{"n_sentences": 4316, "n_units": 4262, "conditional_share": 0.4361, "exceptions": [{"name": "The stacking-value clause", "def": "redeemed only when the paper delivers value beyond stacking existing tools \u2014 theoretical or practical, but new", "n": 759, "exemplars": ["A contribution must provide new theoretical or practical value beyond stacking existing tools.", "A contribution must demonstrate distinct innovative steps beyond existing techniques."], "share": 0.1759}, {"name": "The new-insight clause", "def": "redeemed only by a genuine new insight the parts did not already contain", "n": 674, "exemplars": ["Combination of existing techniques without new insights does not constitute significant contribution.", "combining existing methods without clear distinction from prior art does not constitute a significant contribution."], "share": 0.1562}, {"name": "The threshold, restated", "def": "no exception at all \u2014 a restatement of the significance bar itself", "n": 657, "exemplars": ["Incremental novelty does not meet the threshold for significant contribution.", "Applying established results trivially to a new setting does not constitute sufficient theoretical novelty."], "share": 0.1522}, {"name": "The clear-differentiation clause", "def": "redeemed only by explicit differentiation from the methods being adapted", "n": 534, "exemplars": ["Contribution is not well-defined if it merely adapts known methods without clear differentiation.", "Combining existing methods without significant new insight does not constitute a sufficient contribution."], "share": 0.1237}, {"name": "The new-mechanism clause", "def": "redeemed only by a fundamentally new mechanism, not a new arrangement", "n": 534, "exemplars": ["Combining existing techniques without a fundamentally new mechanism does not constitute sufficient novelty.", "Combining existing techniques without a fundamentally new insight may constitute limited novelty."], "share": 0.1237}, {"name": "The isolated-contribution clause", "def": "redeemed only when the unique contribution can be cleanly isolated from the assembled parts", "n": 440, "exemplars": ["Combining existing components does not constitute significant novelty if the unique contribution is unclear.", "combining existing components does not necessarily constitute significant novelty."], "share": 0.1019}, {"name": "The reproduction floor", "def": "not a combination exception \u2014 a separate boundary: reproducing known results is no contribution", "n": 381, "exemplars": ["reproducing known results does not constitute sufficient contribution.", "Re-parameterization of prior work without substantial conceptual advancement constitutes insufficient contribution."], "share": 0.0883}, {"name": "The integration-innovation clause", "def": "redeemed only by methodological innovation beyond engineering-level integration", "n": 337, "exemplars": ["Engineering integration of known components does not constitute substantial methodological innovation or new theoretical insight.", "Integration of existing engineering techniques does not constitute algorithmic creation or substantial novel insight."], "share": 0.0781}], "referents": [{"name": "The generic combination charge", "def": "\u201ccombines existing techniques\u201d with no part named and no mechanism specified", "n": 1955, "exemplars": ["The algorithm mainly combines existing methods; improvements seem expected.", "The method is a straightforward combination of existing techniques (Gaussian/categorical diffusion, standard equivariant architectures).", "Method shows limited novelty; overall contribution is insufficient."], "share": 0.4605}, {"name": "The component lineup", "def": "each named part of the method already exists in prior work \u2014 the parts are recited", "n": 760, "exemplars": ["Similar ideas (multi-scale learning, channel-wise attention, adversarial objectives) exist in prior work.", "The work primarily leverages existing techniques like retrieval augmentation and Decision Transformers."], "share": 0.179}, {"name": "The named-precedent match", "def": "the assembly as a whole resembles one specific named prior technique", "n": 698, "exemplars": ["The method resembles existing ideas from kernel-based neurons (Xu et al., 2001) and approximate learning for mixture of Gaussians (Verbeek et al., 2003).", "The paper applies known ideas (data augmentation, random sampling, self-supervised learning) to a new domain."], "share": 0.1644}, {"name": "The new-problem, old-tool charge", "def": "the problem is granted as new; the technique applied to it is the offense", "n": 496, "exemplars": ["Addresses new problem, but variance minimization approach lacks significant innovation.", "The arm filtering design and regret analysis follow relatively standard techniques, and the paper does not clearly articulate new analytical challenges introduced by latency or contextual dependencies."], "share": 0.1168}, {"name": "The no-new-insight charge", "def": "known techniques combined without demonstrating any new technical insight", "n": 336, "exemplars": ["The paper combines multiple known techniques without demonstrating a new technical insight.", "The paper aggregates existing techniques without clear specific innovations."], "share": 0.0792}]}