Development history of the SIstory portal

The SIstory portal has a relatively long history of development. Since its initial release in 2008, several versions of the portal have been released as individual upgrades. In 2011, the first software and technological upgrade was carried out, establishing the latest standards and enabling faster and more stable system operation. This upgrade also played an essential role in establishing a national digital infrastructure for the humanities and arts Ana Cvek, Mihael Ojsteršek, and Mojca Šorn, Izhodišča metapodatkovnih sistemov portala Zgodovina Slovenije – SIstory (2008–2016) (Ljubljana: Institute of Contemporary History, 2022), accessed on 26 February 2025, https://sidih.github.io/izhodisca/index.html.. The first upgrade consisted of several components:

Content administration in Apache SOLR Overview of the Solr Admin UI | Apache Solr Reference Guide 6.6, https://solr.apache.org/guide/6_6/overview-of-the-solr-admin-ui.html. and upgrading folder structures and file names. Implementation of the Dublin Core metadata standard (DC) DCMI: Dublin Core™ Metadata Element Set, Version 1.1: Reference Description, https://www.dublincore.org/specifications/dublin-core/dces/.. The original schema contained all 15 basic DC elements. A year later, the original schema was upgraded with elements from the qualified DCMI Metadata Terms (DCTERMS DCMI: DCMI Metadata Terms, https://www.dublincore.org/specifications/dublin-core/dcmi-terms/.). Creation of a unique and permanent URN – Uniform Resource Name. Introduction of the Sphinx Sphinx | Open Source Search Engine, https://sphinxsearch.com. metadata search engine. Two search engines were implemented: a basic and an advanced one. Upgrading of the portal administration. Design of the structure and access levels for users.

In 2013, the portal was updated to introduce the SIstory metadata schema, a customised set of metadata elements developed to better reflect the nature of the content, which had outgrown the descriptive capabilities of the DCTERMS element set. A major update followed in 2016, during which a mapping between the SIstory schema and Dublin Core was established to enhance data interoperability. The schema was then expanded with elements and structures from the HOPE application profile, Bert Lemmens, Joris Janssens, Ruth V. Dyck, Alessia Bardi, Paolo Manghi, Eric Beving, Kathryn Máthé, Katalin Dobó, and Armin Straube, Hope – The Common HOPE Metadata Structure, Including the Harmonisation Specifications (D2.2) (Tech. Rep., HOPE, Deliverable D2.2, 2011). a recognised standard in the GLAM community, resulting in the SIstory application profile. Andrej Pančur, Metapodatki portala Zgodovina Slovenije-SIstory (Tech. Rep.) (Ljubljana: Institute of Contemporary HIstory, 2013). Cvek et al., Izhodišča metapodatkovnih sistemov. Katja Meden, “‘Posmrtno življenje posmrtnih mask’: sodelovanje Raziskovalne infrastrukture slovenskega zgodovinopisja Inštituta za novejšo zgodovino z Društvom za domače raziskave”, in Odlivanje smrti: posmrtne maske v slovenskih javnih zbirkah (Ljubljana: Institute of Contemporary History, 2023), accessed on 26 February 2025, https://sistory.github.io/Odlivanje_smrti/ch02.html. This profile has since served as the portal’s primary metadata standard, shaping the structure, syntax, and semantics of the metadata input tool. Besides the metadata enrichments, new system frameworks and graphical designs for both the administration and user interfaces were also installed. In addition, the search engine (filtering and sorting of results; full-text search) was also taken into account in the updates. Overall, since the portal’s inception in 2008, a series of upgrades have been made, each improving the portal’s functionalities and features. The entire development history is documented in Cvek et al. Cvek et al., Izhodišča metapodatkovnih sistemov. This brings us to today and to new steps in the portal’s development – the decision to redesign the portal from the ground up.

Technical design

In terms of technical composition, the redesigned SIstory 5.0 portal is based on a robust technical framework, while the backend utilises Django for efficient data management and content delivery. On the frontend, SIstory employs Next.js and React in combination with Node.js for dynamic and interactive user interfaces. This modern frontend stack enables smooth navigation and responsive design across various devices, improving accessibility for users accessing historical content. Figure 1 shows an example of the redesigned user interface. The portal’s database architecture is based on PostgreSQL and provides a robust foundation for storing and retrieving large volumes of historical data with high speed and reliability. In addition, SIstory integrates Matomo, an analysis function that enables administrators to gain valuable insights into user behaviour and interaction patterns, thus forming the basis for future developments and improvements.

For efficient search functionality, SIstory incorporates Elasticsearch and Kibana, allowing users to quickly locate relevant historical documents and sources. The use of Elasticsearch ensures fast and accurate search results, improving the overall usability of the portal. In addition, SIstory employs a Handle Handle.Net Registry, https://www.handle.net/. system to provide permanent identifiers, enabling reliable and permanent access to specific historical documents and sources. This allows users to reference and cite materials consistently, contributing to the scholarly integrity and reliability of the portal. Overall, SIstory’s technical specifications underline the portal’s commitment to providing a robust and user-friendly platform for accessing Slovenia’s rich historical heritage.

Metadata design

The portal previously based its metadata element set on the SIstory application profile (SIstory AP), which was derived from the HOPE application profile. In practice, this posed a problem as SIstory AP contained several elements (and element groups) that were not used as frequently as initially assumed. This, in turn, led to a simplification of the profile. To this end, an analysis of the existing SIstory AP was conducted to identify the metadata elements that should be retained and address those that present legacy issues.

The current state of the metadata application profile comprises 26 elements (reduced from the original 33 elements), with a focus on the DC and DCTERMS metadata elements and only a few additional elements from the previously mentioned HOPE AP. One of the main reasons for this shift in focus is to improve the interoperability of our metadata. Only a limited number of elements of the “SIstory” namespace have been retained For example, SIstory Unstored – a field for storing metadata that do not fit into any other metadata field due to their content., primarily due to the remnants of older publications described with these specific metadata elements. The overview of the major metadata groups is presented in Table 1.

Table 1: Overview of the most important metadata groups, the number of unique instances, and the total number of occurrences on the SIstory portal (at the time of writing this paper). Metadata Unique Values Nr. of Usages No. of entries 62.656 Creator 5.319 25.461 Subject 24.236 333.667 Publisher 1.091 57.686 Collection 432 1.895 Contributor 1.322 44.088 Type 12 63.362 Language 61 74.853 Source: Own work

In total, SIstory comprises over 60,000 unique entries and more than 5,000 unique authors/physical persons (under the category “Creator”) The metadata mask includes two separate fields for the Creator, which, according to the Dublin Core definition, can be either a physical person or a legal “organisation.” In Table 1, however, only occurrences of a physical person/author are counted under the “Creator” category., while “Subject” contains keywords that describe the publications. Secondary forms of authorship are described in the category “Contributor” (e.g. editor, translator...), while the type of publication based on the controlled vocabulary (DCMI Type DCMI: DCMI Metadata Terms, https://www.dublincore.org/specifications/dublin-core/dcmi-terms/#DCMIType. ) encompasses 12 categories. Finally, the portal includes publications in 61 different languages, which are presented in more detail in the Language trends subchapter.

Publication trends

One of the trends analysed in this section is the distribution of publications over time according to their publication date, to determine which years are best represented in terms of published content. The results are shown in Figure 2.

One of the first trends to emerge is the distribution of publications between approximately 1715 and the mid-20th century, which exhibits several significant spikes in the number of publications. Conversely, the post-World War II period displays a much more consistent flow. The spikes in the timeline are most likely caused by the publication of several large works or specific types of publications (e.g., textbooks, population censuses) on the portal. In contrast, the steady publication flow from the post-war period to the present day suggests a greater variety of publication types (e.g., literature, research and studies, monographs, etc.) and the absence of very large volumes of similar publications. The nature and content of these publication trends are examined in more detail in the keyword analysis, which provides additional insight and substantiates the reasons for the identified trends.

Keyword analysis

The keyword analysis of the portal content examined the ten most frequently used keywords to describe sources in the individual menus for each of the 20 years (except the period 2010–2024). This analysis was conducted for the first-level menus: Viri (Sources), Literatura (Literature), Dogodki (Events), Podatki (Data), and DH (Digital Humanities). The results are presented in the following sections Some keywords are very similar to one another, mostly due to slight variations in the notation format. For example, “uradni list” and “uradni listi” are the singular and plural forms of the same keyword but are counted separately..

Language trends

In addition to keyword analysis, we also examined the languages of the publications within the SIstory portal, as shown in Figure 3.

It is not surprising that the most frequent language of publications on the portal is Slovenian, with a total of 46312 occurrences, mainly in the period from 1970 to 2025, especially between 1990 and 2025. The second most common language is Serbian (8201 occurrences), although an explicit distinction must be made here, as Serbian also falls into two other language categories: Serbian (Cyrillic) for publications in Cyrillic script (3947 in total) and the Bosnian/Croatian/Serbian category for publications where the language could not be explicitly identified (mostly publications referring to the Yugoslav Official Gazettes), which are the most frequent among the publications.

Figure 3: Language trends – distribution of publication languages over time Source: Own work

Publications in Serbian across all mentioned categories were most frequently issued between 1930 and 1990, mainly in connection with official gazettes.Another commonly used language is German, particularly during the 19th century – especially between 1850 and 1870 – when many publications related to Slovenian history were produced in the Austrian Empire (1804–1867) and Austria-Hungary (1867–1918), often written in both German and Slovenian. While German was less prominent in the late 18th and early 19th centuries (1710–1830), it became more notable in conjunction with Slovenian in later decades. Less frequently used languages – each appearing fewer than five times – include Spanish (Castilian), Latin, Polish, Arabic, Albanian, Ukrainian, and Esperanto. Esperantsko-slovenski in slovensko-esperantski slovar, accessed on 26 February 2025, https://hdl.handle.net/11686/38289. Additionally, 42 publications are multilingual, and some items have no associated language. This is especially true of the digitised collection “The Casting of Death”, a series of death masks detailed in Meden and Pančur. Katja Meden, “‘Posmrtno življenje posmrtnih mask’: sodelovanje Raziskovalne infrastrukture slovenskega zgodovinopisja Inštituta za novejšo zgodovino z Društvom za domače raziskave,” in Alenka Pirman, ed., Odlivanje smrti: posmrtne maske v slovenskih javnih zbirkah (Ljubljana: Institute of Contemporary History, 2023), accessed on 26 February 2025, https://sistory.github.io/Odlivanje_smrti/ch02.html. Andrej Pančur, Alenka Pirman, and Maruša Dražil, “Spregledana kulturna dediščina in uporaba digitalne raziskovalne infrastrukture za humanistiko v raziskavi Odlivanje smrti,” in Alenka Pirman, ed., Odlivanje smrti: posmrtne maske v slovenskih javnih zbirkah (Ljubljana: Institute of Contemporary History, 2023), accessed on 26 February 2025, https://sistory.github.io/Odlivanje_smrti/ch01.html.

Authorship

As the final part of our comprehensive analysis of the content and characteristics of the SIstory portal, we examined authorship trends, focusing on frequent author pairs (Figure 4) and co-authorship networks (Figure 5).

Figure 4: Bubble plot of the 50 most frequent author pairs. The size and darker colours of the bubbles indicate higher counts of co-occurrences for specific pairs. Source: Own work

Figure 4 shows the co-occurrence of author pairs based on the “Author” metadata field, which includes only individual publication authors. Organisational, editorial, and translation authorship were excluded from the analysis.

Of the original 6317 author pair combinations, 5511 occurred only once, representing 87.3% of all author pairings. The remaining 12.7% consist of combinations with multiple occurrences. This may suggest limited recurring collaborations and a diverse research network, as well as the fact that the production on SIstory includes a high number of unique collaborators. Supporting this idea is the very high percentage of publications with single authorship across SIstory’s entire production (94.5%) For this analysis, publications without an identified author (creator) have been excluded from the total publication count. These publications, primarily official gazettes, account for approximately 63% of the entire database., indicating a trend of authors favouring individual work over joint efforts.

Several authors have frequently collaborated on publications available through the portal, with specific pairs standing out for their joint work. The most frequent collaborators are Bojan Balkovec and Božo Repe, who co-authored 49 publications. They are followed by Božidar Flajšman and Božo Repe with 29 joint works, and Bojan Balkovec and Božidar Flajšman with 28. Although these author pairs exhibit high co-occurrence, manual checks reveal that most of their collaborations are concentrated in the Events section (Dogodki), specifically video recordings on historical themes produced by the Department of History, Faculty of Arts, University of Ljubljana – an institution to which all three authors are affiliated.

In contrast, author pairs with fewer than ten co-occurrences tend to be more varied and are more likely to produce traditional academic outputs such as scientific papers or monographs. For example, Gašper Šmid and Žarko Štrumbl co-authored eight scientific papers, mainly in the journal Arhivi, while Mitja Ferenc and Božo Repe collaborated on six chapters in a scientific monograph. However, this pattern is not universal; manual checks of author pairs within the 5–10 co-occurrence range revealed exceptions. One such case is that of Marta Rendla and Vida Deželak Barič, who co-authored eight works – all of which are periodic interim reports for a research project. Vida Deželak Barič et al., Vmesno poročilo o rezultatih opravljenega raziskovalnega dela na projektu v okviru ciljnega raziskovalnega programa (CRP) “Konkurenčnost Slovenije 2006–2013”: Vmesno poročilo velja za obdobje od 15. 11. 2010 do 15. 3. 2011: Pregled mrliških matičnih knjig za ugotovitev števila ter strukture žrtev druge svetovne vojne in neposredno po njej (Ljubljana: Institute of Contemporary History, 2011), https://hdl.handle.net/11686/1120.

Within the scope of the authorship analysis, we also examined the author network. Figure 5 shows the ten most connected authors, based on the number of direct co-authorship links for each author.

Figure 5: Network of the ten most connected authors Source: Own work

The co-authorship network, shown in Figure 5, highlights the ten most interconnected authors and their relationships, outlining key figures within the SIstory production. With a density metric of 0.0046, the network is relatively sparse, indicating that most authors are not directly connected. However, the average path length of 3.89 suggests that most authors can be linked within about four steps, implying that although direct connections are limited, authors are frequently linked through short co-authorship chains.

Many of the most connected authors, such as Božo Repe, Bojan Balkovec, and Božidar Flajšman, are already identified in Figure 4 as frequent collaborators from the same institution. However, additional authors, like Mojca Šorn and Vasilij Melik, although not part of recurring co-authorship pairs, are notable for their high connectivity within the network. Their position on the periphery of the top ten suggests they may not co-author often but remain important figures within the portal’s production.