The EHRI Document Blog

The EHRI Document Blog (https://blog.ehri-project.eu/) was started as a space to share ideas about Holocaust-related archival documents, and their presentation and interpretation using digital tools. EHRI researchers document their activities and experiment with different ways to explain and show digital archival content. The blog is a showcase of using novel approaches for digital archival research.

The blog also provides inspiration to the technical work packages (described below) as to what functionalities and analyses can be useful to researchers, in order to automate part of their work, and allow them to analyze such large amounts of data that was not possible previously.

The Value of Linking

An early inspiration that helped convince the EHRI Project Management Board of the value of interlinking, was the work performed by the Contemporary Jewish Documentation Center Foundation (CDEC), Milan, in linking their internal databases (archival materials and persons) to Linked Open Data (LOD).

For example, one of the 9k person records that CDEC has is about Primo Levi, an Italian Jewish chemist, memoirist, short story writer, novelist, essayist, and Auschwitz survivor. This person is present in 50 wikipedias (articles about him, e.g. see Primo Levi on en.wikipedia ), 11 wikiquote sites (his sayings), and 2 wikisource sites (his books).

There is quite a lot of structured Linked Open Data (LOD) about him: Primo Levi on wikidata (or see Primo Levi on Reasonator, which is a nice reading interface):

1. Names in a number of languages (including Russian, Chinese, Korean, Arabic): Primo Lévi | Primo Michele Levi | Primo levi | Primo M. Levi | Levi, etc.

2. His prisoner number (174517)

3. Identifiers in a large number (30) of national library systems and other sources, i.e. cross-referencing to the following Authority Files BAV, BNE, BnF, British Museum person-institution, CANTIC-ID, DBNL, FAST-ID, Filmportal, Find a Grave grave, Freebase, GND, IMDb, ISNI, KLfG Critical Dictionary of foreign contemporary literature, LCAuth, Munzinger IBA, NDL, NILF author, NKCR AUT, NUKAT authorities, National Library of Israel, National Thesaurus for Author Names, Open Library, PTBNP, Perlentaucher, RKDartists, SBN, SELIBR, SUDOC authorities, VIAF. Curiously, he has 5 identifiers in the National Library of Israel and 3 ISNI identifiers. Additional info can be obtained from these sources, especially publications.

4. Auto-generated biography (rather dry): Primo Levi was an Italian chemist, poet, politician, writer, and novelist. He played a role in imprisonment at Auschwitz concentration camp. He was born on July 31, 1919 in Turin. He died on April 11, 1987 in Turin.

5. Nationality & professions, each supported by a primary source URL.

6. 19 books that he authored

7. Prizes he has won

8. Who he was influenced by, and who he influenced

9. Familial relations, if available

One of the external sources, the International Movie DataBase (Primo Levi on IMDB), shows that he has credits in 11 films, news about him (5 in 2015 alone), and a longer biography. IMDB is also available as LOD, see LinkedMDB.

CDEC Linked Data

CDEC has connected person names to archival descriptions, the CDEC library catalogue, and external LOD (persons to VIAF and DBpedia, places to Geonames). So when a user searches for a person, they find biographical data, family relations, information about the available resources on that person at the CDEC archives and library. For example, http://digital-library.cdec.it/cdec-web/persone/detail/person-5002/levi-primo.html shows:

1. Links to 39 photos, 5 archival materials, 6 audio files, 2 bibliographic items

2. Link to get LOD data in RDF: http://dati.cdec.it/lod/shoah/person/5002 using the Italian Shoah ontology . Thus CDEC names and biographical data are themselves published as LOD

3. Structured properties about the person and interlinking through familial relations

The RDF data is also published on a SPARQL endpoint (http://lod.xdams.org/sparql) and can be queried, e.g. persons with profession Chemist or persons with destination camp Auschwitz

LOD linking allows CDEC to present a lot more data about the persons.

This work was performed as part of the Open Memory Project , which won the LODLAM Challenge 2015 for its significance in the virtual reunification of families.

EHRI Research Use Cases

One of the first tasks of WP14 was to define Research Use Cases that guide the work not only in WP14, but also technical development and capabilities in the other WPs. For example, which datasets to use as Geographic authorities in WP11 is dictated to some extent by needs identified in WP14. The cases were formulated by historians and EHRI archivists, together with digital humanities researchers and IT specialists. They are as follows:

1. Names and Networks: Holocaust victim communities. Most Jews needed the support of other people to survive. Persons found aiders inside their personal and group networks. Investigate the networks in which European Jews operated during their persecution in the Second World War, and improve understanding of the various chances of survival that persecuted Jews all over Europe had.

2. In Search of a Better Life and a Safe Haven: Tracing the Paths of Jewish Refugees (1933-1945). Map and better understand the different migration trajectories and determine the factors that played a role in the migration movements of migrants or forcefully deported Jews.

3. People on the Move: Revisiting Events and Narratives of the European Refugee Crisis (1930s-1950s). Investigate migration movements of European refugees. The International Tracing Service (ITS), and EHRI partner, has been a key actor in managing this twentieth century migration crisis and has hence built the most important archive documenting the lives of refugees and displaced persons.

4. Between Decision Making and Improvisation: Tracing and Explaining Patterns of Prisoners’ Transfers through the Concentration Camp System. Investigate how the SS in the years 1942-1945, via Hollerith Departments and punch-card technology, managed the transport of inmates through the concentration camp system that covered the whole of occupied Europe

5. Archives and Machine Learning. Investigate how digital methods might support archivists in the creation of interoperable and consistent descriptions of sources (metadata) and in the linking of sources. Estimate metadata quality using data mining and machine learning approaches.

6. Networked Reading. What form historical sources need to take in order to be processed using digital methods? What kind of infrastructure do we need to extract meaning from them? How to apply digital methods in such a way that the results are verifiable as well as reproducible?

There is a lot of connections and cross-pollination between WPs. The same datasets, techniques and developed services can often be used in several WPs.

Geographic Mapping

Consider the following example from the EHRI Document Blog . It presents an oral history about Jews stranded on the border of 3 countries around the Danube. An interactive map is displayed that shows the historical and current borders, the places being discussed, and links them between text and map (i.e. if you click on the map, you see the place name occurrences in the text, and vice versa).

Researchers find such visualizations and tools very useful, but creating them by hand is very effort intensive.

Person Names

Names are some of the most important personal characteristics that allow identification, especially because a lot of the other features in USHMM Persons are sparsely populated.

We are currently investigating what databases of names and name variants we can use. Useful information and considerations include:

1. First names (given names), last names (families or surnames), middle names (and rules for forming patronymics in various cultures)

2. Alternative spellings in a variety of scripts and languages, e.g. transliteration from Hebrew script into Latin

3. Normalization rules (e.g. Unicode normalization for German umlauts, e.g. Niemoeller=Niemöller), Approximate matching (e.g.: Malvine=Malvina=Malvin)

4. Name variants (derivations) in various cultures, e.g.: Giulia (Italian) is similar to Ульяна=Uljana (Russian/Ukrainian) and is similar to Юлия=Julija=Yulya=Yuliya (Russian/Slavic languages)

5. Gender (for first names)

6. Gender-specific morphological rules (for middle and last names). E.g., the wife of a Slavic man Ivanov (Иванов) would be Ivanova (Иванова); whereas in Icelandic the suffixes *son and *dottir are often used

7. Culture- and period-specific probabilities for adopting a different name at marriage (e.g. keeping maiden name, adopting married name, or adopting a doubled name)

8. Diminutive variants; e.g. Yakov is a male first name, which for a boy or adult, in Russia, could be written Yashka (ru-Latn) or Яшка (ru-Cyrl)

We are currently investigating the applicability of various data sources, e.g.:

1. Ancestry.com's name dictionaries for various languages. Ancestry's tool is used for keying part of the USHMM Person info.

2. Yad Vashem's comprehensive database of Jewish names.

3. Wikidata, which has relevant classes, e.g. given name and subclasses male given name, female given name and unisex given name. E.g. the following WDQ query returns a list of 308 Italian given names, including corresponding names in other cultures. The cryptic query claim[31:11879590,31:12308941,31:3409032] and claim[407:652] can be decoded at the WDQ tool and means the following:

The same information can also be obtained from other sources, i.e. the Wikidata project Names coordinates work on gathering and describing names, and has interesting info & resources. We haven't yet assessed how many variants are available.

Finally, we could use various Wikipedia categories and lists to extract raw names, e.g.: https://en.wikipedia.org/wiki/Category:Jewish_given_names ; https://en.wikipedia.org/wiki/Category:Jewish_surnames. Unfortunately these names, e.g. Abendana, have no Inter-Language Links, i.e., the respective Wikidata item has only an English label.

Person Deduplication

The USHMM Holocaust Survivors and Victims Database contains over 3.200.000 person records (as of June 2015) gathered over time from heterogeneous sources. Some of the persons in the database are represented by more than one such record. The records come from different sources - lists of arrests, lists of convoys, etc. Different records may contain various information - some may be more complete than others and even if a bunch of records contain information about the same person, they are not linked to each other. For example, consider the following two records: Zoltan Grun (1) and Zoltan Grun (2)

Since the exploration of person data is one of the most common tasks in the Holocaust research, we consider it important to find an automatized way to link these records and allow for better search and discovery of data related to the survivors and victims of the Holocaust. We call this task record linking or also record deduplication. The final goal of this task is to have a single record representing each person that contains the references to all the separate records related to him available from the different sources.

We solve this problem by a combination of statistical and rule-based models which could be roughly split into the following tasks:

1. Record normalization: we transform the properties to their normalized values.

2. Classification: we classify each person record against a number of similar candidates to find out whether they are indeed duplicates or not.

3. Clustering: we use the result of the classification step as a distance measure between the records and cluster them in groups. Each group of records represent a single person.

These tasks are explained in detail further on.

EAD Conversion

T10.1 developed tooling to convert archival descriptions from various formats to standard EAD XML. We selected XQuery for implementing the transformations, and Java & JavaScript for developing an archivist-friendly front-end. We use the BaseX XQuery implementation, since it has good performance, and an big library of useful extensions, e.g. writing multiple files out of one XML input.

The tool addresses tabular data and XML data (e.g. from Faust archivist databases, SOLR indexes, etc). It provides fixes for various commonly occurring problems per institution, and some special processing tasks (e.g. construction of hierarchies, links to digital material, etc). It has a step-by-step wizard interface and supports a sophisticated conversion mechanism that is driven by mappings expressed in Google sheets. Extensive documentation and online tutorials are available . The tool is open sourced and is available on Github (https://github.com/EHRI/ehri-conversion-tools)

For each institution we first develop a conceptual mapping from the institution data to EAD.

We then develop a physical mapping that implements the conceptual mapping using Xpaths.

This mapping then can be selected by the user in the EAD conversion tool, which uses it on-line from Google Sheets. This ensures that the mapping is always up to date. The tool also generates transformation scripts that can be executed in batch mode, e.g. as a scheduled task.

We are even considering for conversion some textual descriptions that use consistent field names, e.g. 1994.62_01_fnd.

EAD Validation

The EAD tool also provides EAD validation and preview as HTML. We perform validation at 3 points:

1. As final step of the conversion, so the archivist can check his work for errors

2. Before publishing (transporting) a set of EAD to the EHRI portal (see the next section)

3. Before ingesting EAD to the EHRI portal

Validation rules are provisioned using TEI ODD files, generate RelaxNG and Schematron (see below) and provide links to EAD taglib documentation for better diagnosing of the error: https://github.com/EHRI/data-validations. We validate EAD against two different set of rules:

1. The official EAD2002 schema, plus a few EHRI-specific rules (e.g. each documentary unit should have an id)

2. An internal EHRI EAD schema. It's almost the same, but has some EHRI specific adaptations (e.g. paragraphs on the EHRI portal are not formatted with

This allows a variety of data flows to be implemented in different combinations: using the conversion tool, using the publishing tool, or direct deposit to the EHRI ingest directory.

1. We validate EAD using both RelaxNG XML schemas and Schematron rule checking. RelaxNG is a better way to express XML schemas than XSD, and RelaxNG Compact is an intuitive and compact notation (e.g. see https://github.com/VladimirAlexiev/rnc). We made RelaxNG from the official EAD2002 schema (whereas EAD3 is developed completely in RelaxNG. This checks the structural integrity of EAD files.

2. We also use Schematron rules for additional checks. These are used for cross-field validation, conditional checks, etc.

3. We extended the popular schema jing-trang validator to report schema errors in the Schematron Validation Report Language (SVRL), which is a simple XML format to describe the source of error using XPath. See https://github.com/EHRI/jing-trang and in particular pull request 1. This way the validator deals with one kind of error, no matter whether it's a schema or Schematron rule violation.

The EAD validator integrates the error reports into the HTML preview, making it more suitable for a non-technical audience. First we generate an index <timestamp>/html/index.html showing all produced EAD files and the number of errors in each. From this index one can access the individual preview+errors. For example, the next figure shows that the Profile Description element is not allowed, because there is a missing eadid element. In order to correct the error, one must add a eadid element. Depending on the data flow, one can correct them in the input file, the mapping configuration, or the XML file.

EAD Publishing/Transport

T10.2 has developed an EAD Publishing (transport) tool based on OAI ResourceSync. OAI ResourceSync can be used to transport metadata, allowing discovery of new and updated metadata files. It is easier to deploy than OAI PMH because it does not require a server, but institutions that already have an OAI PMH server can also be accommodated by EHRI. It consists of two parts:

1. https://github.com/EHRI/resyto implements a GUI that allows the user to select files, applies updating (synchronization) logic and produces ResourceSync manifests. It implements the source part of the ResourceSync Framework

2. https://github.com/EHRI/resydes visits preconfigured ResourceSync endpoints, gets the manifests, checks for new or updated files since the last run, and transports them to the EHRI server. It implements the destination part of the ResourceSync Framework

EAD Ingest

T10.3 will improve the existing ingest of EADs to the EHRI database. It will enable:

1. updating (synchronization)

2. co-referencing of textual Access Points to proper Authority references (see next section)

3. preservation of links created manually in the EHRI portal

4. exposing of links to finding aids and digital materials on the EHRI portal.

EHRI-1 Thesauri

EHRI-1 has published linked data (http://data.ehri-project.eu/) that includes the following authorities:

1. 880 concepts (the EHRI thesaurus)

2. 1970 camps

3. 1000 ghettos

4. 400 people

5. 260 Nazi administrative districts

6. 50 events. These are major historic events that one can find on Wikipedia, e.g. Blood for Goods.

Camps and Ghettos

EHRI-1 had collected a master list of Concentration Camps and Ghettos, but it included minimal information: name in a few languages (e.g. Maly Trostinec@de-latn) and in some cases sub-camp relations (hierarchy). Wikipedia and Wikidata include a lot more info about camps.

E.g., Wikipedia has the following info on Maly_Trostenets_extermination_camp. References are provided, but the info is not structured:

1. Names: Maly Trostinets, Maly Trastsianiets, Trasciane, Малы Трасцянец, Maly Tras’tsyanyets, Малый Тростенец, Maly Trostinez, Maly Trostenez, Maly Trostinec, Klein Trostenez

2. Location: outskirts of Minsk

3. Admin district: Reichskommissariat Ostland

4. Established: 10 May 1942

5. Killing grounds: Blagovshchina (Благовщина) forest, Shashkovka (Шашковка) forest

1. Victim countries: predominantly Belarus (inferred, not explicitly stated). Also Austria, Germany, Czech Republic.

2. Victim places of origin: predominantly Minsk. Also Berlin, Hanover, Dortmund, Münster, Düsseldorf, Cologne, Frankfurt am Main, Kassel, Stuttgart, Nuremberg, Munich, Breslau, Königsberg, Vienna, Prague, Brünn, Theresienstadt.

3. Known victims: Vincent Hadleŭski (Wincenty Godlewski): arrested in Minsk on December 24, 1942 and shot at Trascianiec the same day; Norbert Jokl (debated); Margarete Hilferding (in transit to the camp from Terezín); Grete Forst; Cora Berliner (most likely)

4. Perpetrators (and their fate): SS Unterscharführer Heinrich Eiche (fled to Argentina after the war and all trace of him was lost); Eduard Strauch (died in Belgian prison in 1955); Rottenführer Otto Erich Drews (in 1968 the Court in Hamburg sentenced to life imprisonment); Revieroberleutnant Otto Hugo Goldapp (in 1968 the Court in Hamburg sentenced to life imprisonment); Hauptsturmführer Max Hermann Richard Krahner (in 1968 the Court in Hamburg sentenced to life imprisonment); Heinrich Seetzen (committed suicide in a British POW camp); Gerhard Maywald (settled after the war in West Germany; on August 4, 1977 sentenced to 4 years imprisonment); Jewish Sonderkommando 1005

Wikidata (Q316109) knows the following structured info:

1. Names and Wikipedia links in the following languages: Беларуская, Беларуская (тарашкевіца)‎, Čeština, Dansk, Deutsch, Español, Français, Frysk, Italiano, עברית, Nederlands, Norsk bokmål, Polski, Português, Русский, Српски / srpski, Suomi, Svenska, Українська, 中文

2. Additional aliases, e.g. Vernichtungslager Maly Trostinez, KZ Maly Trostinez, Blagowschtschina

3. Country: Belarus

4. Location: 53°51'3N, 27°42'17E

5. Authority identifiers: Geonames, VIAF, Freebase

DBpedia (Maly_Trostenets_extermination_camp) knows the following structured info:

1. links to Wikidata, Geonames, Freebase, different Wikipedias

2. coordinates

3. a few more aliases: Maly_Tras’tsyanyets, Maly_Tras’tsyanyets_camp, Maly_Tras’tsyanyets_concentration_camp, Maly_Tras’tsyanyets_extermination_camp

The fact that it is DeathPlace of the following people. This comes from the articles about these people (i.e., inverse links): dbr:Margarete_Hilferding, dbr:Grete_Forst, and dbr:Vincent_Hadleŭski.Categories: dbc:World_War_II_sites_of_Nazi_Germany, dbc:Geography_of_Minsk, dbc:History_of_Belarus_(1939–1945), dbc:History_of_Minsk, dbc:Maly_Trostenets_extermination_camp, dbc:The_Holocaust_in_Belarus, dbc:World_War_II_sites_in_Belarus, dbc:Belarus_in_World_War_II.

Wikidata templates about specific camps have even more info, e.g. Template:Treblinka_extermination_cam p includes detailed lists of perpetrators, victims, resistance, etc., all linked to Wikipedia:

1. Camp organizers: Odilo Lotario Globocnik, Hermann Julius Höfle, Erwin Hermann, Lambert Richard, Wolfgang Thomalla, Christian Wirth

2. Commandant: Irmfried Eberl, Franz Paul Stangl, Kurt Hubert Franz

3. Deputies: Theodor van Eupen, Heinrich Arthur Matthes Karl Pötzinger

4. Gas chamber executioners: Gustav Münzberger, Fritz Schmidt

5. Other officers: Max Biala, Paul Bredow, Herbert Floss, Erich Fritz Erhard Fuchs, Lorenz Hackenholt, Hans Hingst, Josef Hirtreiter, Otto Richard Horn Kurt Küttner, Karl Emil Ludwig Willy Mätzig, Willi Mentz, August Wilhelm Miete, Max Möller, Willi Post, Albert Franz Rum, Karl Schiffer, Otto Stadie, Ernst Stengelin, Franz Suchomel

6. Guards: Ukrainians: Ivan the Terrible, John Demjanjuk, Feodor Fedorenko, Nikolay Yegorovich Shalayev; Others: Trawnikis, Volksdeutsche

7. Prominent victims: Ernst Arndt, Yitzchok Breiter, Amalia Carneri, Julian Chorążycki, Samuel Finkelstein, Artur Gold Ludwik Holcman, Janusz Korczak, Berek Lajcher, Henryka Łazowertówna, Yechiel Lerer, Yitzchak Lowy, Simon Pullman, Natan Spigel, Symche Trachter, Zygmunt Zalcwasser, Lidia Zamenhof

8. Resistance, Survivors: Richard Glazar, Chil Rajchman, Sol Rosenberg, Kalman Taigman, Jankiel Wiernik, Samuel Willenberg, Franciszek Ząbecki

9. Nazi organizations: General Government SS-Totenkopfverbände

10. Planning Methods, Documents, Evidence: Operation Reinhard Höfle Telegram

11. Aftermath, Memorials: Treblinka trials

12. Related topics: The Holocaust Operation Reinhard, Nazi concentration camps, Extermination camp

Recognizing the value of this data and working together with the LOD community, EHRI decided in Feb 2017 to enrich and integrate data on Camps and Ghettos using Wikidata as a data integration platform. We first created or enriched Wikidata entities and properties to use in the matching process, e.g.:

1. USHMM Holocaust Encyclopedia (Q5883924). Extended the Wikipedia article Holocaust_Encyclopedia with some numbers and website https://www.ushmm.org/learn/holocaust-encyclopedia.

2. Yad Vashem Encyclopedia of the Holocaust (Q2906963). Represented at Wikipedia article Encyclopedia_of_the_Holocaust

3. EHRI Project (Q21755493)

4. USHMM Holocaust Encyclopedia ID (P3724): identifier of camp/ghetto to link to this encyclopedia. See property proposal.

5. Yad Vashem Encyclopedia of the Ghettos ID (P3735): identifier of a ghetto, or a place containing a ghetto. See property proposal.

6. USHMM person ID (P4130): identifier for a person in the UHSMM HSV, a database of Holocaust survivors and victims. See property proposal.

We then used data from many places:

1. The two institutional encyclopedias mentioned above.

2. Querying Wikidata by a set of relevant types, e.g.: https://www.wikidata.org/wiki/Q328468 Nazi concentration camp; https://www.wikidata.org/wiki/Q152081 Concentration camp; https://www.wikidata.org/wiki/Q1719244 Subcamp; https://www.wikidata.org/wiki/Q628505 Labor camp; https://www.wikidata.org/wiki/Q14916829 transit camp; https://www.wikidata.org/wiki/Q15727645 internment camp

3. Querying DBpedia by a number of Yago types, e.g.: http://dbpedia.org/class/yago/WorldWarIISitesOfNaziGermany , http://dbpedia.org/class/yago/ConcentrationCamp103086183 , http://dbpedia.org/class/yago/Camp102945379 , http://dbpedia.org/class/yago/NaziConcentrationCamps , http://dbpedia.org/class/yago/NaziConcentrationCampsInAustria , http://dbpedia.org/class/yago/NaziConcentrationCampsInBelgium , http://dbpedia.org/class/yago/NaziConcentrationCampsInDenmark , http://dbpedia.org/class/yago/NaziConcentrationCampsInEstonia , http://dbpedia.org/class/yago/NaziConcentrationCampsInFrance , http://dbpedia.org/class/yago/NaziConcentrationCampsInGermany , http://dbpedia.org/class/yago/NaziConcentrationCampsInItaly , http://dbpedia.org/class/yago/NaziConcentrationCampsInLatvia , http://dbpedia.org/class/yago/NaziConcentrationCampsInLithuania , http://dbpedia.org/class/yago/NaziConcentrationCampsInNorway , http://dbpedia.org/class/yago/NaziConcentrationCampsInPoland , http://dbpedia.org/class/yago/NaziConcentrationCampsInTheNetherlands , http://dbpedia.org/class/yago/NaziConcentrationCampsInUkraine , http://dbpedia.org/class/yago/NaziConcentrationCampsInYugoslavia , http://dbpedia.org/class/yago/NaziConcentrationCampsOnAlderney , http://dbpedia.org/class/yago/WorldWarIIConcentrationCamps , http://dbpedia.org/class/yago/WorldWarIICroatianConcentrationCampsInFormerYugoslavia

1. Extracted a list of 4174 camps/ghettos from USHMM places by looking for names that include patterns like camp|ghetto|lager|kamp|лагер|KZ etc. We found 58 patterns that indicate a camp, e.g.: KLV camp, GTE camps?, POW camp, Prisoner of War Camps?, camp prisoner of war, camp v/pl, Labour camp, KriegsGefangenenLager, GemeinschaftsLager, K\. Lager, K\.z\. ?Lager, KZ, unbekanntes Lager, Lager unbek\., Lager unbekannt, Arbeitslager, FrauenArbeitsLager, StrafLager, Reichsbahn Lager, LAGERKÜCHE, WaldLager, Firmen?Lager, SammelLager, isprav trud lager', Isprav\. trud\. lager', Ispr\.Trud\.lager', ispr\. trudovoi lager', isprav\.trudovoi lager', Ispravitel'nyi trudovoi lager', Smertnyj Lager, Lagerei, Lageri.

We also found 20 place names that do not indicate a camp (false positives), e.g. Campagna, Campagne, Campan, Campana, Campania, Campbell, Campiniac, Campitelli, Campobasso, Campobello, Campos, Danekamp, Eichkamp, Kampen, Kampina, Kampinos, Lagerstr.

Not all the above USHMM places represent new camps (not present in the EHRI-1 list). We performed comprehensive research (although a partial evaluation) and estimate as follows:

1. 67% or 2797 places are new camps, e.g. Akmecetka (Acmecetca, Akmeketka, Акмечетка), Odessa, Ukraine/USSR; Ahlem, Germany, “uncovered Hannover-Ahlem (April 10, 1945) and Salzwedel (April 14, 1945), both satellite camps of Neuengamme”

2. 11.34% or 473 places are duplicates, but may provide valuable altLabels, e.g. Pavlou Mela (Salonika / Makedhonía / Greece

3. 15.46% or 645 are not valid camp names, e.g.: Bulgarien/unbekanntes Lager: unspecific; Fécamp: false match on name

4. 3.09% or 129 are Displaced Person (DP) camps that are not Nazi camps but are relevant to the Holocaust, e.g.: Kloster Indersdorf displaced persons camp: charitable camp at a monastery; Neustadt [DP camp]: DP camp in the British zone

We also considered the CIA Nazi camps list published on 7 May 1945 and titled CIA: AXIS CONCENTRATION CAMPS AND DETENTION CENTRES REPORTED AS SUCH IN EUROPE. Basic Handbook KLs (Konzentrationslager). Originally published at the CIA Library, it is 187 pages long and includes info about 615 camps (and 93 cross-references) grouped by country. EHRI performed extensive cleanup and regrouping by country. See the following chart.

A massive matching and cross-referencing task was undertaken. As a result, we have merged several thousand Ghetto records across encyclopedias and contributed them to Wikidata, taking coordinates and associated place hierarchy in return . As a result, the number of ghetto records on Wikidata has grown drastically.

Thesaurus Management System and Editorial Board

The EHRI-1 thesaurus was managed by ad-hoc means and by a small number of people. To make this activity more sustainable and facilitate the translation to a much larger number of languages, EHRI established an Editorial Board and deployed the VocBench web-based thesaurus management system (TMS), which allows multiple users and a variety of roles.

VocBench is the best open source TMS that works directly over SKOS RDF data. SKOS is a widely used ontology for representing thesauri (see SKOS Primer, SKOS Reference, and ). VocBench works directly over Ontotext GraphDB and is used by UN FAO, EC's Publication Office, etc.

Access Points

The current version of the EHRI portal has full-text search. This can find documents by EAD access points (a.p.) mentioned in the document. However, it has the following deficiencies:

1. The same concept may be spelt in many different ways (see below)

2. It doesn't know multilingual labels of the concepts

3. It doesn't know hierarchical relations: Eg. if a document mentions Lodz, it won't be returned in a search for Poland.

There is a staggering variety of writing out the same concept across institutions. Eg below are different ways of spelling Łódź (note: the first couple use Unicode combining diacritics instead of simple accented chars). Some of them are mapped to EHRI-1 access point objects, others are plain text:

Łódź (Ghetto)

Łódź (Poland)

Łódź (Poland),.

Łódz (Poland)

Łódź ehri-ghettos-513 cvocConcept

Łódź jc-places-place-iti-48 cvocConcept

Łódź terezin-places-place-iti-48 cvocConcept

Łódź (Poland)

Łódź – Łódź – Polen Łódź – Łódź – Poland

Lodz

Lodz terezin-places-place-iti-412 cvocConcept

Lodz (Polen)

Lodz - Poland

Lodz Ghetto

Lodz ghetto

Lodz ghetto, Poland

Lodz, Poland, Eastern Europe,

Lodz,Ghetto,Poland

Lodz,Lodz,Lodz,Poland

Lodz,גטו,Poland (note: גטו means Ghetto)

Lodž

Lodž jc-places-place-iti-48 cvocConcept

Lodž terezin-places-place-iti-48 cvocConcept

Ladzogne

Leitzmenstadt

Lidzmanstadt

Litmannstadt

Litrwanstach

Litsmannstadt

Litz

Litzen

Litzm

Litzmandstadt

Litzmannsstadt

Litzmannst

Litzmannstadrt

Litzmannstadt

Litzmannstadten

Litzmansntadt

Litzmanst

Litzmanstadt

Litzmaustadt

Lizmannstadt

Lodcsh

Lodez

Lodezogne

Lodon

Lodsch

Lodyogne

Lodz

Lodzen

Lodzi

Lodzogne

Losch

Lotdz

Lotz

Lotzogne

Lòdz

Lód

Lódz

Lódzen

Lödz

lodz

We have assembled a whole catalog of errors. A few of them are shown below:

1. Lack of capitalization: maingain, olivier; maingain, roger; maires—belgique; maires--mechelen (belgique).

2. Lot of mis-typed a.p., e.g. these should be corporateAccess instead of subjectAccess:8f(United Nations Educational, Scientific & Cultural Organization [UNESCO])8g(Organisation f. Ernährung u. Landwirtschaft [FAO]). These should be placeAccess instead of subjectAccess: 915 (Balta, Chersson), 915 (Feodosia), 93. Koblenz.

3. Compound a.p. that should be split up and applied as separate access point: Abortion—Poland--Oswiecim. Abortion—Poland--Warsaw. Abortion—Poland. Abortion.

4. Wrong info (system of arrangement) mis-represented as subjectAccess: 5(dt. Orte in alphabet. Folge); 5(dt. Orte, alphabetisch); A-Z.

5. Just a number without any text, i.e. an access point with no information content: 1; 100; 104. ID; 1042.

We used the Geonames co-referencing service (see 2.2.1 Geo-Referencing Service) to match geographic access points to Geonames, thus bringing multilingual access, unification of a variety of a.p. to the same entity, and hierarchical place access.

Other kinds of access points (subject, person, family, corporate) are still an open challenge. We have investigated the feasibility of co-referencing Person access points to VIAF and USHMM Persons, but research is still ongoing.

The challenges for corporate a.p. are especially hard, since it's hard to recognize them as designating an organization. E.g. consider this subjectAccess a.p. 8. Florian Geyer

This must be the organization 8th SS Cavalry Division Florian Geyer and not the person Florian Geyer (Franconian nobleman, diplomat, and knight), after whom the SS division was named. But because the a.p. is mis-typed (it should be corporateAccess not subjectAccess), that is hard to classify/interpret.

https://en.wikipedia.org/wiki/Primo_Levi

https://www.wikidata.org/wiki/Q153670

https://tools.wmflabs.org/reasonator/?q=Q153670&lang=en

http://www.imdb.com/name/nm0505485/

http://www.imdb.com/name/nm0505485/bio?ref_=nm_ov_bio_sm

http://www.linkedmdb.org/

http://summit2015.lodlam.net/2015/04/21/challenge-entry-open-memory-project/

https://en.wikipedia.org/wiki/Reichskommissariat_Ostland

https://portal.ehri-project.eu/vocabularies/admindistricts

http://www.ushmm.org/online/hsv/source_view.php?SourceId=20990

http://collections.ushmm.org/search/catalog/irn518183

http://collections.ushmm.org/oh_findingaids/RG-50.469.0006_trs_en.pdf

https://www.ushmm.org/online/hsv/person_view.php?PersonId=4225334

http://collections.ushmm.org/oh_findingaids/RG-50.661.0001_trs_en.pdf

https://www.ushmm.org/online/hsv/person_view.php?PersonId=3120613

https://www.ushmm.org/online/hsv/source_view.php?SourceId=20874

https://www.google.bg/maps/dir/Lublin,+Poland/Peremyshlyany,+Lviv+Oblast,+Ukraine/@50.4566024,22.4417333,8z/data=!3m1!4b1!4m13!4m12!1m5!1m1!1s0x472257141e154061:0x5528ee7af6e8e95f!2m2!1d22.5684463!2d51.2464536!1m5!1m1!1s0x473a98e613b80eb1:0xe95d759ed61f558e!2m2!1d24.5593173!2d49.6692136?hl=en

https://www.ushmm.org/online/hsv/person_view.php?PersonId=3120613

http://tools.wmflabs.org/autolist/autolist1.html?lang=it&q=claim%5B31%3A11879590%2C31%3A12308941%2C31%3A3409032%5D%20and%20claim%5B407%3A652%5D

http://wdq.wmflabs.org/wdq/

https://www.wikidata.org/wiki/Wikidata:WikiProject_Names

https://en.wikipedia.org/wiki/Abendana

http://collections.ushmm.org/findingaids/RG-31.094_01_fnd_en.pdf

http://collections.ushmm.org/oh_findingaids/RG-50.005.0047_trs_en.pdf

http://collections.ushmm.org/oh_findingaids/RG-50.488.0016_trs_en.pdf

^{https://www.ushmm.org/remember/the-holocaust-survivors-and-victims-resource-center/holocaust-survivors-and-victims-database}

^{https://www.ushmm.org/online/hsv/person_view.php?PersonId=3428503}

^{https://www.ushmm.org/online/hsv/person_view.php?PersonId=5843067}

https://portal.ehri-project.eu/units

https://portal.ehri-project.eu/institutions?data=yes

https://portal.ehri-project.eu/countries

https://portal.ehri-project.eu/institutions

https://docs.google.com/document/d/1qO2L2N_UeTeTafZxa7nKWDyZvYCZnqHi8RPhCsHMtmg/edit

http://collections.ushmm.org/findingaids/1994.62_01_fnd_en.pdf

TEI: Getting Started with ODDs. http://www.tei-c.org/Guidelines/Customization/odds.xml

https://en.wikipedia.org/wiki/Schematron

https://github.com/EHRI/jing-trang/pull/1

http://www.openarchives.org/rs

https://www.openarchives.org/pmh/

https://en.wikipedia.org/wiki/Category:Blood_for_goods

https://en.wikipedia.org/wiki/Maly_Trostenets_extermination_camp

https://www.wikidata.org/wiki/Q316109

http://live.dbpedia.org/page/Maly_Trostenets_extermination_camp

https://en.wikipedia.org/wiki/Template:Treblinka_extermination_camp

https://www.wikidata.org/wiki/Q5883924

https://www.wikidata.org/wiki/Q2906963

https://en.wikipedia.org/wiki/Encyclopedia_of_the_Holocaust

https://www.wikidata.org/wiki/Q21755493

https://www.wikidata.org/wiki/Property:P3724

https://www.wikidata.org/wiki/Wikidata:Property_proposal/USHMM_Holocaust_Encyclopedia_id

https://www.wikidata.org/wiki/Property:P3735

https://www.wikidata.org/wiki/Wikidata:Property_proposal/Ghetto_Encyclopedia_ID

https://www.wikidata.org/wiki/Property:P4130

https://www.wikidata.org/wiki/Wikidata:Property_proposal/USHMM_person_ID

Confirmed at https://ro.wikipedia.org/wiki/Lagărul_de_concentrare_Acmecetca

Confirmed at http://www.tabletmag.com/scroll/194615/kissinger-on-liberating-ahlem-concentration-camp and http://faculty.jou.ufl.edu/croberts/documentary/angelofahlem/thecamp.html

https://www.wikidata.org/wiki/Q39487700

https://www.cia.gov/library/readingroom/docs/GERMAN%20CONCENTRATION%20CAMPS_0001.pdf

http://vocbench.uniroma2.it/

https://www.w3.org/TR/skos-primer

https://www.w3.org/TR/skos-reference