| paper | d4-conf-ijcai-2015-0518 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
The paper reports simulations and empirical comparisons, but contains no numbered theorem, lemma, corollary, or proposition asserting a computational result. The proposed type-based diffusion optimization is not anchored to a named result from the paper, so the computational-result requirement fails.
fails bit a — no named computational result to mirror
No named computational result is covered; the paper's centrality, clustering, and rewiring findings are empirical results on finite graphs.
I can’t make a compliant anchor-based positive case: this paper has no numbered Theorem, Lemma, Corollary, or Proposition asserting a complexity result or algorithmic classification. Sections 3.2–3.5 report simulations, clustering, and observed improvements from rewiring; those are empirical findings, not eligible anchors. There are therefore no anchors to quote and restate, and no anchor-based question to classify as tractable, hardness-transferring, or continuum-specifically hard.
The strongest positive intuition is an unanchored extension: on a large platform with many users in recurring interaction-role classes, assign each class a population share and a contact profile, then choose which mass of low-spreading users has a link redirected toward influential users, maximizing expected final informed mass under the paper’s Maki–Thompson process. That could preserve the paper’s central intervention and objective. Its weak point is substantial: the paper relies on individual network topology and does not define a type-level contact model or formulate that optimization problem. So it motivates a possible continuous computational problem, but it does not establish a mirror of any named result in the paper.
There is no anchor to mirror here. The paper states no theorem, lemma, or corollary about the complexity of a computational problem. Its contributions are empirical: it compares simulated spreading across vertices, relates spreading scores to centrality measures, clusters users, and reports results from rewiring edges in three fixed networks. So the proponent’s proposed optimization is an extension of the paper’s subject, not a continuous version of one of its named results.
The strongest version of that extension would give user classes population shares and a contact kernel, then choose how much mass to rewire to maximize the final informed share under a continuous Maki–Thompson model. That is a coherent computational question, but its answer depends on a network model the paper does not supply. The reported rankings use each vertex’s position in a particular graph; the intervention redirects a particular edge toward a particular high-centrality vertex. A coarse type model can lose the local graph structure that drives those scores and the spreading dynamics. A type detailed enough to retain that structure may need to distinguish an ever-growing number of network positions. Either way, the optimization studies a new modeled diffusion problem, not the paper’s result.
That is the strongest negative case: the paper offers no computational result to anchor, and its main intervention is tied to specific graph instances. It does not establish the broader claim that no worthwhile continuous model exists. A population model with recurring interaction roles could still be a useful neighboring project; arguing that it is not worth the programme’s time would go beyond what this paper supports.
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.