01
Systems by family and era
Each dot is a seed at its publication year (OpenAlex; for IEEE proceedings papers the year in the DOI, which OpenAlex often gets wrong; for StarPU 2009, its Euro-Par 2009 version); its area grows with the works that cite it. Tap a family to hide or show it.
Swipe the figure sideways to see all years.
02
Attention per year and attention within the field
Citing works per year since publication against the share of citing works that also cite another seed (the core). Left: cited mainly beyond AMT; right: cited mainly by a community that knows several seeds.
Swipe the chart sideways to see it in full.
03
Citing works over time
Works per publication year that cite the seeds (2026 is incomplete; shaded bands mark the eras). Pick seeds to compare their curves.
Swipe the chart sideways to see all years.
Per family
Works per year that cite at least one seed of a family.
04
Attention lags behind invention
For each seed, its publication year (open dot) and the median publication year of the works that cite it (filled dot): half of its citing works appeared by then. Seeds in order of publication.
05
All seeds
| System and paper | Family | Year | OpenAlex | Sem. Scholar | Per year | Core share | Median citing year |
|---|
06
Mechanisms studied per system
Which mechanisms has the literature studied for which system? Each cell counts the papers of the curated corpus ( papers on task-based runtimes) whose system field names the system and whose themes include the mechanism; a paper can count for several systems and mechanisms. The matrix shows what the literature studies, not which features a system officially supports. Tap a cell or a system to open the list of papers with links.
07
Direct citations among the seeds
A filled cell means that the seed of its row cites the seed of its column (OpenAlex reference lists); both are ordered by year. Hover or tap a name to highlight a seed's references and citers.
Swipe the matrix sideways; tap a name to highlight it.
08
Co-citation network
Seeds placed by how often works cite them together (multidimensional scaling of Salton's cosine). Drag nodes; tap one to highlight its partners, double-tap to open its profile.
Swipe sideways to see the whole network.
Strongest pairs
| Pair | Works | Cosine |
|---|
09
Most cited works beyond the seeds
Works of the field set that are not seeds, ranked by the number of core works (works citing at least two seeds) that cite them; records of the same work are merged. "All" is the all-time OpenAlex count. Works that cite no seed are not part of the field set and cannot be ranked.
| # | Year | Work | Core works citing it | All citations |
|---|
10
Main paths of the citation network
The backbone of the core network by search path count; seeds in bold. Compare the global main path with the key-route and local paths.
11
Community venues and events
Workshops, conferences, user meetings, BoFs and tutorials dedicated to AMT systems, to single systems or to the broader field, one dot per edition; shaded bands mark the eras. The data set was compiled from workshop pages, host conference programs, mailing-list announcements and dblp (retrieved ); only editions that took place and are confirmed by a source are shown. Tap a row for all editions with their sources.
Swipe sideways to see all years.
12
Data
Citation data from OpenAlex, an open bibliographic database that links the references of the works it indexes (references it cannot match are missing, so some citations are not counted), with Semantic Scholar as a second source for the seed counts only, retrieved . Duplicate records of a seed (reprints, conference and journal versions) are merged; other works with the same title, such as a later tutorial abstract or a book chapter, are not. Counts measure attention, not quality.
Two databases, one ranking?
Citing works per seed in OpenAlex (duplicate records merged) against the citation count in Semantic Scholar, on logarithmic axes; the line marks equal counts, and labels mark the seeds whose counts differ by 30% or more (ratio Semantic Scholar to OpenAlex). Hover or tap for all others.
The BibTeX file holds the curated corpus, all seeds and the further references of the paper; its keywords field marks each entry as seed, corpus, reference or main path (a work on a main path). The JSON data hold everything this page shows; the full field set (every citing work with its OpenAlex ID, DOI, title, year and references) is data/field.json.gz of the reproducibility package.
This page, its figures and the underlying analysis were prepared with the help of Claude Opus 5.5 (Anthropic) and checked by the authors. If you find a mistake, please contact the authors (see the paper).