Interactive data visualization companion for the paper "Does SPARQL federation work in the real world? A case study over large biological SPARQL endpoints", accepted to and scheduled for presentation at the October 2026 ISWC conference in Bari, Italy.
This site presents the experiment-level outcomes of federated query executions. For general interpretation, start with Overview Charts and Monthly Views for global trends, then dig into the Data Explorer to inspect individual queries or experiments in more detail.
Use these filters to narrow the visible data. Defaults show the full main experiment scope.
These charts summarize the full filtered dataset to show overall reliability, dominant failure modes, and baseline runtime behavior across experiment runs.
This section highlights temporal changes in outcomes and workload so you can compare how experiment behavior evolved across testing periods.
This section summarizes structural characteristics of the federated query set using the SIB query-analysis statistics dataset. It mirrors the paper tables with descriptive summaries, bucketed feature distributions, and per-query feature listings.
| Feature | Average | Maximum | Minimum | Standard Deviation |
|---|
| Query | # Triple Patterns | # OGPs | # UGPs | # UGPs w/ Multiple TPs | # Federation Members |
|---|
Endpoint inventory used by the study workload and its federated targets. This table is reproduced from
endpoints_table.tex without altering the underlying endpoint data.
Endpoints listed above the separator are targets in at least one query
(the endpoint that receives the full query WITH SERVICE descriptions). Endpoints listed
below the separator are federatesWith sources that are only invoked through
remote SERVICE calls.
Table caption: SPARQL endpoints represented in the real-world query workload. Triple-store was assessed through endpoint literature, server response, and endpoint service description data.
| Endpoint | Description | Triple-store | Endpoint URL |
|---|---|---|---|
| UniProt | Protein sequence and functional annotations | Virtuoso (past) / QLever (current) | sparql.uniprot.org/sparql |
| Rhea | Biochemical and transport reactions | Virtuoso | sparql.rhea-db.org/sparql |
| SwissLipids | Lipid structures | unknown | beta.sparql.swisslipids.org/sparql |
| Bgee | Gene expression for comparative analysis | Virtuoso | bgee.org/sparql |
| OrthoDB | Orthologous genes | Virtuoso | sparql.orthodb.org/sparql |
| OMA | Orthology and protein families | Virtuoso | sparql.omabrowser.org/sparql |
| biosoda | SPARQL federation server | N/A | biosoda.unil.ch/emi/sparql |
| Wikidata | General-purpose knowledge graph | Blazegraph | query.wikidata.org/sparql |
| Wikidata--QLever | Alternative Wikidata endpoint | QLever | qlever.cs.uni-freiburg.de/api/wikidata |
| METRIN-KG | Plant metabolome and traits | QLever | kg.earthmetabolome.org/metrin/api |
| IDSM | Small molecule database | PostgreSQL | idsm.elixir-czech.cz/sparql |
| MBGD | Microbial genomes | Virtuoso | sparql.nibb.ac.jp/sparql |
| STRING-DB | Functional protein association networks | Virtuoso | string-db.org/sparql |
| Allie | Abbreviation and long-form searches | Virtuoso | data.allie.dbcls.jp/sparql |
| EPO | Linked patent data | unknown | data.epo.org/linked-data/query |
| EUNIS | Biodiversity and species data | unknown | semantic.eea.europa.eu/sparql |
| data.europa.eu | European open data | Virtuoso | data.europa.eu/sparql |
| GlyConnect | Glycoproteomics data | GraphDB | glyconnect.expasy.org/sparql |
| MeSH | Medical Subject Headings (MeSH) | unknown | id.nlm.nih.gov/mesh/sparql |
| Bioregistry | Registry of biomedical identifiers | RDFLib endpoint | bioregistry.io/sparql |
| Identifiers.org | Persistent identifier registry | unknown | sparql.api.identifiers.org/sparql |
Use this area to inspect individual queries or selected experiment groups, with linked tables and charts for detailed, evidence-level comparisons.
| Run | Start | Duration (s) ⓘ | Outcome | Results ⓘ | HTTP Requests ⓘ | Error |
|---|
Select a query to inspect its text and metadata.
This section shows curated field notes documented in
Experiment Outcomes (notes).txt.
These notes record interpretation constraints and data handling choices so readers can understand assumptions behind the displayed results.
old-results data is kept in-repo for provenance and is excluded from all displays._ns and no-service query identifiers are treated as "no service descriptions".EX1 -> NOMETA-ASK,
EX2 -> NOMETA-COUNT, EX3 -> VOID-TRIPLE,
EX4 -> VOID-BLOCK, and matching -NRL variants.