschung. Supplement,
pp. 260–286, 2017.
[Vo15] Vogeler, Georg: Warum werden mittelalterliche und frühneuzeitliche Rechnungsbücher
eigentlich nicht digital ediert? In (Baum, Constanze; Stäcker, Thomas, eds): Grenzen und
Möglichkeiten der Digital Humanities. 2015.
[Vo16] Vogeler, Georg; Jahnke, Carsten; Gleba, Gudrun; Cordes, Albrecht; Franzke, Cordula A;
Laczny, Joachim; Würz, Simone: The Content of Accounts and Registers in their Digital
Edition. XML/TEI, Spreadsheets, and Semantic Web Technologies. In (Sarnowsky, Jürgen,
ed.): Konzeptionelle Überlegungen zur Edition von Rechnungen und Amtsbüchern des
späten Mittelalters, pp. 13–41. Göttingen, 2016.