Recall Isn't Enough: Bounding Commitments in Personalized Language Systems
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913143336206336 |
|---|---|
| author | Tang, Rui Zhang, Yichi Chen, Xi Dong, Chen Yang, Youwei Shen, Yumeng Liu, Qiangqiang |
| author_facet | Tang, Rui Zhang, Yichi Chen, Xi Dong, Chen Yang, Youwei Shen, Yumeng Liu, Qiangqiang |
| contents | Long-context and memory systems usually treat personalization as a recall problem. In practice, many failures occur later, when a system commits: it turns noisy hints into hard constraints, drops rare witnesses, forgets downstream obligations, or answers despite infeasibility. We introduce Contract-Bounded Evidence Activation (CBEA) with Lexicographic Commitment Validation (LCV). CBEA activates a bounded evidence set using typed coverage, tail witnesses, and consequence debt; LCV validates structured commitments before prose and routes infeasible states to repair, abstention, or recontract. Across 360 fixtures and three generation backends, CBEA+LCV reaches zero failures within validator scope at 0.49-0.60 availability over attempted runs. Raw and long-context baselines with the same LCV gate reach zero only at 0.003-0.092. A shadow oracle diagnostic marks the limit: CBEA+LCV recalls 0.012 of uncompiled visible facts, while raw recalls 0.53. The result is a bounded operating point: explicit commitment control and 74-75% lower median input payload, not universal memory dominance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_16712 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Recall Isn't Enough: Bounding Commitments in Personalized Language Systems Tang, Rui Zhang, Yichi Chen, Xi Dong, Chen Yang, Youwei Shen, Yumeng Liu, Qiangqiang Artificial Intelligence Computation and Language Human-Computer Interaction I.2.7; H.3.3 Long-context and memory systems usually treat personalization as a recall problem. In practice, many failures occur later, when a system commits: it turns noisy hints into hard constraints, drops rare witnesses, forgets downstream obligations, or answers despite infeasibility. We introduce Contract-Bounded Evidence Activation (CBEA) with Lexicographic Commitment Validation (LCV). CBEA activates a bounded evidence set using typed coverage, tail witnesses, and consequence debt; LCV validates structured commitments before prose and routes infeasible states to repair, abstention, or recontract. Across 360 fixtures and three generation backends, CBEA+LCV reaches zero failures within validator scope at 0.49-0.60 availability over attempted runs. Raw and long-context baselines with the same LCV gate reach zero only at 0.003-0.092. A shadow oracle diagnostic marks the limit: CBEA+LCV recalls 0.012 of uncompiled visible facts, while raw recalls 0.53. The result is a bounded operating point: explicit commitment control and 74-75% lower median input payload, not universal memory dominance. |
| title | Recall Isn't Enough: Bounding Commitments in Personalized Language Systems |
| topic | Artificial Intelligence Computation and Language Human-Computer Interaction I.2.7; H.3.3 |
| url | https://arxiv.org/abs/2605.16712 |