University of Mannheim researcher applying machine learning to process mining and temporal knowledge graphs
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No publication-count data is available yet, so this axis is unscored. It would measure aggregate research output — total papers, venue types and distribution over years — and needs a fuller bibliographic record linked to this author.
Impact—
Citation-based impact data is not available in the current snapshot, so no score is given. This axis would capture citation counts and influence on follow-up work, which cannot be computed from the two reviewed records alone.
Focus0/100
The focus score is 0, grounded in an even field distribution across the two catalogued papers: 50% process mining / machine learning and 50% machine learning / temporal knowledge graph forecasting. This split reflects two deliberate benchmark-building efforts in adjacent applied areas rather than scattered output; a larger publication record would be needed to judge whether it is a stable research identity or a small-sample artifact.
Collaboration36/100
The collaboration score is 36, computed from 5 distinct coauthors across the catalogued work — Lukas Kirchdorfer, Keyvan Amiri Elyasi, Julia Gastinger, Thilo Dieing and Christian Meilicke. Both papers are team efforts with Stuckenschmidt as last author, consistent with a senior supervisory role in small groups; the moderate score indicates a compact collaboration network rather than large consortium-style projects.
Activity—
Annual publication-rate data is not available, so activity is unscored. This axis would measure publishing cadence and sustained engagement over time; with only two 2026 papers on record, the sample is too thin for a meaningful score.
Reproducibility100/100
Reproducibility is the strongest signal in the record: both catalogued papers — 2 of 2 — reach Evolve's highest readiness, for a mean score of 100 (reproMean over reproPapers = 2). One paper releases a temporal knowledge graph benchmark from public setlist.fm and MusicBrainz sources; the other ships a broad multi-task benchmark for predictive process monitoring, both built around public data and comparable baselines.
Research summary
Heiner Stuckenschmidt is affiliated with the University of Mannheim and appears in the Evolve catalog with two 2026 arXiv preprints, on both of which he is the last author, with no corresponding-author role recorded. His catalogued work spans two adjacent areas of applied machine learning. The first, in process mining, tests multi-task learning for predictive process monitoring across a broad benchmark; Evolve's review finds the claim that it strongly improves next-activity prediction largely convincing, while gains on temporal tasks are statistically null. The second builds a temporal knowledge graph for music festival lineup forecasting from setlist.fm and MusicBrainz data; reviewers praise the thoughtful metrics but note that embedding models run out of memory on the full graph and gains over recurrence baselines stay modest. Reproducibility is the clearest strength: both papers (reproPapers = 2) carry the highest readiness rating, for a mean score of 100. Radar dimensions show moderate collaboration breadth (36, from 5 distinct coauthors) and a focus score of 0, reflecting an even 50/50 field split between process mining / machine learning and machine learning / temporal knowledge graph forecasting. Output, impact and activity metrics are not yet available and are left unscored.
Collaboration pattern
Both catalogued papers are multi-author works with Stuckenschmidt in the last-author position and no corresponding-author role recorded. Five distinct coauthors appear: Lukas Kirchdorfer, Keyvan Amiri Elyasi, Julia Gastinger, Thilo Dieing and Christian Meilicke. A collaboration score of 36 on the radar reflects this compact team pattern rather than a broad international network; coauthor affiliations and role details are not recorded in the available data.
Research keywords
Highlights
A real-world benchmark for music festival lineup forecasting
Multi-task learning put to a broad test in process monitoring
Full reproducibility marks across the catalog
Both catalogued papers reach Evolve's highest reproducibility readiness, giving a mean score of 100 across 2 papers — grounded in public data sources (setlist.fm, MusicBrainz) and openly benchmarked baselines.
Timeline
- 2026First catalogued appearance: two arXiv preprints reviewed by Evolve, both with Stuckenschmidt as last author.
- 2026Released a broad benchmark of multi-task learning for predictive process monitoring, covering next-activity prediction and related tasks.
- 2026Introduced a temporal knowledge graph benchmark for music festival lineup forecasting built on setlist.fm and MusicBrainz data.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Radar scores and bibliometrics come from a sparse snapshot; only collaboration, focus and reproducibility have underlying data, while output, impact and activity are unscored.
- The profile rests on only two 2026 arXiv preprints reviewed by Evolve; peer-reviewed venue publications, if any, are not reflected.
- Citation counts, h-index-style aggregates and annual publication rates are not available and were not estimated.
- Identity resolved from public bibliographic data (ORCID 0000-0002-0209-3859); affiliation and coauthor names come from the catalogued records only.