This paper presents an application for studying the death records of WW2 casualties from a prosopograhical perspective, provided by the various local military cemeteries where the dead were buried. The idea is to provide the end user with a global visual map view on the places in which the casualties were buried as well as with a local historical perspective on what happened to the casualties that lay within a particular cemetery of a village or town. Plenty of data exists about the Second World War (WW2), but the data is typically archived in unconnected, isolated silos in different organizations. This makes it difficult to track down, visualize, and study information that is contained within multiple distinct datasets. In our work, this problem is solved using aggregated Linked Open Data provided by the WarSampo Data Service and SPARQL endpoint.
This paper builds upon the WarSampo project4 that collects data related to
the Finnish WW2, and publishes this data as Linked Open Data [
Seven different user-friendly application perspectives based on the the
WarSampo Data Service6 have been created within the semantic WarSampo
Portal7 [
Finnish soldiers who perished in WW2 were transported back to their
hometown for burial whenever it was possible [
This paper presents new interactive views to the casualty data based on
cemeteries and other community level groups. First, all military cemeteries in
Finland can be browsed with a faceted browser, and visualized on maps to
provide a global view of the cemeteries. Second, another new view is the cemetery
information page that provides a community-level view presenting various kinds
of information and visualizations about each cemetery and the soldiers buried
there, including over 3000 photographs of the cemeteries. Third, the new
functionalities of the existing casualties perspective make it possible to study various
community level groups. For example, the people buried in a particular
cemetery or born in a certain town form a prosopographical [
The principal use case of the new views is to provide local level views to the casualty data. This use case originates from the data owners at the National Archives of Finland, who emphasized the idea that the users need a local starting point for browsing the data. The existing person perspective of the WarSampo portal provides answers to the question: “What happened to my relatives in WW2” whereas the new views are meant to expose what happened to the user’s or the user’s relatives’ hometown soldiers in WW2. 2
The WW2 Casualty database is originally collected and maintained by the National Archives of Finland. Based on their experiences, the data users can roughly be divided into three groups: academic researchers, military history enthusiasts and private citizens. The first group has the widest range of needs regarding the data, but they often have the best skills to handle and refine the data by themselves. The focus of academic research regarding this data seems to be shifting from a macro level towards individual and social aspects of war.
Military history enthusiasts usually approach the data from a military unit perspective, or they may concentrate on a certain location during a narrow time frame. They may also be searching for irregularities, such as peaks in numbers of casualties or in certain age groups within the data. One special group of data users in this segment are voluntary researchers, whose aim is to search, locate, and bring back remains of Finnish soldiers who went missing during the battles in the area that is currently part of Russia. They use the data to identify the war victims found during their excavations at the WW2 battle sites.
Private citizens usually begin their search for information with their own relatives who were lost during the war. After finding that out, they may go on searching for similar destinies based on age group, unit, or locations (e.g. home towns or the location where their relatives lost their lives).
As mentioned above, academic researchers usually have the best skills to refine the data for their needs. At the other end are private citizens who are usually most dependent on easy-to-manage user interfaces. There have been no studies done on the data user profiles, but it seems apparent that they form the largest group. In this group are also school children who may benefit from this data during their studies as local history aspects are more and more valued in Finnish national curriculum for basic education.
The WarSampo project is targeted at all three user groups. The WarSampo Data Service publishes all information as Linked Open Data so that academic researchers can, at their choice, download the data and process it further, or query it directly with SPARQL. The WarSampo Portal provides user-friendly applications for all the user groups to search, browse, analyze and visualize the data. 3
An important part in studying histories of communities are cemetery studies,
which provide a physical record of a community’s former inhabitants [
A complete listing of all war cemeteries in Finland has not previously been available, but it has been estimated that there are about 690 of them. The Central Organization of Finnish Camera Clubs (Suomen Kameraseurojen Liitto ry, SKsL) is coordinating a project called “War Cemeteries in Finland” where local camera clubs photograph and collect data about all the war cemeteries.
The camera clubs were instructed to take five specific photographs from each cemetery. At the same time the following information will be collected: 1) official name of the cemetery, 2) year of foundation, 3) architect, 4) memorial (name, sculptor and unveiling date), 5) street address, 6) coordinates, 7) former municipality, and 8) current municipality. The cemetery and photograph data is first organized into a CSV table by a representative of the SKsL, who manually harmonizes the data entry forms sent by the camera clubs. Then the CSV table is converted into RDF9 and linked to existing WarSampo datasets. Photographs and new cemetery data is then shown on the cemetery page. 3.2
For interlinking the heterogeneous datasets of WarSampo, a wartime place
ontology containing place names with different levels of granularity and types was
created earlier as a pilot implementation of the “Finnish Ontology Service of
Historical Places and Maps” [
In order to add the cemetery data to the WarSampo Data Service, the
eventbased data model of WarSampo was extended with a new cemetery class. The
part of the WarSampo data model that is relevant to this paper is presented in
Figure 1. Because the WarSampo data model is founded on the CIDOC
Conceptual Reference Model (CRM) [
The namespaces :, :crm, and :narc used in Figure 1 refer to WarSampo schema (http://ldf.fi/schema/warsa/), CRM, and Casualties schema (http://ldf.fi/schema/narc-menehtyneet1939-45/) respectively. Using the CRM model is not straightforward, and there are hardly any best practices 9 At the time of writing the information collection is still in progress and preliminary data and photographs related to 218 cemeteries have been converted into RDF and published on the WarSampo Data Service. available. A previous modeling decision was that when a class containing custom properties that were not available in the CRM is needed, it will be created as a subclass of a corresponding CRM class, because otherwise we would have to add invalid properties to a CRM class.
Another justification for using subclasses is that the usual way of defining types of instances in Linked Data is with rdf:type pointing to a rdfs:Class, but the CRM model, originating from outside the Linked Data world, uses another way of specifying types. According to CRM the way to include type information is to use the P2 has type property, which is supposed to point to a concept picked from e.g. a hierarchical thesaurus. As we are producing Linked Data, we use the RDFS standard for modeling the types of the instances (e.g. place types, types of military units), which enables more powerful semantics. The full WarSampo schema is published on GitHub10.
The dataset of the Finnish war casualties consists of about 95,000 death
records, and originates from the Finnish National Archives. It has served as the
primary source of person instances in WarSampo. Because it was created before
the WarSampo Data Service, it uses a different namespace for some properties.
As can be seen from Figure 1, the death records are modeled as CRM
documents, which have custom properties for different pieces of information about
the casualty. The death records are linked to corresponding WarSampo person
instances, military units, military ranks, and wartime municipalities [
The death records were originally annotated with information about cemeteries, but this information was limited only to the cemetery name, and the municipality in which the cemetery is located. Because the “War Cemeteries in Finland” project and the original death record database used the same listing of war cemetery names as a starting point, it is a straightforward operation to append the new cemetery data and photographs to the existing cemetery instances by using only the name of the cemetery as a connecting link. The cemeteries provide a more stable basis for local history than the death records’ temporally changing municipalities, which have already changed greatly since the war. 4
The new cemetery perspective and the new visualizations in the casualties
perspective have been developed to gain new insights from the data based on the
community-level aspect provided by the cemeteries. Since there is not enough
data about the casualties to construct life stories of individual soldiers as
biographies, the new views focus on supporting prosopographical study of the war
casualties in which people are studied as groups. This approach is useful when
there is not enough information about individual people to construct
biographies, but the amount of individuals is large enough to study the data as groups
of people using, for example, visualizations. Information visualization is a useful
approach for analyzing complex spatiotemporal and multivariate data [
Figure 3 illustrates the top section of a cemetery information page. The top section contains all information provided by the “War Cemeteries in Finland” project. Below that the user can browse through the photographs of the cemetery with a full screen photo gallery. As is the case with all other WarSampo information pages, the right column contains links to related resources. In this context links to buried people and their units are provided. Below the photographs there is a map visualization based on the places of death of the buried people. This example shows that the death places are concentrated on the eastern front of Finland. By clicking the map pin the user can move onwards to the person information page.
The bottom section of a cemetery information page, partly illustrated in Figure 4, contains several visualizations for studying the buried people as a prosopographical group:
Distribution over units. A pie chart for showing the most common units of one’s hometown soldiers.
Distribution over ranks. The most common ranks as a pie chart. 11 Premilinary version at http://www.sotasampo.fi/en/cemeteries Age distribution. The ages at death as a bar chart. This visualization can also be found in the Casualties perspective with customizable options.
For prosopographical research the pie chart slices and bar chart bars were made interactive, so that by clicking them the user can examine the persons that belong to the group in question and visit their information pages. On the bottom there is a link to the casualties perspective, where the user can study the buried people further as a group with a faceted search interface and newly added visualizations. A design decision originating from the needs of the user groups was to place certain visualizations to the cemetery information page, and leave the more customizable visualizations (e.g. soldier life paths) to the more general casualties perspective. 4.2
The casualties perspective12 of WarSampo provides a faceted search interface
to the Finnish WW2 casualty data as a web page [
The faceted search results have previously been displayed in a data table, but the perspective has been extended to also support visualizing the results based on the facet selections. This provides a way to easily analyze the data by 12 http://sotasampo.fi/en/casualties 13 https://github.com/SemanticComputing/WarSampo-death-records using the faceted search interface to filter the results based on what the user is interested in. For example you could select only soldiers that have been born in a certain place, or that have been buried in a certain cemetery.
After filtering the results with the facets, the user can choose from several different ways of displaying the results from the Results Display drop-down menu at the top of the page. The supported methods of displaying the faceted search results are:
Table. The original method of displaying the casualties as a table. Age distribution. Age distribution of casualties as a column chart. Soldier paths. Soldier life paths visualized as a sankey diagram. The default steps in the diagram are the municipalities of birth, residence and death, and the military cemetery. Steps can be customized by the user by selecting properties to be used for each step.
Statistics. Distribution over an arbitrary property as a bar chart. The user can select the property to use for the diagram.
Figure 5 shows the age distribution of casualties in Hietaniemi cemetery in Helsinki, which is the military cemetery with the highest number of casualties.
Figure 6 presents a screenshot of the soldier life paths diagram, showing the life paths of 40 soldiers. The diagram shows where the soldiers were born, where they lived, where they died, and where they are buried. The birth, residence, and death are given on a municipality level, whereas the cemetery is given as the exact cemetery. In this case the facets have been used to filter the casualties to only show persons buried in the cemetery of Inari in Ivalo. A large portion of places of death are unknown in the data, as is the case here as well. All in all 56% of the casualties in the dataset have an unknown place of death.
The faceted search makes it possible to further narrow the search down e.g. to only include casualties who served in a certain military unit. Figure 7 show the whole user interface of the Statistics visualization view. Based on the facet selections, the diagram currently shows occupations of soldiers with the military rank of private, who were married, were born in Helsinki, lived in Helsinki, and are buried in Hietaniemi war cemetery in Helsinki. The most common occupations in the group are workman, chauffeur and labourer. 5
Applying Linked Data principles to cultural heritage data [
The War Graves Photographic Project15, founded in 2008, aims to create an archive of names and photographs of all military graves and memorials from 1914 to the present day from any nationality, although the focus is on Commonwealth soldiers. Data collection is based on volunteer work, but unfortunately the data is not available through APIs or dumps, and the photographs are subject to a charge.
This paper presented how Finnish WW2 casualty data was linked to war cemetery data that is being collected for studying the casualties from a prosopographical perspective. The WarSampo portal was extended with several new interactive views to the casualty data based on cemeteries and other community level groups. A quick comparison shows that the War Graves Photographic Project has information on 668 people buried in Hietaniemi, the biggest war cemetery in Finland, whereas WarSampo now has information on 4268 people buried there.
The Cemetery perspective and the data visualizations on cemetery information pages are useful for all user groups mentioned in section 2, but because they provide easy access and a starting point from a local point of view, private citizens might benefit most from it. These visualizations offer instant information on divisions, for example, in ranks, age groups and units. By using a local starting point for the whole casualties dataset it is also much easier to detect 15 https://www.twgpp.org/ local irregularities in data that might be worth exploring in more detail. The new visualizations of the Casualties perspective offer more customizable views to the casualty data as a whole, which are especially useful for academic researchers.
While the current tools allow for many ways to analyze and explore the data, there is still a vast amount of different kinds of visualizations that could be tried, e.g. many that would be based on aggregating the data in some manner. For example displaying the average amount of children in relation to some other property, like place of residence, might reveal some new insights.
The casualty and cemetery data collected from various sources contains errors and inconsistencies. To improve the quality of the data, a Linked Data tool for suggesting corrections and collecting new data from the users is being planned. Acknowledgements Our work is funded by the Memory Foundation for the Fallen16, Open Science and Research Initiative17 of the Finnish Ministry of Education and Culture, the Finnish Cultural Foundation, and the Academy of Finland. 16 http://www.muistosaatio.fi/ 17 http://openscience.fi/