# FOSSGIS "SPARQL Unicorn"

CfC: https://pretalx.com/fossgis2024/cfp

## Title (zwischen 8 und 85 Zeichen)

The SPARQL Unicorn Ontology documentation: Exposing RDF geodata using static GeoAPIs

## Type

* Vortrag (20min) - Academic Track

## Track

* Open Data

## Author(s)

* Timo Homburg (HS Mainz)
* Florian Thiery (Research Squirrel Engineers Network)

## Zusammenfassung (100-500 Zeichen)

We introduce the ontology documentation feature of the SPARQLing Unicorn QGIS plugin, allowing the conversion of RDF data dumps to static HTML deployments with static versions of well-known APIs such as OGC API Features. Contents of RDF data are deployed interoperably accessible for many research communities, e.g. directly in QGIS. Our talk shows the tool's motivation, conversion process with test datasets, discusses limitations, standardization issues and future developments of this approach in general.

## Abstract (ca. 3000 Zeichen)

A pertaining problem for linked open (geo-)datasets hosted on the web is the difficulty of their accessibility for people who do not have a previous knowledge about the datastructures and the software to process [RDF](https://www.w3.org/TR/rdf11-concepts/).
Often, researchers are only interested in the question if a dataset is suitable for their needs, which warrants an easy discovery of linked open data content with common tools used by the research community they are familiar with.
For example: A GIS researcher expects data to be served using e.g. [OGC API Features APIs](http://features.developer.ogc.org/) in formats such as [GeoJSON](https://datatracker.ietf.org/doc/rfc7946/) or [GML](https://www.ogc.org/standard/gml/), to be processed in tools such as [QGIS](https://www.qgis.org) to allow for an easy accessibility. The GIS researcher is not necessarily proficient in the [SPARQL query language](https://www.w3.org/TR/sparql11-query/) and in RDF-to- Geodata-conversion.
This is currently not the case for RDF data dumps, e.g. hosted on [Zenodo](https://zenodo.org) or [Github](https://github.com), since special kinds of data need to be manually extracted and converted to be of use to other communities first.

To mitigate this gap in data provision, we present the newly-developed SPARQL Unicorn ontology documentation feature, available as [QGIS plugin](10.5281/zenodo.3786814), a [standalone script](https://github.com/sparqlunicorn/sparqlunicornGoesGIS-ontdoc) and as a [Github Action](https://github.com/sparqlunicorn/sparqlunicornGoesGIS-ontdoc), which allows to create readily-hostable data dumps from RDF data, which can be served using static webspaces such as Github Pages.
RDF resources are represented by one HTML and RDF resource each, which are identified by their URI and for dereferencability, ideally hosted at this particular place on the web.
In addition to an HTML navigation and search function and a variety of graph analysis formats (e.g. [GraphML](http://graphml.graphdrawing.org), [GEXF](https://gexf.net)), the script is able to generate static API versions e.g. OGC API Features for geospatial data or the IIIF Presentation API for accessing media data.
This allows to access geodata encoded in RDF as pre-generated FeatureCollections in e.g. QGIS therefore making geospatial data in RDF directly accessible to the GIS community.

In our talk we describe the publication process of several semantically enriched and publicly available geospatial datasets in RDF and how they can be reused in QGIS as FeatureCollections and as RDF using the SPARQLing Unicorn QGIS plugin.
The [SPP Dataset](https://github.com/archaeolink/SPP1630Harbours-RDF), the [AncientPorts Dataset](https://github.com/archaeolink/AncientPorts_RDF) and the [CIGS Datatset](https://github.com/archaeolink/CIGS_RDF) highlight the representation of (spatio-)temporal research data referenced by scientific literature and serve as examples for the OGC API features integration.
In addition, we highlight the publication of [IIIF](https://iiif.io/api/presentation/3.0/) content (e.g. as complementary data to geodata) using the example of [ARS-LOD](https://github.com/RGZM/ars-lod/).

Finally, we discuss further extensions of the plugin for other statically generated content, limitations of the current approach and highlight the importance of this way of publishing for the general public and in a research software engineering setup.

--> Noch einzufügen: Bilder wenn wir dürfen, Verlinkungen, Beschreibung der Application Cases die wir im Talk vorstellen. Dann haben wir auch die 3000 Zeichen zusammen
==> Bilder nur als "Ressourcen"

## Beschreibung (mind. 200 Zeichen)

### Introduction
In recent years, many academic disciplines, but also companies in the industry, rely on knowledge graphs to represent, among others, geospatial data in RDF.
RDF data can be hosted as single files or in specialized databases called triple stores, which use a REST API, typically a SPARQL endpoint, to provide access to said geospatial data. It is then up to individual applications to display this geospatial data in web frontends, e.g. using Leaflet or use other means to visualize the contents.

Triple stores are a good solution for institutions which host their own linked data databases. However, researchers without such means will commonly host only RDF datadumps in repositories such as Zenodo and Github.
These datadumps contain useful data, but are often not easily accessible to a wider audience beyond the LOD community, as the discovery, modification and processing of these data dumps usually requires knowledge in programming and LOD software.
This is besides the fact that the linked open data principles advocate for URIs in the knowledge graph to be resolvable in different serializations and the geospatial community requiring access to geodata using their own native APIs, such as OGC API Features.

### Related Work
Recommendations to publish RDF data as linked open data exist and many data portals also adopt way to expose metadata about geospatial datasets as linked open data.

### SPARQLing Unicorn QGIS Plugin

### SPARQL Unicorn Ontology documentation

To overcome this interoperability gap, we present the newly developed RDF documentation feature of the SPARQLing Unicorn QGIS Plugin, a QGIS plugin for scientists and citizen scientists to enrich geodata with semantic web data and to document these RDF datasets in static HTML.
RDF data can be converted using the [QGIS plugin](10.5281/zenodo.3786814), a [standalone script](https://github.com/sparqlunicorn/sparqlunicornGoesGIS-ontdoc) available on Github and as a [Github Action](https://github.com/sparqlunicorn/sparqlunicornGoesGIS-ontdoc) and exposes a number of static APIs, which allows the seamless integration of the published linked open data in various applications such as QGIS.
In the following we described the kind of data which can be generated using the documentation feature.

#### HTML and further data exports
The RDF documentation converts RDF data to LOD serializations in HTML and other graph and relational data formats, to locally preview and to host these data on static webspaces, such as Github Pages.
Graph data formats include GraphML and GEXF for graph analysis tools such as Gephi and CSV, as well as TSV to be used in a variety of

#### Static APIs

A static API is the representation of the directory and file structure of an existing API as a collection of pre-generated (JSON) files, which mimic typical answers of the API they represent.

The script supports three kinds of static APIs, which are depicted as JSON files in the

Our talk highlights the following three datasets, which are publicly available on Github:

* SPP Dataset: A dataset of ancient Harbours of Europe which are referenced in space and time, as well as, further metadata and references to the publications in which they are described.

The SPP Dataset is comprised of geodata which have been described using the GeoSPARQL vocabulary, one of the many vocabularies the script recognizes as geodata in RDF.
The HTML result contains of one HTML page per geospatial feature, the same representation in TTL and a GeoJSON representation of each feature.
In addition, this dataset is a showcase of a static version of the OGC API Features API in stat

#### HTML templates

* SPARQLing Unicorn QGIS Plugin (10.5281/zenodo.3786814)
* interdisciplinary Research Software for scientists and Citizen Scientists to integrate Linked Open Data in Open Source GIS Software (QGIS), enrich data with data from the semantic web, transform data into LOD and create documentation.
* SPARQL Unicorn Ontology Documentation (10.5281/zenodo.8190763)
* docu auch als [GitHub action](https://github.com/sparqlunicorn/sparqlunicornGoesGIS-ontdoc), [Beispiel Limes](https://github.com/sparqlunicorn/sparqlunicornGoesGIS_testdata)