In this demonstration we showcase PathFinder, a unified approach to returning paths in graph database queries. Returning paths is a central feature of regular path queries in the new GQL graph query standard. In the demo we showcase how PathFinder works by establishing two public endpoints, which the attendees will be able to access using their browser to try diferent queries. The first endpoint will host a property graph version of Wikidata to test GQL-style path queries, while the second one contains Wikidata in RDF format and illustrates how SPARQL can be extended to return paths.
n7 : City name = Rome country = Italy
n6 : City counnamtrye == ANmetshteerrldaanmds e11 : car tolls = 73.81€
n1 : City counnamtrye == BGaeyrmreaunthy e1 : train e2 : train
n2 : City Counnamtrye == GNeürrmnbaenryg e4 : train e3 : train duration = 2h
e9 : flight e7 : train
n4 : City counnamtrye == FGrearnmkafunryt
n3 : City counnamtrye == SGteurtmtgaanryt e6 : train e5 : train
n8 : City Counnamtrye == SCahnilteiago e8 : flight code = AF406 duration = 13h
n5 : City name = Paris country = France
The second key component of GQL is the ability to return paths in query answers. Examining
the graph of Figure 1, we can immediately see that returning all the paths conforming to the
query in our example is not feasible. Namely, the loop generated by the edges e1 and e2 allows
us to generate an infinite number of paths. To prohibit infinite query answers, GQL uses path
modes [
MATCH (?x:City WHERE ?x.name='Bayreuth') -[?p ANY SHORTEST (:train | :flight)+]->
(?y: City WHERE ?y.name='Santiago') In this query the path itself is bound to the variable ?p and has non-deterministic semantics since there could be multiple shortest paths between the two nodes. For instance, in our example one answer is traced by the edges e1 → e3 → e5 → e8, another one by e1 → e4 → e7 → e9, etc. According to GQL, all these answers are valid, and there is no guarantee as to which one will be returned. Returning all of them can be done using the ALL SHORTEST mode.
Given the novelty of GQL [
Conference paper. We remark that this demo accompanies our paper PathFinder: Returning Paths in Graph Queries [16] which will be presented in the ISWC research track. Compared to the conference version, here we put a strong focus on usability; namely, we develop a graphical interface for returning paths and host two endpoints where attendees can test path queries. We also propose an extension of SPARQL where paths can be returned in the query results. Finally, we remark that an additional author is added compared with the Research Track paper to work on providing user interfaces and SPARQL support for the demo.
The demonstration will consist of the following use cases, all of which are supported through public query endpoints that will be available during the review process and later on. A public query endpoint. The main highlight of our demonstration will be a public query endpoint where the attendees will be able to test GQL path queries using only their Web browser. In particular, our endpoint will be hosting a property graph version of Wikidata [17] which we used in our scalability tests. In particular, we use a curated version of the data set based on the truthy dump of Wikidata [18], which was used in WDBench [19]. The dataset is publicly available at [20]. During the demonstration we will also have a brief explanation of the underlying data and will prepare a series of instructive queries for the users to get acquainted with the language. The Wikidata endpoint is available for the review purposes at https://mdb.imfd.cl/path_finder/, with a series of illustrative queries included on the endpoint. PathFinder and SPARQL engines. Since the PathFinder approach can also work as a part of a SPARQL engine, we first showcase its functionality in the context of evaluating property paths without returning the path themselves. This approach is already used in MillenniumDB’s Wikidata SPARQL endpoint at https://wikidata.imfd.cl/, which leverages PathFinder to handle property path queries, and can be tried there. We remark that in this case the entire Wikidata dump is from 2023–07–16 and not Wikidata Truthy.
Returning paths in SPARQL. In addition to checking reachability via property paths, we would also like to illustrate how PathFinder can be used to return paths in SPARQL queries, through a syntactic extension of the language. One such proposal can be accessed directly through the PathFinder console as explained in our online repository [14]. For the demo presentation we also developed a Web interface for displaying paths that witness SPARQL property path query answers in a similar fashion as done for our property graph endpoint. To test this functionality, the attendees can access our (extended) SPARQL endpoint hosting the Wikidata Truthy dataset at https://mdb.imfd.cl/path_finder/. Figure 2 shows how this extension of SPARQL syntax and semantics works on our endpoint.
Calisto, Farías, Rojas and Vrgoč were supported by ANID – Millennium Science Initiative Program – Code ICN17_002. Vrgoč was also supported by the ANID Fondecyt Regular project 1240346. Martens was supported by ANR project EQUUS ANR-19-CE48-0019; funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), project number 431183758. [13] W. Martens, M. Niewerth, T. Popp, C. Rojas, S. Vansummeren, D. Vrgoč, Representing paths in graph database pattern matching, Proc. VLDB Endow. 16 (2023) 1790–1803. [14] B. Farías, W. Martens, C. Rojas, D. Vrgoč, PathFinder: A unified approach for handling paths in graph query languages, 2024. URL: https://github.com/AnonCSR/PathFinder. [15] D. Vrgoč, C. Rojas, R. Angles, M. Arenas, D. Arroyuelo, C. Buil-Aranda, A. Hogan, G. Navarro, C. Riveros, J. Romero, Millenniumdb: An open-source graph database system, Data Intell. 5 (2023) 560–610. URL: https://doi.org/10.1162/dint_a_00229. [16] B. Farias, W. Martens, C. Rojas, D. Vrgoč, Evaluating regular path queries in GQL and
SQL/PGQ:, CoRR abs/2306.02194 (2023). URL: https://doi.org/10.48550/arXiv.2306.02194. [17] D. Vrandecic, M. Krötzsch, Wikidata: a free collaborative knowledgebase, Commun. ACM 57 (2014) 78–85. [18] T. W. Foundation, Wikidata:database download, 2021. URL: https://www.wikidata.org/ wiki/Wikidata:Database_download. [19] R. Angles, C. B. Aranda, A. Hogan, C. Rojas, D. Vrgoč, Wdbench: A wikidata graph query benchmark, in: The Semantic Web - ISWC 2022, 2022. URL: https://doi.org/10.1007/ 978-3-031-19433-7_41. doi:10.1007/978-3-031-19433-7\_41. [20] R. Angles, C. B. Aranda, A. Hogan, C. Rojas, D. Vrgoč, WDBench Dataset Download, 2022. doi:10.6084/m9.figshare.19599589.