QLever Wikipedia article

[WolfgangFahl]

Just added QLever to https://en.wikipedia.org/wiki/List_of_SPARQL_implementations

but unfortunately there is no proper article:

https://en.wikipedia.org/w/index.php?title=QLever&redirect=no

I did some brainstorming with Claude AI (see below) - this will not suffice yet and needs some love since Wikipedia will not accept AI generated articles

{{Infobox software
| name               = QLever
| logo               = 
| screenshot         = 
| caption            = 
| developer          = University of Freiburg
| released           = {{Start date|2017}}
| latest release version = 
| latest release date    = 
| programming language   = C++
| operating system   = Cross-platform
| platform           = 
| size               = 
| language           = 
| genre              = SPARQL query engine
| license            = {{AGPL}}
| website            = {{URL|https://qlever.cs.uni-freiburg.de/}}
}}

'''QLever''' (pronounced "Clever") is an open-source [[SPARQL]] query engine developed by the University of Freiburg. It is designed for efficient indexing and querying of very large [[knowledge graph]]s, capable of handling over 100 billion triples on a single standard PC or server.<ref name="readme">{{cite web |url=https://github.com/ad-freiburg/qlever/blob/master/README.md |title=QLever README |website=GitHub |access-date=YYYY-MM-DD}}</ref>

==Overview==

QLever is particularly notable for its speed in handling queries that involve large intermediate or final results, which are traditionally challenging for engines like Blazegraph or Virtuoso. It supports both SPARQL queries and text search within the associated knowledge base.<ref name="cikm2017">{{cite conference
  | last1 = Bast
  | first1 = Hannah
  | last2 = Buchhold
  | first2 = Björn
  | title = QLever: A Query Engine for Efficient SPARQL+Text Search
  | booktitle = CIKM '17: Proceedings of the 2017 ACM on Conference on Information and Knowledge Management
  | pages = 647–656
  | year = 2017
  | month = November
  | doi = 10.1145/3132847.3132921
  | url = https://doi.org/10.1145/3132847.3132921
  | publisher = ACM
  | access-date = YYYY-MM-DD
}}</ref> The engine also features context-sensitive autocompletion for SPARQL queries, enhancing ease of use.<ref name="cikm2022">{{cite conference |last1=Bast |first1=Hannah |last2=Korfhage |first2=Kenji |title=QLever Autocompletion: Fast, Context-Sensitive Query Autocompletion for SPARQL |booktitle=Proceedings of the 31st ACM International Conference on Information & Knowledge Management |pages=3635–3639 |year=2022 |doi=10.1145/3511808.3557335}}</ref>

==Features==

* Efficient indexing and querying of very large knowledge graphs (over 100 billion triples)
* High performance on queries with large intermediate or final results
* Integration of text search with SPARQL queries<ref name="cikm2017" />
* Context-sensitive autocompletion for SPARQL queries<ref name="cikm2022" />
* Support for multiple large-scale knowledge graphs, including Wikidata, Wikimedia Commons, OpenStreetMap, UniProt, PubChem, and DBLP
* Aims for full SPARQL 1.1 support, including Federated Query (SERVICE) and Update functionalities<ref name="readme" />

==Technology==

QLever is implemented in C++. It uses advanced indexing and query optimization techniques to achieve efficient querying of large-scale knowledge graphs. The engine can be deployed on a single standard PC or server, making it accessible for various research and application scenarios.<ref name="readme" />

==Applications==

QLever provides demos for various large knowledge graphs, regularly updated:

* Complete Wikidata
* Wikimedia Commons
* OpenStreetMap
* UniProt
* PubChem
* DBLP

These demonstrations showcase QLever's capabilities in handling diverse and extensive datasets.<ref name="readme" />

==Development and Deployment==

QLever can be used via the `qlever` script, which controls all aspects of QLever's operation. The script uses a configuration file called `Qleverfile`, and comes with several example configurations. QLever can be run using Docker, which automatically pulls the most recent image.<ref name="readme" />

==Research==

QLever has been the subject of several research papers, notably:

* A 2017 paper presented at CIKM detailing the combination of SPARQL and text search capabilities.<ref name="cikm2017" />
* A 2022 paper, also presented at CIKM, focusing on QLever's context-sensitive autocompletion functionality.<ref name="cikm2022" />

==See also==

* [[SPARQL]]
* [[Resource Description Framework]]
* [[Triplestore]]
* [[Semantic Web]]
* [[Wikidata]]

==References==
{{Reflist}}

==External links==

* {{Official website|https://qlever.cs.uni-freiburg.de/}}
* [https://github.com/ad-freiburg/qlever QLever GitHub Repository]

[[Category:Semantic Web]]
[[Category:Query engines]]
[[Category:Free software programmed in C++]]
[[Category:Cross-platform free software]]

[hannahbast]

@WolfgangFahl Thanks for the initiative. What do you propose, that we write a first draft of the Wikipedia article on QLever?

[tuukka]

I think that ideally the community would write the article and no project member was involved, in order to stay clearly within English Wikipedia's conflict of interest guidelines and to avoid any extra reviews triggered by ghostwriting.

• https://en.wikipedia.org/wiki/Wikipedia:Plain_and_simple_conflict_of_interest_guide
• https://en.wikipedia.org/wiki/Wikipedia:Ghostwriting

[WolfgangFahl]

I think for a stub we are fine to initiate it and then ask the community to elaborate. Unfortunately it is not easy these days to get a wikimedia stub article out there without a request for speedy deletion. On the other hands our friends from scholia/wikicite might also be interested in spreading the word.

[tuukka]

To get started, I tried to find some third-party sources and found a good one with some familiar names: 😁

• Getting and hosting your own copy of Wikidata.
  Wolfgang Fahl, Tim Holzheim, Andrea Westerinen, Christoph Lange and Stefan Decker. pdf

[hannahbast]

@WolfgangFahl @tuukka I think that paper is outdated since a lot of development has happened since then. For a more up-to-date performance evaluation and comparison with other engines, see https://github.com/ad-freiburg/qlever/wiki/QLever-performance-evaluation-and-comparison-to-other-SPARQL-engines .

[1ec5]

I just noticed this discussion – by coincidence, I started an article about QLever back in July.

[hannahbast]

@1ec5 Thank you for starting the article. I found it in November and wrote a revision, which unfortunately was rejected by one of the Wikipedia editors (despite the changes being factual improvements). You find the change I proposed in the article history. Can you please consider them? In summary, there are three issues:

1. Some language inaccuracies, which I corrected
2. The paper with the performance evaluation used a very old version of QLever, which doesn't have much to do with the current version anymore. I don't mind mentioning the evaluation, but that should be stressed (which I did in my revision)
3. There are several more recent performance evaluations in the meantime, see my revision
4. QLever implements basic GeoSPARQL functionalities (see https://github.com/ad-freiburg/qlever/wiki/GeoSPARQL-support-in-QLever) + the triples you mentioned

[WolfgangFahl]

Just had a look at https://en.wikipedia.org/w/index.php?title=QLever&diff=1256355130&oldid=1234410767.

It was IMHO clearly rejected based on Hannah's authorship which is against Wikimedia rules.

QLever powers the official SPARQL endpoint of [[DBLP]].<ref>{{cite web|title=dblp SPARQL query service|publisher=Schloss Dagstuhl |accessdate=2024-11-09 |url=https://blog.dblp.org/2024/09/09/introducing-our-public-sparql-query-service/}}</ref>

was readded by me in "no-ghost-writing-mode" - this is a statement that is IMHO clearly factual enough to not be disputed or challenged.

Some of the other parts of the diff are IMHO more problematic so i propose to continue in this stepwise fashion. Also I am happy that there is now a proper article. That is very helpful!

[WolfgangFahl]

Claude-AI Übersetzung ins Deutsche:

{{Infobox Software
| title = QLever
| author = [[Hannah Bast]], Björn Buchhold, et al.{{sfn|Bast|Buchhold|2017}}
| released = {{Start date and age|2017}}
| repo = {{URL|https://github.com/ad-freiburg/qlever}}
| qid = Q111016295
| programming language = [[C++]]
| standard = [[SPARQL]]
| language = [[Englische Sprache|Englisch]]
| genre = [[Graphdatenbank]]
| license = [[Apache-Lizenz]]
| website = {{URL|https://qlever.cs.uni-freiburg.de/}}
}}

'''QLever''' (ausgesprochen {{IPAc-en|ˈ|k|l|ɛ|v|ɚ}} {{respell|KLEH|ver}}, wie in "clever") ist ein [[Open-Source-Software|quelloffener]] [[Triplestore]] und eine [[Graphdatenbank]], die von einem Team an der [[Albert-Ludwigs-Universität Freiburg]] unter der Leitung von [[Hannah Bast]] entwickelt wurde. QLever führt hochperformante Abfragen von [[Semantisches Web|semantischen Web]]-[[Wissensbasis|Wissensbasen]] durch, einschließlich [[Volltextsuche]] innerhalb von Textkorpora.{{sfn|Bast|Buchhold|2017}} Eine spezialisierte Benutzeroberfläche für QLever vervollständigt [[SPARQL]]-Abfragen prädiktiv.{{sfn|Bast|Kalmbach|Klumpp|Kramer|2021}}

QLever ist einer der Kandidaten für die Ablösung von [[Blazegraph]] als Triplestore für den [[Wikidata Query Service]]. Eine Studie aus dem Jahr 2023 ergab, dass QLever im Vergleich zu anderen Triplestores eine schnelle Ausführung erfolgreicher Abfragen erreichte, jedoch nur eingeschränkte Unterstützung für komplexe SPARQL-Konstrukte bot.<ref>{{cite conference|title=Evaluation of a Representative Selection of SPARQL Query Engines Using Wikidata|first1=An Ngoc|last1=Lam|first2=Brian|last2=Elvesæter|first3=Francisco|last3=Martin-Recuerda|work=[[Extended Semantic Web Conference]]|series=Lecture Notes in Computer Science|publisher=Springer, Cham|volume=13870|date=2023|url=https://2023.eswc-conferences.org/wp-content/uploads/2023/05/paper_Lam_2023_Evaluation.pdf|doi=10.1007/978-3-031-33455-9_40}}</ref>

== Inhalte ==
Der offizielle QLever-Server bietet Zugriff auf folgende Datensätze:<ref>{{cite web|title=QLever|publisher=Lehrstuhl für Algorithmen und Datenstrukturen der Universität Freiburg|location=Freiburg im Breisgau|accessdate=13. Juli 2024|url=https://qlever.cs.uni-freiburg.de/}}</ref>

* [[Wikidata]]
* [[Wikimedia Commons]]
* [[Freebase (Datenbank)|Freebase]]
* [[OpenStreetMap]]
* [[OpenHistoricalMap]]
* [[UniProt]]
* [[PubChem]]
* [[DBLP]]
* [[OpenCitations]]
* [[IMDb]]
* [[Gemeinsame Normdatei]]
* [[YAGO (Datenbank)|YAGO]]
* [[DBpedia]]
* Wallscope Olympics Datenbank

QLever betreibt den offiziellen SPARQL-Endpunkt von [[DBLP]].<ref>{{cite web|title=dblp SPARQL query service|publisher=Schloss Dagstuhl |accessdate=2024-11-09 |url=https://blog.dblp.org/2024/09/09/introducing-our-public-sparql-query-service/}}</ref>

Obwohl die QLever-Engine keine native Unterstützung für [[GeoSPARQL]]-Funktionen bietet, berechnet der offizielle Server eine Teilmenge von GeoSPARQL-Beziehungen in dedizierte Tripel für OpenStreetMap- und OpenHistoricalMap-Daten vor.{{sfn|Bast|Brosi|Kalmbach|Lehmann|2021}}

== Siehe auch ==
{{Commons category|QLever}}
* [[Liste von SPARQL-Implementierungen]]

== Einzelnachweise ==
{{reflist}}

== Weiterführende Literatur ==
* {{cite conference|title=An Efficient RDF Converter and SPARQL Endpoint for the Complete OpenStreetMap Data|first1=Hannah|last1=Bast|author1-link=Hannah Bast|first2=Patrick|last2=Brosi|first3=Johannes|last3=Kalmbach|first4=Axel|last4=Lehmann|conference=SIGSPATIAL|date=2.–5. November 2021|publisher=[[Association for Computing Machinery]]|location=Peking|isbn=978-1-4503-8664-7|doi=10.1145/3474717.3484256|url=https://ad-publications.cs.uni-freiburg.de/SIGSPATIAL_osm2rdf_BBKL_2021.pdf}}
* {{cite conference|title=QLever: A Query Engine for Efficient SPARQL+Text Search|first1=Hannah|last1=Bast|first2=Björn|last2=Buchhold|work=[[Conference on Information and Knowledge Management]]|date=6.–10. November 2017|location=Singapur|publisher=[[Association for Computing Machinery]]|isbn=978-1-4503-4918-5|doi=10.1145/3132847.3132921|url=https://ad-publications.cs.uni-freiburg.de/CIKM_qlever_BB_2017.pdf}}
* {{cite arXiv|title=Efficient SPARQL Autocompletion via SPARQL|first1=Hannah|last1=Bast|first2=Johannes|last2=Kalmbach|first3=Theresa|last3=Klumpp|first4=Florian|last4=Kramer|first5=Niklas|last5=Schnelle|date=29. April 2021|eprint=2104.14595v1|class=cs.DB}}

== Weblinks ==
* {{Official website|https://qlever.cs.uni-freiburg.de/}}
* {{GitHub|ad-freiburg/qlever}}
* {{GitHub|ad-freiburg/qlever-ui}}

[[Kategorie:Triplestore]]
[[Kategorie:Graphdatenbanken]]
[[Kategorie:Albert-Ludwigs-Universität Freiburg]]

[hannahbast]

@WolfgangFahl Can you please explain which parts of my the diff I proposed you find problematic? The two most problematic aspects of the current version are:

1. The statement about the performance: Only a completely outdated paper is mentioned, which has no relevance anymore. At the same time, several more recent evaluations, including the ongoing one by the Wikidata folks themselves, are not mentioned.
2. The statement about the GeoSPARQL support. The GeoSPARQL support is basic but existing and there is a dedicated page on the QLever Wiki on this, which should be linked to

[WolfgangFahl]

Can you please explain which parts of my the diff I proposed you find problematic?

It's not my opinion that the parts are problematic - i just figure that the rules of the Wikipedia community might lead to this conclusion depending on who look on the article. I have had lots of my changes being rejected in Wikipedia since 2008. IMHO only in the first few years was it really straightforward to get content into Wikipedia.

1. At ESWC 2023 i personally was there when the results were presented and i actively informed the main author about the difference between the finding and the state of affairs at presentation time. It was pitty that the SINTEF team did not actively seek communication. I have talked to them a few month ago about snapquery and they showed some interest in this newer research. Unfortunately the snapquery paper has been rejected twice so far - the test suite for QLever is still a major use case from my point of view.
   I have tried to change the section to:

A 2023 study compared QLever with [[Virtuoso Universal Server|Virtuoso]], [[Blazegraph]], GraphDB, Stardog, [[Apache Jena]], and Oxigraph. The QLever version investigated in the study achieved 

2. GeoSPARQL is interesting for the https://github.com/WolfgangFahl/velorail project - so i will happily look into that detail.

[WolfgangFahl]

https://www.wikidata.org/wiki/Wikidata:Scaling_Wikidata/Benchmarking

[hannahbast]

@WolfgangFahl I just stumbled on https://en.wikipedia.org/wiki/QLever again and it still says:

A 2023 study compared QLever ... The QLever version investigated in the study achieved fast execution of successful queries but offered limited support for complex SPARQL constructs

The study actually names the exact commit hash as 742213facfcc80af11dade9a971fa6b09770f9ca, and it is from 04.11.2021, so really really old.

I would suggest to clarify that by writing

A 2023 study compared QLever ... The QLever version investigated in the study (from 04.11.2021) achieved fast execution of successful queries but offered limited support for complex SPARQL constructs

Otherwise the statement is misleading: everyone who reads the current passage will think that it's the QLever version from 2023.

And why not mention https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/WDQS_backend_update/WDQS_backend_alternatives, which lists a variety of more recent benchmarks?

[1ec5]

The study actually names the exact commit hash as 742213facfcc80af11dade9a971fa6b09770f9ca, and it is from 04.11.2021, so really really old.

Clarified in https://en.wikipedia.org/wiki/Special:Diff/1281830749

And why not mention https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/WDQS_backend_update/WDQS_backend_alternatives, which lists a variety of more recent benchmarks?

It might be acceptable under Wikipedia’s external link policy, but I’m not sure.

[hannahbast]

@1ec5 @WolfgangFahl @Daniel-Mietchen I just stumbled upon https://en.wikipedia.org/wiki/QLever again. It still mentions the 2023 study, which evaluates a 2021 version of QLever. This is incredibly outdated by now and IMO should simply be dropped, in particular, in view of several more recent evaluations, including the following two, which are both very thorough:

1. There is the benchmark carried out by Peter Patel-Schneider and Sam Klein, which was supported by a grant from WIkimedia CH. This is very thorough work
2. There is a new ISWC'25 paper with a very thorough performance comparison of eight SPARQL engines on a broad selection of queries covering the whole breadth of the SPARQL standard. See https://iswc2025.semanticweb.org/#/program/acceptedpapers (ISWC'25 website), https://ad-publications.cs.uni-freiburg.de/ISWC_sparqloscope_BKTU_2025.pdf (PDF of the publication), and https://qlever.dev/evaluation (detailed results)

[1ec5]

I renamed the section in question to “History” to clarify that it doesn’t necessarily reflect the current state of affairs. For now, the section is probably necessary to establish QLever’s notability. Otherwise the article would risk getting deleted on the grounds that QLever hasn’t been given substantial coverage in enough secondary sources. Wikipedia policy unfortunately doesn’t allow citing “primary” sources. Sometimes editors can get away with citing primary sources if they’re peer-reviewed, but a citation of the ISWC ’25 paper contradicting a secondary source would probably be scrutinized because of the authors’ affiliation with QLever. The benchmark by Patel-Schneider and Klein sounds promising. Any idea where that’s been published? All I see is this grant proposal.