GESIS Knowledge Graph

Maxine

Maxine is a PhD student in the sociology. She is searching for most recent surveys covering questions on migration and publications analysing these datasets. As a newbie, she is uncertain about which academic search portal to use for finding such inter-connected information. The GESIS KG contains links between social science research data and publications. This data is integrated in the GESIS Search portal. Maxine can use the GESIS Search to find the information she needs.

You can find an example SPARQL query here.

Will

Will is a senior researcher in information science. In his research, he is investigating citation behaviour and usage of data in different scientific disciplines across time. To analyse data citation and usage behaviour in the social science domain, Will can query the necessary data from the GESIS KG via its provided SPARQL endpoint. Alternatively, he can download the KG as a dump file to analyse it offline with other preferred tools. The provided documentation of the underlying schema of the GESIS KG helps him to understand the structure of the data.

You can find an example SPARQL query here.

Nancy

Nancy is a research data engineer and works in a research infrastructure organization. In her current project, she needs to integrate metadata from scientific resources from GESIS and other organizations into another search system. For doing so, she needs to access and harvest metadata from the GESIS KG via an OAI-PMH API. The OAI-PMH API of the GESIS KG will allow Nancy to harvest the metadata from GESIS she needs in a standardized format. The webpage of the API will also provide documentation for Nancy on how to use the API and how to harvest the data.

Resources and entities

The GESIS Knowledge Graph comprises several types of scientific resources, entities and the direct relationships without provenance between them. The latest version of the GESIS Knowledge Graph ontology can be found in the Download section.

The figure above illustrates the four main scientific resources which are currently entailed in the GESIS KG. We distinguish between scientific resources and entities. The main resources and entities of the GESIS Knowledge Graph are listed and described below:

Relationships / links between resources

This section describes the relationships between scientific resources in the GESIS Knowledge Graph and how they are represented in the data model.

The figure above depicts indirect relationships with provenance. In the GESIS KG Ontology, we provide both direct relationships without provenance and indirect relationships with provenance—the former for easy querying, and the latter for capturing additional metadata about how the links were created. Since the GESIS KG represents many-to-many (m:n) links between scientific resources and includes specific data about these links, additional classes for references and link metadata are incorporated into the data model.

Links between resources in the GESIS KG are either manually curated or automatically generated. Manual links are created by GESIS staff or are derived from research data bibliographies curated for specific research data programs. These manually linked resources are clearly marked as such and typically point to unique research datasets. The curation of these links is handled by the Manual Link Curation Pipeline. For automatically generated links between publications and research data, GESIS has developed the Dataset Citation Detection Pipeline. This pipeline, an extension of the InfoLink tool, is designed to detect and disambiguate dataset citations. It identifies mentions of research data within full texts and automatically links them to the corresponding datasets. These links are marked as automatic and may not always refer to a single, unique dataset, but rather to potential datasets used in the publication. It is important to note that the automatically generated links have not yet been evaluated by domain experts—this is planned as future work. Another pipeline, the Variable Detection Pipeline, automatically identifies links between publications and variables. This was developed as part of the DFG-funded VADIS project. Additionally, the Publication Citation Detection Pipeline identifies citation links between publications and their referenced works automatically, developed under the DFG-funded Outcite project.

Provenance

In the following table, it is described how provenance information is reflected in the GESIS KG. In different properties, it is captured from which data source within GESIS Search a particular resource is originating, from where a mentioned dataset is originating, and from which link detection pipeline or manual effort a link is originating as well as versioning information.

Persistent identifiers

Resources and entities in the GESIS Knowledge Graph hold several identifiers. While this includes persistent identifiers like DOIs assigned by PID authorities, there are also identifiers assigned to resources by the authority of the data source. Thirdly, a dereferenceable URI within the namespace of the GESIS Knowledge Graph has been assigned to every resource and entity which is part of the graph. The table below gives an overview of all identifiers which are present in the GESIS Knowledge Graph.

The GESIS KG uses the same IDs for its resources like the IDs in URLs used by the GESIS Search, i.e. URIs of the GESIS KG can be easily constructed if the URL, resp. the ID of a scientific resource in the GESIS Search is known.

Examples
GESIS Search URL of a resource: https://search.gesis.org/research_data/ZA5280
GESIS KG URI of the same resource: https://data.gesis.org/gesiskg/resource/ZA5280

Dataset statistics

APIs

OAI-PMH API

We provide an OAI-PMH API, available at https://data.gesis.org/gesiskg/oai/, which allows you to access and harvest metadata provided by the GESIS KG in the DataCite and OpenAIRE format.

SPARQL endopint

You can explore the data within the GESIS Knowledge Graph using SPARQL queries at the following SPARQL endpoint: https://data.gesis.org/gesiskg/sparql
Below you can find some example SPARQL queries.

SELECT ?id ?title
WHERE {?id ?p <https://schema.org/ScholarlyArticle>.
       ?id <https://schema.org/name> ?title.
} 
LIMIT 10000

SELECT *
WHERE {<https://data.gesis.org/gesiskg/resource/ZA5282> ?p ?o}

SELECT ?publication ?publication_title ?dataset ?dataset_title
WHERE {?publication <https://schema.org/about> ?topic.  
       ?topic <https://schema.org/name> "Migration"@en.
       ?publication ?p <https://schema.org/ScholarlyArticle>.  
       ?publication <https://schema.org/citation> ?dataset.
       ?dataset ?p <https://schema.org/Dataset>.
       ?publication <https://schema.org/name> ?publication_title.
       ?dataset <https://schema.org/name> ?dataset_title.
}Limit 100

SELECT ?year (COUNT(?publication) AS ?count)
WHERE {?publication ?p <https://schema.org/ScholarlyArticle>.
       ?publication <https://schema.org/citation> ?dataset.
       ?dataset ?p <https://schema.org/Dataset>. 
       ?dataset <https://schema.org/about> ?topic.  
       ?topic <https://schema.org/name> "Migration"@en.
       ?publication <https://schema.org/datePublished> ?year .
} 
GROUP BY ?year
ORDER BY ?year

Download

You can download the current version of the GESIS Knowledge Graph as a full RDF dump (JSON-LD and Turtle) as well as its underlying ontology at: https://doi.org/10.7802/2878

Older version:
v0.1.0-beta: https://doi.org/10.5281/zenodo.14229945

GESIS Search

The GESIS Knowledge Graph is integrated in the GESIS Search. Links between scientific resources are included in the result list and detailed views of search results.

License

The GESIS Knowledge Graph is available for access, download, and reuse under a Creative Commons Attribution 4.0 license since the license of some input sources is CC-BY as well.
If you are using the GESIS Knowledge Graph or parts from it, please cite the GESIS KG as follows:

Biswas, Debanjali, & Zapilko, Benjamin (2025). GESIS Knowledge Graph (GESIS KG). GESIS, Köln. Datenfile Version 1.0.0, https://doi.org/10.7802/2878.

Versions and changelog