Content uploaded by Bogumil Szady
Author content
All content in this area was uploaded by Bogumil Szady on Feb 02, 2021
Content may be subject to copyright.
Transactions in GIS. 2020;24:355–370. wileyonlinelibrary.com/journal/tgis
|
355
© 2020 John Wiley & Sons Ltd
1 | INTRODUCTION
The development of historical-geographic information systems (HGIS), which store and process data on the history
of the geographical entities, poses challenges due to not only spatial but also temporal charac teristics. The moti-
vation for this research stems from a research and development projec t titled “Ontological foundations for build-
ing historical geoinformation systems” (ONTOHGIS) aiming to develop an information system for collec ting and
integrating historical information on settlements and administrative units on the Polish lands until 1939 (http://
ontoh gis.pl/).
The starting point for our considerations is a situation when we have only residual information in a historical
source. The historical source is, in our c ase, a medieval manuscript such as a tax register or a court record, or other
source with information about settlements. By “residual” we mean that the information is incomplete, that is to
say, assuming that the complete data record we have designed for a settlement in a dat abase includes a name, a
type, a location (geographical scope), mereological relationships with other settlements, an administrative unit
DOI : 10.1111/tgis .126 09
RESEARCH ARTICLE
Modeling and presenting incomplete and
uncertain data on historical settlement units
Grzegorz Myrda1 | Bogumił Szady1 | Agnieszka Ławrynowicz2
1Institute of History, Polis h Acade my of
Science s, Warsaw, Poland
2Faculty of Computing and
Telecommunicatio ns, Pozna n Univer sity of
Technolog y, Poznań, Pol and
Correspondence
Agnieszka Ławryn owicz, Faculty of
Computing and Telecommunications,
Poznan Universit y of Technology, Poznań,
Poland.
Email: grzemy@gmail.com
Abstract
When entering information about changes of settlements in
time into a database, we are dealing with direct facts from
historical sources and with interpretations. The source infor-
mation generally relates to an event or situation at a specific
time. This means that we only fill in the parts of the timeline
with data objects called (partial) manifestations; however,
lack of data does not mean that settlement did not exist.
It would take a lot of time to interpret and enter missing
information manually. In this article we present a database
schema, database procedures and algorithms to generate
aggregated settlement unit manifestations in an automated
way from their partial manifestations. As a result, it is pos-
sible to obtain a continuous history of a given settlement so
as to present it on a map.
356
|
MYRDA et A l.
to which it belongs, number of inhabitants, etc., at least one of these elements is missing. Usually, at most one or
two are present in a single source. For instance, as we can see in Figure 1, the source may only contain a name.
Regarding historical research methodology, geographical-historical facts are usually divided into simple facts
and complex fact s. The former refer to the historical event and relate directly to the historical source. The latter
refer to the historical process, to a longer period of time, and are the result of the interpretation of a historian
based on simple facts (Pomorski, 1983–1984, pp. 48–49; Topolski, 2012, p. 222).
For the purposes of represent ation in the database of information about settlement s and changes that took
place in their status, we will use the concept of partial and aggregated manifestations, leveraging the concept of
manifestation-based endurantist pattern for set tlement units discussed in Garbacz and Trypuz (2017). A partial
manifestation, which may refer to simple facts, concerns a basic at tribute (a feature) of the description of the
settlement: a name, a type, a geographical location or a (mereological) relation to other places. The information
on simple facts about the attributes of a set tlement will be called simple partial manifestations. Each of the par-
tial manifestations has a time dimension expressed by the period to which the given source information refers.
Aggregated manifestations are built on the basis of par tial manifestations, and may refer to complex facts and
interpretations. Each partial manifestation, both simple and interpolated, be it either for the name, type, location,
or mereological relations, results in the creation of an aggregated manifestation, which includes all the attributes
of the settlement.
The goal of this research is to meaningfully present incomplete and uncertain data on settlement unit s. The
presentation of that data should be possible on a map of GIS type, taking into account such things as responsive-
ness during interaction with a user and rendering logic. The map should be refreshed immediately after changing
a date, and it would lead to confusion if the map did not show anything if the user selected a date inbetween the
manifestations based on facts (and not an interpolated manifestation).
The main contributions of this article are as follows:
1. A database model for representing partial information on historical settlement units;
2. Database algorithms for generating the interpolated and aggregated information manifestations; and
3. Ultimately, based on par tial facts, the construction of a stor yline for settlements with demonstration on which
storyline elements are based on simple facts and which are the result of interpretation and interpolation.
We discuss the problems presented and solutions via several examples from the real data we have collected
within the ONTOHGIS project.
FIGURE 1 The mention of a settlement name, “Czermino”, in a historical source
|
357
MYRDA et A l.
2 | PROBLEM STATEMENT
2.1 | Motivating scenarios
The following deliberations pert ain to two cases of settlements, for which the name, location, and type have been
subject to change.
The first concerns the change of location and the type of a settlement, located by the Vistula River, called
Ryczy wół. Ryczy wół, or Ricziwol in its original notation, first appears in a document of Władysław II Jagiełło dat-
ing back to January 7, 1388, where it already had been granted town status (“oppidanis eiusdem oppidi Ricziwol
damus”) (Sułkowska-Kuraś & Kuraś, 1974, p. 35). Near the town, operating as a separate entity, there was a castle,
as demonstrated by another document issued by Queen Jadwiga on January 26, 1393 (Kuraś & Sułkowska-Kuraś,
1969, p. 267). It was possible to establish the location of the settlement in the Middle Ages owing to an archeolog-
ical sur vey, and cartographic sources from the second half of the eighteenth century (Zieleniewska-Kasprzycka,
2015). Such a procedure is illustrated by, for example, the Historical Atlas of Poland in the second half of the six-
teenth century (ca. year 1600) (Słon, 2014, p. 1378).
In 1813, due to the V istula River flooding, the authorities decided to relocate Ryczywół 2 kilometers to the
south-west of its original location. The process was gradual and took some time: by February 8, 1817, 46 new
dwellings and 41 barns had already been built at the new location, 7 houses and 8 barns were in progress, and
73 houses and 9 barns were still to be moved (Szafer, 1955, p. 80). The change is well illustrated by car tographic
sources: on Mayer von Heldensfeld's map of 1801–1804, Ryczywół can be found at its former location, while
on the Quatermaster's map of 1822–1843, it is already at the new one. In the partition period, following the
Prussian authorities’ decision, Ryczywół became a town and as such was entered into the Geographical Dictionary
of the Kingdom of Poland of 1889 (Sulimierski, Chlebowski, & Walewski, 1889, p. 81) and into the first census
of the Second Republic of Poland of 1921 (Anon, 1925, p. 60). Currently, according to the National Register of
Geographic Names (PRNG) of July 18, 2016, Ryczywół is a village (ID:119135).
The ma p pr esented in Fi gu re 2 depi cts Ryc z y wó ł's loc at ions, whi le Tabl e 1 il lu str at es the ch an ge s in the se ttle-
ment's name and type as confirmed by the primary source attestations. During the process of input ting the data
into the GIS system, some simplifications were made.
1. The name of the settlement and its type were adjusted to the Polish notation (the names of the set-
tlement s given in the historical sources and English translations of their types are given in parentheses).
2. It was assumed that the date of the new location was February 8, 1817.
3. For the Middle Ages (January 7, 1388 and January 26, 1393) the location was assumed according to nineteenth-
century maps verified by the archaeological survey's findings.
4. A bsent the specific day in a date, the beginning of a period was assumed as January 1 and the end as December 31.
The scenario of the second case was even more complex and involves changes in the name, location, type
and mereological relationships of the current district of Toruń called Podgórz (the changes are summarized in
Table 2). The very beginning of the history of Podgórz causes difficulties in interpretation. The settlement of
Podgórz was created as a result of moving the inhabitants of a trading settlement called Dybów or Stara Nieszawa
to a new location (about half a kilometer away). A royal document of May 16, 1555, mentions: “incolae Antique
Nieschoviensis seu Diboviensis, quae nunc nova nomine Podgorze appellatur” (“inhabit ants of Stara Nieszawa or
Dybów, which is now called by the new name Podgórze”), and further, “in Antiqua Nieschovia seu Dibovia, quam
nunc de loco inferiori in montem translatum [!] Podgorze appellant” (in Stara Nieszawa or Dybów, which now is
transferred from the valley to the hill called Podgórze) (Janosz-Biskupowa, 1954, pp. 191–192). Although both the
location and the name of the locality have changed, Stara Nieszawa-Dybów and Podgórze should be treated as
the same place (identity embedded in historical source; Szady & Ławrynowicz, 2017). This is also illustrated by the
Historical Atlas of Poland of ca. 1600 (Institute of History of the Polish Academy of Sciences, 2019).
358
|
MYRDA et A l.
In the following centuries, the legal status of the settlement was subject to frequent changes. On November
7, 1611 the town was granted privileges which it lost on March 26, 1833 due to legal changes concerning town
and village communes (gminy, the lowest level of administrative division) in Prussia. In 1845, Podgórz was granted
FIGURE 2 Ryczywół in the Middle Ages (I) and af ter the change of location (III) (Szafer, 1955, p. 63). A finer-
grained example of changes undergone by Ryczy wół is included in Table 1
TABLE 1 Information on the name, settlement type, and location of Ryczy wół
Date Name Typ e Location
Januar y 7, 1388 Ryczywół [Ricziwol] miasto [oppidum, town] X:21.440928666118467,
Y:51.694253 34776987
Januar y 26, 1393 R yczy wół [Ryczy wol] miasto [oppidum, town]
1600 Ryczy wół miasto [town] X:21.440928666118467,
Y:51.694253 34776987
1801–1804 Ryczywól [Ryczy wół,
Ryczewul]
X:21.440928666118467,
Y:51.694253 34776987
8 February, 1817 Ryczy wól miasto [town] X:21.42249999998687,
Y:51.689999999757134
1822–1843 Ryczywół [Ryczy wol] X:21.42249999998687,
Y:51.689999999757134
1889 Rycz ywół osada miejska [municipal settlement]
1921 Ryczywół osada miejska [municipal settlement
July 18, 2016 Ryczy wół wieś [village] X:21.42249999998687,
Y:51.689999999757134
|
359
MYRDA et A l.
the status of a market settlement (Marktflecken) and regained its former full town privileges on Januar y 1, 1925.
However, on April 1, Podgórz ceased to be a separate settlement and was incorporated into Toruń, becoming a
district (Ciesielska & Zakrzewski, 2005), which it has remained until today (PRNG of July 18, 2016; ID: 1044 84).
Similarly to the case of Ryczywół, while inputting the data into the GIS some simplifications concerning the
name, the location before that illustrated on the cartographic sources, as well as the dates were made.
We will use these two examples of Ryczywół and Podgórz in what follows to motivate and illustrate our pro-
posed solutions.
2.2 | Research questions
We have gathered a set of research questions which we seek to answer by our research presented in this ar ticle.
RQ1 How to consistently handle information confirmed in historical sources, which is only par tial (i.e.,
it regards only a subset of settlements’ features)?
RQ2 How to distinguish generated information from source information?
RQ3 How to ensure continuit y (which is required to show localities on the map)?
RQ4 Is there only one correct method of filling holes?
3 | RELATED WORK
Traditional approaches to modeling spatio-temporal changes are usually based on a relational database model
(Stuart , 1992). They usually either keep an updated version of the state or take snapshots at certain points in time
(Abraham & Roddick, 1999). The latter model (snapshot model) has been the most common approach to the inte-
gration of space and time in databases. Although snapshot databases are relatively easy to implement and query
regarding the recorded states, repeatedly recording the whole data set in some time intervals leads to redundancy.
Additionally, the snapshot model only considers change implicitly, and its explicit detection requires comparing
many snapshots.
TABLE 2 Information on the name, settlement type, and location of Podgórz
Date Name Typ e Location
June 2, 1543 Dybów [Antiqua
Nyeschewa, Dibovia]
X:18.5932311869783,
Y:52 .99 881467 22835
May 16, 1555 Podgórz [Podgorze] X:18.5916356118256,
Y:52.9921219054298
1600 Podgórz X:18.5916356118256,
Y:52.9921219054298
November 7, 1611 Podgórz miasto [town]
March 26, 1833 Podgórz wies [village]
1845 Podgórz osada t argowa [Markt flecken,
trading settlement]
Januar y 1, 1925 Podgórz miasto [town]
April 1, 1938 Podgórz czę ść miasta [par t of town]
July 18, 2016 Podgórz część mias ta [part of town] X:18.5916356118256,
Y:52.9921219054298
360
|
MYRDA et A l.
Object-oriented approaches are based on object-oriented databases (Abraham & Roddick, 1999; Atkinson
et al., 1992). The object-centric approach takes into account both natural phenomena and ar tifacts resulting from
human work as distinguished object s, which are charac terized by thematic, topological and geometrical attributes.
When the object changes, a new version of it is created. There are two main ways of versioning: assigning time-
stamps to each attribute, and creating a new version of the whole object with a new identifier. The identifier of an
object indicates continuity of the object during changes the object undergoes.
The event-oriented approach, in contrast to the other t wo approaches, stores changes explicitly. In its sim-
plest form, the event-based model consists of a base map, which shows an initial state. The map is subsequently
changed and the subsequent changes are recorded in the transaction log (Pelekis, Theodoulidis, Kopanakis, &
Theodoridis, 20 04). Since event data are stored explicitly, it facilitates formulating queries on the changes, there-
fore making the event-based approach better suited to modeling processes. In this way, the event-based approach
seems the most appropriate to our case.
Other approaches that also offer a solution for handling spatio-temporal changes include qualitative spatial repre-
sentation and the use of graph theory. Qualitative spatial representation (Chen, Cohn, Liu, & Wang, 2015) has broad
applications in geographic information systems. There have also been attempts in practical solutions in the field at
formalization and reasoning based on qualitative spatial data. For instance, a thematic dictionary for formalization (in
an ontolog y language, OWL) was presented in a narrative description of spatial information regarding ethnographic
data (Chang, Chang, Chuang, Deng, & Huang, 20 09). Practical and theoretical problems regarding inference based on
qualitative data on transformations of urban space were presented in Bhatt and Wallgruen (2014). However, in con-
trast to the numerous qualit ative spatial calculi proposed, publicly available implementations employing inference with
qualitative spatial representations are still both scarce and relatively limited. Considering the fact that most of our col-
lected s patial information is based on c artographic re presentation, qualitative (text ual, narrative) met hods of expressing
geographic location and relations (e.g., that “some settlement is north (or south) of some geographical feature”) were
omitted at this stage because our source spatial data are not qualitative. The aim of this ar ticle is to present methods for
processing massive data, where the location of a settlement is expressed in a classical way by geographic coordinates.
Graph database s offer many possibilities, but still have limitations in the area of spatial data handling. For example,
according to Agoub, Kunde, and Kada (2016), graph database management systems such as Neo4j and ArangoDB do
not have full support for many reference systems or advanced spatial functionalities. Compared to relational data-
bases, their capabilities are still much smaller. Additionally, our initial requirement was that projec t results will be used
in a produc tion system, so greater maturity and versatility at the database level were required which, according to
Baralis , Dalla Valle, Garza, and Scullino (2017), NoSQL databases in general do not yet provide. Because there is no da-
tab ase solving all the problems, and in particular be cause, according to Khine and Wang (2019), graph databa ses work
well only when it is necessary to manage very complex relationships, we decided to choose an approach referred to as
“polyglot persistence,” that is, to use multiple data storage technologies to handle different application needs, to have
the possibility of using a multi-model data store architecture in the future. Assuming that we are currently developing
an element of a future la rger system, we have focused on th e possibility of storing data in a manner t ailored to a partic-
ular need, which is visualization of data in the form of a map, using methods making it possible to use GIS visualization
software. The software programs we use are standard GIS component s (including Geoserver, QGIS), hence the as-
sumption that the dat abase for data visualization in the form of a map will be a stable and mature relational database.
Our goal is also to have GIS data in a standard form, which could be used as an element of the national spatial data
infrastructure, that is, data that others c an easily use in standard GIS applications and systems, outside of our project.
Some other related areas and works include historical gazetteers and spatio-temporal databases support-
ing digital history, where the role of ontologies was discussed in developing historical gazetteers (Janowicz &
Keßler, 2008). In Gantner, Waldvogel, Meile, and Laube (2013), a spatio-temporal ontolog y was proposed (named
SONADUS) for modeling the evolution of the administrative units of Switzerland. Another notable geo-spatial
ontology of administrative units (QVIZ), subsequently developed under the name Administrative Unit Ontology
(AUO), was proposed in Southall (2012).
|
361
MYRDA et A l.
No matter what approach to the modeling of spatio-temporal changes has been adopted, the previous ap-
proaches have not considered presenting the provenance of a critical historical analysis on a map, but only the re-
sults of such analysis. In particular, previous approaches have not considered the distinction bet ween the source,
the critical analysis, and the interpretation. In this article, we also consider a step before the critically analyzed
information is captured in the map, and our proposed model also embraces capturing in the database and present-
ing on the map the record of the critical historical analysis.
4 | PRELIMINARIES
In this section we first introduce our proposed approach to historical analysis, supported by the computational
methods we propose. In particular, we discuss the practice of the process of historical research and how our
proposed methods suppor t automation of this process. Subsequently, we formalize the notion of manifestations.
4.1 | Classical approach versus our proposed approach to historical analysis
Our proposed methods provide computer support for historians in their analysis. Fur thermore, the methods sup-
port more finer-grained analysis with provenance tracking.
In general, until now, if a historian wanted to obtain an information about the continuit y of a settlement unit,
s/he had to analyze it by her/himself, that is, if this settlement was mentioned in some source and later in another
one, s/he had to connect and fill a gap in its timeline manually (see Figure 3). Our proposed methods support
semi-automation of this analysis.
In particular, each settlement consists of a series of successive manifestations in our representation. Each
manifestation concerns a single feature such as name, type, or location (partial manifestation) or a set of these
features (aggregated manifestation) during a defined period of time. Unfor tunately, similarly to the classical anal-
ysis setting, not all of the manifestations can be filled on the basis of historic al sources, because of time gaps in
historical sources. Thus also the represent ation of reality by means of manifestations produces “gaps” in the repre-
sentation of the lifetime of a settlement (lack of continuity). However, our proposed methods enable interpolation
of the available information automatically to fill in those gaps. In this way, we obtain interpolated manifestations,
that is, manifestations based on automatic inference (see Figure 4). Furthermore, our proposed methods also pro-
duce aggregated manifestations. For historical periods, which are devoid of a direct source for the name, location,
type, and mereological relations of the settlement, the concept of par tial interpolated manifestations will be used.
FIGURE 3 A classical process of manual analysis and interpretation of historio-geographical facts
362
|
MYRDA et A l.
4.2 | Variable entities and manifestations
The database, which we describe in Sec tion 5, is designed based on a philosophical concept of variable and non-
variable entities (Fine, 1999; Moltmann, 2020, in print; Trypuz, Kuzinski, & Sopek, 2016). Each entity belongs to
either of these groups. Variable entities are those which have their different manifestations at different times.
Manifestations of variable entities are called non-variable entities.
The basic idea regarding manifestations is that all entities are divided into variable entities (i.e., entities that
have different manifestations at different times) and non-variable entities (which do not have different manifesta-
tions as different objects at different times).
Variable objects are subject to the following conditions (based on Fine, 1999; Moltmann, 2020, in print; Trypuz
et al., 2016).
•Existence. A variable entity e exists at time t if e has a manifestation m at t and all properties of m at t.
• Location. A variable entity e takes a geographical location of its manifestation m at time t if its manifestation at
t has a location.
•Property inheritance 1. If a variable entity e exists at time t, then e has a property P (time-dependent) at t if e's
manifestation at t has P.
• Property inheritance 2. A variable entity e has a property P (independent of circumstances) if all of e's manifes-
tations have P (time-independent) at all the times at which they exist (in practice, the property P can be directly
attributed to the variable entity).
Each manifestation m has a n associated time i nterval
m(tstart,tend )
. Each var iable entity e has associated k it s man-
ifestations
m
e
1,
…
,m
e
k,k
≥
0
. Each manifestation m has n properties from the set of properties
Q
=
(q1,q2,
…
,qn)
modeled in the domain of interest, and it has n ≥ 1 values of the properties set up, while the values of the remain-
ing properties from the set are unknown.
Such representation means that the time inter val of the variable entity may not be continuous, as already men-
tioned. There may be gaps caused by the lack of existence of a manifestation during some time period. In this work,
we aim to fill these gaps by generating “missing” manifestations. Therefore, we define real manifestations and
FIGURE 4 A semi-automatic process of analysis and interpretation of historio-geographical fact s
|
363
MYRDA et A l.
interpolated (virtual) manifestations. Real manifestations are those which are confirmed based on historical sources
and recorded in the database by historians. Interpolated (virtual) manifestations are generated from the real manifes-
tations by our algorithms described in Section 5.2. Moreover, historical sources may contain only partial character-
istics of a settlement; for example, there may be a mention in the source only of a name or location or another single
characteristic. Indeed, one may easily imagine that situations where there will be a complete description in a given
mention in a source regarding a particular period are quite rare. Therefore, we also introduce a categorization of
manifestations into partial and aggregated manifestations. Partial manifestations are those which describe a selected
proper ty of a manifestation only. Aggregated manifes tations describe all the properties of a manifestat ion in the mod-
eled domain. In other words, any partial manifestation m has exactly one propert y
pi
for which it s associated value
vi
is known, while any aggregated manifestation has all values
v1,v2,
…
,vn
of all of its properties
q1,q2,
…
,qn
known.
5 | CREATION OF A DATABASE
5.1 | Database schema
The central entity of our schema is VariableSettle me nts, whose aim is to represent the history of a given
settlement unit , regarding changes of its properties. In this work we consider four major properties of settlement
units: name, type, location, and mereology following Garbacz, Lawrynowicz, and Szady (2018). This is reflected
in Figure 5, which presents the conceptual Entity Relationship Diagram (ERD), a schema of the database, which
includes the following entities: SettlementNames, S ettle me ntTy pes, and SettlementLocations.
These properties, together with their values, constitute partial manifestations. This is because the mentions of
places included in historical sources usually relate to one of the properties to which we can attribute the start and
end dates. All partial manifestations of a given settlement are grouped into one variable settlement. The basic and
only datum that uniquely identifies each settlement is its identifier and only this identifier (not the name, nor the co-
ordinates). The list of all settlements is a list of identifiers in VariableSettlements in the database schema. The
settlement can thus be associated with such attributes as the manifestation of the name, location, type, origin, and
an affiliation to another settlement. These attributes, referred to as partial manifestations, represent a given locality
at a given time determined by two dates: Start and End. The set of partial manifestations for a given settlement
identified by an identifier in VariableSettlements is a representation of the history of that settlement (its indi-
vidual attributes) over time.
In order to facilitate immediate access to the set of information usually presented on a map, particular partial
manifestations for a given settlement are combined into one, which presents all the attributes such as name,
type, location at the same time, thus enabling effective presentation of this information on the map. However,
combining partial manifestations into aggregated manifestations leads to a side effect in the form of a lack of time
synchronization between different manifestations resulting from the fac t that one proper ty may have a different
time period associated than another property of the same settlement, which results in some kind of “time hole”
in describing the settlement with the use of properties. In the real world, the situation where there is a period
of uncertainty between two periods of the confirmed occurrence of a given property is usually interpreted such
that a settlement had a given property with a value of the previous period for the whole period of the “time hole”
until the beginning of the second period. Therefore we adopt similar treatment of such a situation in the database.
An automatic replenishment of the time gaps, that is, an interpolated chronology as par t of unit manifesta-
tions, was made directly in the database in the form of SQL queries providing the appropriate views, including
partial manifestations along with virtual par tial manifestations (we describe the algorithms in Section 5.2).
Virtual manifestations in par tial views are distinguished by the fact that they have zeroed identifiers, in oppo-
site to real manifestations.
Figure 6 presents database records with the information related to the Podgórz settlement, which we have
discussed as one of our motivating examples.
364
|
MYRDA et A l.
5.2 | Algorithms for generating aggregated settlement units manifestations
The general algorithm works in such a way that for each settlement unit all of its manifestations are first included
in the aggregated manifestation. Then the time continuity is checked. If time gaps are detected, virtual manifesta-
tions are generated.
To increase flexibility, we propose several variants of the algorithm to generate virtual manifestations. In ad-
dition to the one often used in the real world (and mentioned in the previous section), it is also possible to fill the
time gap, dividing it in half and filling the halves with two dif ferent values, or filling it with two different values, in
parallel in time. If one fills in a half, each vir tual manifest ation consists of a sequence of two sub-periods. The first
half of the virtual manifestation is assigned the value of the preceding manifestation, while the second half of the
virtual manifestation is assigned the value of the real manifestation following the generated virtual manifestation.
If one fills in parallel, each virtual manifestation consists of two parallel sub-periods. The first virtual manifestation
is assigned the value of the preceding manifestation, while the second virtual manifestation is assigned the value
of the real manifestation following the generated virtual manifestation. Ar tificial periods form new manifestations,
created based on information about existing manifestations. It is worth noting that time interpolation is performed
FIGURE 5 ERD conceptual schema
|
365
MYRDA et A l.
between existing manifestations in the database. So it is not done for the period before the first existing manifes-
tation, nor after the last existing manifestation.
The following rules are applied:
1. Virtual manifestations do not have their own unique identifiers, unlike real manifestations.
2. Each ending year of an existing manifest ation, incremented by one year, is the potential beginning of
the virtual period and can be a start date of new manifestation for given settlement (represented by
Varia bleSettlem entIdent ifier).
3. The earliest beginning date of the following real manifestation is a potential ending of the virtual period and
can be the end date of a new manifestation for a given settlement. Each time gap can be filled with one or two
manifestations as in Algorithm 1.
Assuming that we have a set (a table) of non-continuous manifestations of settlement names in time and we
want to have continuous manifestations of names in time in the output (a table view), we can use the following data
processing and automated reasoning. It is a high-level description of an algorithm generating continuous settlement
manifestations consisting of real manifestations and interpolated virtual manifestations in place of time gaps. At the
low level it is implemented as an SQL query.
FIGURE 6 A settlement unit (Podgórz), which throughout history has had two different names, three
different types, two different locations, and has undergone changes in mereology, but still has the identifier
83131, indicating it has retained its identity
Identifiers
83131
(a) Variable Settlement
Identifiers Names VariableSettlementIdentifiersStartsAt EndsAt
155657 Stara Nieszawa 83131 1460-09-25 1554-12-31
83131 Podgórz 83131 1555-01-01 2016-12-31
(b) Manifestation of Name
Identifiers SettlementTypeIdentifiers VariableSettlementIdentifiersStartsAt EndsAt
155657 2 83131 1460-09-25 1611-11-06
155656 3 83131 1611-11-07 1833-03-26
155655 2 83131 1833-03-27 1924-12-31
155654 3 83131 1925-01-01 1938-03-31
83131 61 83131 2016-01-01 2016-12-31
(c) Manifestation of Type
(d) Manifestation of Location
(e) Manifestation of Mereology
(f) Aggregated Manifestation
Identifiersthe_geomVariableSettlementIdentifiersStartsAt EndsAt
155660 POINT(18.5932311869783
52.9988146722835)
83131 1460-09-25 1554-12-31
83131 POINT(18.5916356118256
52.9921219054298)
83131 1555-01-01 2016-12-31
Identifiers PartIdentifiers WholeIdentifiers StartsAt EndsAt
7 83131 112602 1938-01-01 2016-12-31
Identifiers Names VariableSettlement-
Identifiers
Types Mereology Geometries Name-
Identifiers
Type-
Identifiers
Location-
Identifiers
Mereology-
Identifiers
StartsAt EndsAt
83363 Stara Nieszawa 83131 2 Null POINT-
(18.5932311869783
52.9988146722835)
155657 155657 155660 Null 1460-09-25 1554-12-31
83364 Podgórz 83131 2 Null POINT-
(18.5916356118256
52.9921219054298)
83131155657 83131Null1555-01-01 1611-11-06
83365 Podgórz 83131 3 Null POINT-
(18.5916356118256
52.9921219054298)
83131155656 83131Null1611-11-071833-03-26
83366 Podgórz 83131 2 Null POINT-
(18.5916356118256
52.9921219054298)
83131155655 83131Null1833-03-27 1924-12-31
83367 Podgórz 83131 3 Null POINT-
(18.5916356118256
52.9921219054298)
83131155654 83131Null1925-01-01 1937-12-31
83368 Podgórz 83131 3112602 POINT-
(18.5916356118256
52.9921219054298)
83131155654 831317 1938-01-01 1938-03-31
83369 Podgórz 83131 3112602 POINT-
(18.5916356118256
52.9921219054298)
831310831317 1938-04-01 2015-12-31
83370 Podgórz 83131 61 112602 POINT-
(18.5916356118256
52.9921219054298)
8313183131831317 2016-01-01 2016-12-31
366
|
MYRDA et A l.
The manifestations of types and locations are generated analogously to manifest ations of names.
Results can be visualized on a map as we can see in Figures 7 and 8, which are a continuation of our motivating
example of the changes to Podgórz. Podgórz has settlement type given for the years 1833–1924, and no type
Algorithm 1: Generating continuous settlement manifestations of names consisting of real manifest ations
and interpolated virtual manifestations in place of time gaps
|
367
MYRDA et A l.
entered manually for year 1940. Vir tual manifestations (which have zero values in the aggregated manifestation
table in Figure 6) are presented on a map in a different visual style.
5.3 | Implementation
We have applied our database model and algorithms to implement and deploy a database of settlement unit s for
the real-world case of representing historical settlement units in the Polish lands until 1939 ( http://ontoh gis.pl).
The data were collected from historians via database inter faces. The proposed model also proved flexible enough
to import data from other, external sources: the PRNG ( http://codgik.gov.pl) and the Historical Atlas of Poland.
FIGURE 7 Podgórz on the map: Visualization of the real manifestation in the year 1845
FIGURE 8 Podgórz on the map: Visualization of the vir tual manifestation in the year 1940
368
|
MYRDA et A l.
Every entr y from the PRNG, depending on whether it is the main or an additional location of a given localit y,
created either a meta-object in Varia bleSettlements or a manifestation (in the PRNG, most of the localities
have several locations).
Each type name specified in the PRNG, after verifying whether it exists in our case, has been changed
into a type identifier and the type identifier was inserted. If the type did not exist, it was added to our table
Settle me ntTy pes (it turned out that there was only one such case: part of the hamlet).
In the PRNG , a locality can have a specific parent object (also a locality). For such a relationship, it was also
reproduced in our database. This turned out to be non-trivial, due to lack of numerical identifiers in the PRNG
(parent objects in the PRNG only have textual identifiers—names of settlements—and there are multiple set-
tlement s with the same name), and errors (such as t ypos in the names of settlements or differences in naming
on various source maps). Those relations are recorded in ManifestationMereologicalLinks. Initially, there
were 155,622 entries of the settlement type. In total, this resulted in the addition of 124,729 settlements (me-
ta-objects). Each manifestation was initially dated for the whole of 2016 (beginning January 1, ending December
31), and then a decision was made that the manifestations would be more event-centric and now they have the
date July 18, 2016, the date of publication of these data by GUGIK (Head Office of Geodesy and Cartography).
The Historical Atlas of Poland, maintained by the Polish Academy of Sciences (Institute of History) provided
data on set tlement units for the per iod 1550–1600. No new identities were inser ted (in Varia bleSett lem ent s),
but only new partial manifestations were created.
Table 3 shows summary statistics for the records currently available in the database.
6 | CONCLUSIONS AND FUTURE WORKS
In this article we have presented a solution aimed at modeling and storing incomplete and uncertain information
regarding the properties (name, type, etc.) of settlement units in time. This involves using a concept of variable
entities and manifestations to solve the problem, a database schema and algorithms to fill in the “time holes” in the
history of a settlement unit, to be able to present it on a map (which requires continuity).
We have presented both the problems and solutions using real cases from a project aimed at representing the
history of settlement units on Polish lands. We have also reported on the deployment of the proposed solutions
in the historical GIS maintained by the Polish Academy of Sciences.
The proposed approach to the state of settlements for periods devoid of information or source premises cov-
ers only the simple c ase scenarios. However, not only diachronic approaches to database changes, but also double
location (e.g., Ryc zywół) or more than one name (e.g., Podgórz) in a synchronic approach still remain challenges
for the future. For instance, in the case of Ryczywół, it actually operated in two places at the same time for a short
period (1813–1817). The second problematic issue is retrogressive modeling of geographic location, as the first
TABLE 3 Summary statistics on the records in the database on historical settlement units, developed in the
ONTOHGIS project
VariableSettlements 124,72 9
SettlementLocations 173, 218
SettlementMereologyLinks 77,61 2
Settlement Names 142, 332
SettlementTypes 142,33 0
SettlementTypesDictionary 140
SettlementManifestations 159, 925
|
369
MYRDA et A l.
cartometric maps illustrating the locations of the settlements date back to the end of the eighteenth century. For
now, the archeological verification of the location of each settlement for the preceding period is still impossible.
Another problem is the presumed existence of a partial manifest ation and incorporating or extracting one set-
tlement from another (mereological change). Finally, the fourth challenge still to be tackled is the way to register
and input contradictory or diverging information for one period on the basis of two different sources or different
interpretations of one vague source.
ACKNOWLEDGMENTS
The research was supported by the Ontological Foundations for Building Historical Geoinformation Systems (2bH
15 0216 83) grant funded by National Programme for the Development of Humanities.
ORCID
Grzegorz Myrda https://orcid.org/0000-0002-2756-8654
Bogumił Szady https://orcid.org/0000-0003-0059-5596
Agnieszka Ławrynowicz https://orcid.org/0000-0002-2442-345X
REFERENCES
Abraham, T., & Roddick, J. F. (1999). Survey of spatio-temporal databases. Geoinformatica, 3(1) , 61–9 9.
Agoub, A., Kunde, F., & Kada, M. (2016). Potential of graph databases in representing and enriching standardized geodata.
Dreiländertagung of DGPF, OVG and SGPF, Bern, 25, 208–216.
Anon. (1925). Skorowidz miejscowosci Rzeczypospolitej Polskiej opracowany na podstawie wyników Pierwszego Powszechnego
Spisu Ludności z dn. 30 września 1921 r. i innych źródeł urzędowych. T. 3 : Województ wo kiele ckie [Index of places of the
Republic of Poland based on results of th e First Census of Septem ber 30, 1921 and other off icial source s] (Vo l. 3). Warsaw,
Poland: Kielce Voivodeship. War saw.
Atkinson, M., DeWitt, D., Maier, D., Bancilhon, F., Dittrich, K., & Zdonik, S. (1992). The object-oriented database system
manifes to. In F. Bancilhon, C. Delobel, & P. Kanellakis (Eds.), Building an object-oriented database system (pp. 1–20). San
Francisco, CA: Morgan Kaufmann.
Baralis, E., Dalla Valle, A., Garza, P., Rossi, C., & Scullino, F. (2017). SQL versus NoSQL databases for geospatial applica-
tions. In Proceedings of the 2017 IEEE International Conference on Big Data, Boston, MA (pp. 3388–3397). Piscataway,
NJ: IEEE.
Bhatt, M., & Wallgruen, J. O. (2014). Geospatial narratives and their spatiotemporal dynamics: Commonsense reason-
ing for high-level analyses in geographic infor mation s ystems. ISPRS International Journal of Geo-Information, 3(1),
166–205.
Chang, C.-L., Chang , Y.-H., Chuang, T.-R., Deng, D.-P., & Huang, A . W.-C. (2009). Narrative geospatial knowledge in eth-
nographies: Representation and reasoning. In K. Janowicz, M. Raubal, & S. Levashkin (Eds.), GeoSpatial semantics:
GeoS 2009 (Lecture Notes in Computer Science, Vol. 5892, pp. 188–203). Berlin, Germany: Springer.
Chen, J., Cohn, A . G ., Liu, D., & Wang, S. (2015). A sur vey of qualitative spatial representations. Knowledge Engineering
Review, 30(1 ) , 10 6–136.
Ciesielska, K., & Zakrzewski, T. (2005). 450 lat toruńskiego Podgór za 1555–2005 [450 years of Toruń Podgór ze 1555–2005].
Toruń, Poland: Towarzystwo Miłosników Torunia.
Fine, K. (1999). Things and their parts. Midwest Studies in Philosophy, 23(1 ), 6 1–74 .
Gantner, F., Waldvogel, B., Meile, R., & Laube, P. (2013). The basic formal ontology as a reference framework for modeling
the evolution of administrative units. Transactions in GIS, 17(2), 206–226.
Garbacz, P., Lawr ynowicz, A., & Szady, B. (2018), Identity criteria for localities. In S. Borgo, P. Hitzler, & O. Kutz (Eds.),
Formal ontology in information systems: Proceedings of the 10th International Conference, FOIS 2018, Cape Town, South
Africa (pp. 47–54). Amsterdam, Netherlands: IOS Press.
Garbacz, P., & Trypuz, R. (2017). Representation of tensed relations in OWL: A survey of philosophically-motivated pat-
terns. In E. Garoufallou, S. Virkus , R. Siatri, & D. Koutsomiha (Eds.), Metadata and semantic research: 11th International
Conference, MTSR 2017, Tallinn, Estonia, November 28-December 1, 2017, Proceedings (pp. 62–73). Cham, Switzerland:
Springer.
Institute of History of the Polish Academy of Sciences. (2019). Ziemie polskie Korony w XVI w. Przestrzenna baza danych.
Retrieved from http://atlas fonti um.pl
370
|
MYRDA et A l.
Janosz-Biskupowa , I. (1954). O położeniu i przeniesieniu Nieszawy (Z dziejów handlu wiślanego w XV wieku). Zapiski
Towarzystwa Naukowego w Toruniu, 20, 16 7–195 .
Janowicz, K.,& Keßler, C. (20 08). The role of ontology in improving gazetteer interaction. International Journal of
Geographical Information Science, 22(10), 1129–1157.
Khine, P., & Wang, Z. (2019). A review of polyglot persistence in the Big Data world. Information, 10(4), 141.
Kuraś, S., & Sułkowska-Kuraś, I. (Eds.) (1969). Zbiór dokumentów małopolskich (Vol . 4: Dokument y z lat 1211–1400).
Wrocław, Poland: Zakład Narodowy im. Ossolińskich - Wydawnic two Polskiej Akademii Nauk.
Moltmann, F. (2020, in print). Variable objects and truthmaking. In M. Dumitru (Ed.), Metaphysics, meaning, and modality:
Themes from Kit Fine. Oxford, UK : Oxford University Press.
Pelekis, N., Theodoulidis, B., Kopanak is, I., & Theodoridis, Y. (2004). Literature review of spatio-temporal database mod-
els. Knowledge Engineering Review, 19(3), 235–274.
Pomorski, J. (1983–1984). Współczesne spory wokół teorii faktu historycznego. Annales Universitatis Mariae Curie-
Skłodowska: Sectio F, Historia, 38–39, 27–5 4.
Southall, H. (2012). Rebuilding the Great Britain Historic al GIS: Part 2, A geospatial ontology of administrative units.
Historical Methods, 45(3), 119–134.
Stuar t, N. (1992). Time in geogr aphic inform ation systems b y G. Langran ( book review). International Journal of Geographical
Information Systems, 6(6), 547.
Sulimierski, F., Chlebowski, B., & Walewski, W. (Eds.). (1889). Słownik geograficzny Królestwa Polskiego i innych krajów słowi-
ańskich (Vo l. 10). Wars aw, Poland: University of Warsaw Library.
Sułkowska-Kuraś, I.,& Kuraś, S. (Eds.) (1974). Zbiór dokumentów małopolskich, Vol. 6: Dokumenty króla Władysława Jagiełły
z lat 1386–1417. Wrocław, Poland: Zakład Narodowy im. Ossolińskich - Wydawnictwo Polskiej Akademii Nauk.
Szady, B., & Ł awrynowicz, A. (2017). Considering identification of locality in time: Theoretical and practical approach.
In P. Fogliaroni, A. Ballatore, & E. Clementini (Eds.), Proceedings of Workshops and Pos ters at the 13th International
Conference on Spatial Information Theory (COSIT 2017) (Lecture Notes in Geoinformation and Cartography, pp. 283–
291). Cham, Switzerland: Springer.
Szafer, T. P. (1955). Ze studiów nad planowaniem miast w Polsce w XVIII i pocz. XIX . In W. Ostrowski (Ed.), Studia z historii
budowy miast (pp. 48–82). Wars aw, Poland: Prace Instytutu Urbanist yki i Architektury.
Słon, M. (Ed.) (2014). Historical atlas of Poland in the 2nd half of the 16th Centur y. Bern, Swit zerland: Peter L ang.
Topolski, J. (2012). Methodology of history. Dordrecht, Netherlands: Springer.
Trypuz, R., Kuzinski, D., & Sopek , M. (2016). General legal entit y identifier ontology. In Proceedings of the Joint Ontology
Workshops 2016—Episode 2: The French Summer of Ontolog y Co-Located with the 9th International Conference on Formal
Ontology in Information Systems, Annecy, France.
Zieleniewska-Kasprzycka, M. (2015). Ryczywół zaginiony—badania nieinwazyjne dawnego kompleksu osadniczego Ryczywół
i jego okolic. Retrieved from http://www.staro zytni cy.org.pl/img/ryczy wol_zagin iony.pdf
How to cite this article: Myrda G, Szady B, Ławrynowicz A. Modeling and presenting incomplete and
uncertain data on historical set tlement units. Transactions in GIS. 2020;24:355–370. https ://doi.or g/10.1111/
tgis.12609
