E.11.PUR:11 - SoTA-Echoing
| Source or practice line | Problem-solving move taken here | Adoption and boundary |
|---|---|---|
| Que et al., LLM-as-a-Judge for Reliable and Explainable Offline Evaluation in Top-K Recommendation, KDD 2026, arXiv:2606.22961 | Observed feedback and Top-K scores can be biased proxies; pair a judgement with explicit rationale rather than treating the score as self-explanatory. | Adapt the proxy warning and rationale pressure to current candidate fit and expected-result reasoning. Reject the recommender, Top-K, user-profile, and LLM-judge ontology as a model of FPF recommendation authority. |
| Nunes and Jannach, A Systematic Review and Taxonomy of Explanations in Decision Support and Recommender Systems, User Modeling and User-Adapted Interaction 27 (2017) | Lineage for separating recommendation explanation functions and making reasons addressable to a receiving decision. | Retain as lineage, not current-best evidence. Candidate and coordination rationales do not prove applicability or authorize action. |
| Jin, Bai, and Oulasvirta, Modeling Trial-and-Error Navigation With a Sequential Decision Model of Information Scent, arXiv:2603.11759 (2026) | Preserve bounded search, wrong-turn recovery, and reconsideration under limited attention. | Adapt to candidate reconsideration and return boundaries. The preprint does not decide recommendation authority or record cardinality. |
| Current FPF NQD and OEE lines together with A.19 comparison practice | Preserve plural candidates, non-dominated alternatives, explicit comparison spaces, and dynamic reconsideration. | Adopt the plurality discipline. PUR coordinates pattern uses but does not replace subject-domain candidate evaluation. |
Current FPF C.2.1, A.10, and G.11 result-identity, reliance, and currentness line | Compare the earlier result under its direct subject pattern, then state later reliance and currentness separately. | Adopt the separation. Reuse ends repeated work only when the result’s relevant values still match; the branch creates no generic result-reuse relation. |
The practical implication is to recommend a use for its expected result, not for its familiarity or score, to stop on an earlier result that still answers the concern, and to add order only where a real dependency exists.
Que et al. is the current decision-bearing recommender source in this narrow use; Nunes and Jannach supplies lineage. The 2026 navigation preprint supplies bounded reconsideration, while current FPF NQD, OEE, A.19, C.2.1, A.10, and G.11 supply the transdisciplinary candidate, comparison, result-identity, reliance, and currentness basis. These sources change 4.1-4.5, 5.1.1, and 5.6; none decides FPF kinds or recommendation authority.
Reopen the score-proxy adaptation when stronger evaluation evidence shows that the relied-on score tracks current expected-result and receiving-use fit without the identified exposure or rationale loss. Reopen the wrong-turn adaptation when peer review, replication, or use evidence changes the value of reconsideration. Use G.11 to manage source and telemetry currentness, then revise the affected PUR fit, rationale, recommendation, or return relation.