Computational fingerprints help find shared solutions across different science fields
Same Problem, Different Field: Cross-Domain Solution Import via Domain-Stripped Computational Fingerprints
Digital LibrariesComputation and LanguageInformation Retrieval
Summary
Many scientific fields solve the same complex problems but call them by different names, which makes it hard to spot when solutions are similar. The authors created a special way to describe papers by focusing on the core computational steps and ignoring the field-specific words. This approach helps find papers that solve the same problem in different fields, even when usual search methods fail. They tested it on many papers and verified with experts that it can successfully identify useful solution matches across fields. This could help scientists borrow better methods from other areas more easily.
recursive Bayesian state estimationKalman filterBayesian forecastingdata assimilationcomputational fingerprintcross-domain retrievalnatural language processingscientific embeddingsTF-IDFsolution import
Authors
Eryk Kulikowski
Abstract
The same underlying computational problem is solved across unrelated fields under different names: recursive Bayesian state estimation appears as a "Kalman filter" in control, "Bayesian forecasting" in pharmacokinetics, and "data assimilation" in geoscience. Topical and citation-based scientific embeddings cannot see this shared problem. We distill each paper once into a domain- and method-name-stripped faceted computational fingerprint, a free-text mechanism skeleton plus controlled computational facets. We define a tunable, facet-selectable distance over it. The goal is solution import: surface cross-field pairs solving the same problem, so a bespoke implementation can be swapped for another field's standard, specialized solver. On a benchmark of 18 method families across 109 papers, the skeleton lifts cross-domain retrieval average precision over the abstract from 0.222 to 0.513, and the whole fingerprint reaches 0.557. Strikingly, four trained scientific embedders all fall below plain abstract+TF-IDF: they encode topical and citation similarity, the wrong signal for this task. The gain is the representation: the abstract-to-skeleton swap lifts every embedder, and the pipeline is one cached LLM call per paper plus a cheap embedder. An interventional re-skin / math-edit test shows the fingerprint tracks the computation, not the field. On a 501-paper wild corpus, known twins dominate the top of the ranking (23 of the top 30); with planted pairs excluded from the results, three blind LLM judges rate 3 of the top 5 and 8 of the top 30 pairs genuine import candidates, and 0 of 30 random ones. The human verification is the four executed imports: in one, an open standard solver reproduces a bespoke clinical dosing engine's output. We release the benchmark, the code, and the distillation prompt.