| paper | d4-conf-ijcai-2015-0715 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
Proposition 1 gives a semantic equivalence for preferred interpretations but asserts no algorithm, runtime, or complexity classification. The paper leaves polynomial-time query answering for incommensurable scales as future work, so no named computational result anchors a continuous mirror.
fails bit a — no named computational result to mirror
Proposition 1 and the paper’s merging results are semantic; the proposed future work on polynomial-time query answering is left open and supplies no computational result to mirror.
I can’t make a compliant positive case from this paper: it contains no numbered result asserting a computational complexity bound or classification. The closest is Proposition 1, which says that \(I\) is a model of the merge defined using all compatible scales iff some compatible scaling makes \(I\) a model of the corresponding min-based merge. That is a semantic equivalence, not a result that a problem is in P, NP-hard, or otherwise computationally classified. The conclusion instead leaves open whether the polynomial-time query-answering algorithm for a standard uncertain ABox can be extended to the incommensurable case.
The closest candidate mirror would treat sources as a population: a type is a source’s complete ABox and its within-source certainty ordering; \(\mu_t\) is the fraction of sources of type \(t\). One could ask which conjunctive queries hold in every preferred interpretation after merging across all compatible scales. But the paper’s min-based operator depends on which source types occur, not their masses. Duplicating a source type would not change the merge, so \(\mu_t\) would do no work in this formulation. Making mass affect the result would require changing the operator, and the paper provides no computational anchor for that new problem.
Thus there is no valid anchor, and no continuous problem I can present as a faithful mirror of a named complexity result. The weakest point is decisive: the paper motivates query-answering complexity but does not establish a named computational result for this paper to continuize.
The strongest objection is that there is no computational result here to continuize. Proposition 1 is a semantic equivalence: it characterizes preferred interpretations through compatible scalings. It gives no complexity classification or algorithmic bound. The paper’s conclusion instead leaves query-answering complexity as future work. So the proponent’s proposed query question is a possible research direction, but not a mirror of a computational result established in this paper.
The strongest version of that direction would use a large federation of data providers. A type could be a provider’s complete ABox together with its within-source certainty ordering, and \(\mu_t\) the fraction of providers of that type. That is a plausible high-multiplicity scenario. But the paper’s min-based merge is invariant to duplicating a provider: it depends on which source profiles occur, not how often they occur. A distribution over profiles therefore reduces to its support for this merge. If types instead describe individual facts, they lose the within-source ordering that compatible scales must preserve. To make source mass affect the result, one would need a mass-weighted or threshold-based fusion rule; that could be an interesting new problem, but it changes the operator and has no computational result in this paper to anchor it.
This supports a narrow negative verdict: the paper offers no worthwhile *mirror of one of its computational results*, because it proves none. It does not establish the broader claim that no continuous computational question could be worthwhile in any scenario. A continuum federation problem could be well-posed; the negative case is weaker there. The point is that it would be a new question about source aggregation, not a continuization of a named complexity result in this paper.
The adversarial triple: the proponent anchors on up to three named results; the opponent sees that case and must defeat every anchor; the judge decides which case convinced it. These are the pipeline’s own outputs, generated by tools/triple_run.py — no human edited them. The paper’s own text is not reproduced here beyond the quoted statement above.