aiqbench analysis
This research introduces a novel "computational fingerprint" that strips domain-specific terminology to identify shared mathematical problems across unrelated scientific fields, significantly improving cross-domain solution retrieval. By replacing standard text embeddings with this mechanism skeleton, the method outperforms even plain abstract+TF-IDF approaches in finding equivalent solvers for tasks like Bayesian state estimation. The work demonstrates a practical pathway for "solution import," allowing specialized algorithms from one discipline to be directly applied to analogous problems in others without re-deriving them.
Same Problem, Different Field: Cross-Domain Solution Import via Domain-Stripped Computational Fingerprints
arXiv:2609.07595v1 Announce Type: cross 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.
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.
Accepted as a full paper at JCDL 2026 (The 2026 ACM/IEEE Joint Conference on Digital Libraries), Frisco, TX, October 13-16, 2026. 10 pages plus references, 2 figures, 8 tables. Code and benchmark: this https URL ; archived dataset (KU Leuven RDR): this https URL
Digital Libraries (cs.DL); Computation and Language (cs.CL); Information Retrieval (cs.IR)
arXiv:2609.07595 [cs.DL]
(or arXiv:2609.07595v1 [cs.DL] for this version)
https://doi.org/10.48550/arXiv.2609.07595 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 340 words
Related coverage
Game Design Creativity evals
LessWrong · Oct 12, 2026
Presentation: From Hype to Strong Foundations: What the Rise, Fall and Resurgence of Agents Can Teach Us About Outlasting the Cycle
InfoQ · Jun 17, 2026
Claude Code’s Next Era — Thariq Shihipar, Anthropic
Latent Space · Sep 29, 2026
Podcast: The Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes
InfoQ · Aug 24, 2026