interaction effect decomposition

This paper addresses the challenge of uniformly detecting higher-order feature interactions—specifically synergistic, redundant, and independent relationships—which are difficult to characterize with existing methods. We propose a geometric analysis framework based on random sequential feature addition. By modeling how individual feature contributions evolve as features are added in varying orders, we observe that their trajectories in a two-dimensional plane consistently exhibit an L-shaped pattern. Leveraging this geometric regularity, we define the continuous, interpretable L-score (ranging from −1 to +1), which enables inference of third- and higher-order interaction structures using only pairwise interaction measurements. Our method imposes no assumptions about underlying model architecture, requires no gradient computation or predefined distance metrics, and yields unbiased identification of feature dominance and interaction type. Extensive experiments across diverse domains demonstrate accurate discrimination among synergistic (e.g., Y = X₁X₂), redundant (e.g., X₁ ≈ X₂), and independent relationships—achieving, for the first time, model-agnostic, quantitative, and interpretable decomposition of higher-order interactions.

Modeling higher-order ternary interactions—where one node dynamically modulates the pairwise interaction between two others—has long been overlooked in complex biological systems, limiting mechanistic understanding beyond traditional pairwise network assumptions. Method: We propose an information-theoretic paradigm, establishing the first theoretical framework characterizing how ternary interactions modulate pairwise mutual information. Building on this, we develop TRIM (Ternary Interaction Miner), a scalable algorithm that infers statistically significant ternary regulatory relationships directly from nodal metadata (e.g., gene expression profiles), integrating mutual information estimation, rigorous statistical inference, and data-driven validation. Results: Applied to multi-omics data from acute myeloid leukemia, TRIM identifies multiple novel, experimentally verifiable ternary regulatory modules, substantially improving resolution of transcriptional and post-transcriptional regulatory mechanisms. The method demonstrates robust generalizability across disease contexts, offering a principled foundation for uncovering higher-order regulatory logic in biological networks.

This paper addresses complex randomized experiments subject to interference between units—such as social network interventions—where standard causal inference assumptions fail. Method: We develop a design-based theoretical framework for estimating treatment effects, introducing a family of design-compatible estimators and a scalar, interpretable measure of “experimental complexity.” We establish its theoretical connection to design variance, derive the asymptotic variance lower bound for unbiased estimation under arbitrary designs, and propose a consistent variance estimator. Contributions/Results: Through interference modeling, design-based inference foundations, and network experiment simulations, we validate our approach on real-world social network data from an insurance adoption study. Our estimators achieve significantly improved estimation accuracy and consistent variance estimation compared to existing methods, providing a theoretically rigorous yet practically implementable analytical framework for complex experimental designs.

Existing visualization research predominantly focuses on *how to use* interactive features, neglecting the critical question of *how to construct* them. Method: We propose the first three-layer decoupled interaction authoring task model—intent–technique–component—derived from empirical coding and abstraction of 592 interaction units across 47 real-world applications. Contribution/Results: This model provides descriptive, evaluative, and generative capabilities, enabling the first unified formalization of interaction authoring intent, technical implementation, and component instantiation. It yields a reusable, theory-grounded classification framework that supports critical evaluation of existing visualization tools and informs the design and validation of next-generation low-code interaction authoring systems.

This paper studies networked innovation processes, where each process is modeled as an infinite-color Pólya urn to capture novelty emergence. Addressing the limitation of existing models—which ignore historical interdependencies among processes—we develop, for the first time, a second-order asymptotic theory for interactive innovation processes, characterizing their joint growth rates and covariance structure. We propose a general statistical framework based on intensity function estimation and point-process inference to quantify the direction and magnitude of cross-process influence. The methodology is empirically validated on Reddit community evolution and Gutenberg textual innovation data, demonstrating both theoretical consistency and statistical robustness. Our approach provides a scalable theoretical toolkit and practical methodology for modeling innovation diffusion and conducting causal inference across diverse domains.

Existing explainable AI (XAI) methods struggle to characterize the precise functional forms of feature interactions, typically supporting only limited types of detection or visualization. This work proposes the SAILS framework, which, for the first time in a model-agnostic setting, fits interpretable generalized additive model (GAM) surrogates to local effects and decomposes pairwise feature interactions at the derivative level. The method systematically categorizes these interactions into three types—linear, multiplicatively separable, and non-multiplicatively separable—and provides tailored visualizations for each. By integrating local smoothing, derivative-based interaction decomposition, statistical significance testing, and interaction-type classification, SAILS effectively uncovers interaction mechanisms in both synthetic and real-world tasks, thereby addressing a critical gap in XAI regarding the modeling of interaction functional forms. Nevertheless, limitations remain in scenarios involving highly correlated features or higher-order interactions.

This study addresses the pronounced sensitivity of large language models (LLMs) to prompt phrasing order when generating two-level fractional factorial designs, demonstrating that the sequence of phrases in a prompt significantly affects output quality. To tackle this issue, the authors introduce, for the first time, a sequential addition experimental design framework into prompt engineering. This approach systematically quantifies the ordering effects of individual prompt components and automatically identifies the optimal prompt configuration. The proposed method not only elucidates the underlying mechanisms by which LLMs respond to structural variations in prompts but also substantially enhances both the performance and stability of LLMs in statistical experimental design tasks.

本文通过图论框架定义因果有序方差分量，并从观测数据中识别这些分量，以解决传统方差分解缺乏因果解释的问题。

This study addresses the challenge of characterizing and quantifying higher-order homophily and heterophily in hypergraphs by proposing the first unified framework that integrates both measurement and generative modeling. Clarifying the conceptual distinctions between higher-order mixing patterns and traditional pairwise homophily, the work establishes a comprehensive suite of metrics tailored specifically for hypergraphs. It further provides a systematic review of existing random hypergraph generative models, delineating the conditions under which each model family is appropriate. By laying a coherent theoretical foundation for the study of higher-order homophily, this research offers clear methodological guidance for future model selection and design, thereby advancing the broader field of higher-order network analysis.

This study addresses the challenge of biased effect estimation in online controlled experiments caused by overlapping tests on shared traffic, which hinders accurate assessment of feature interactions. To resolve this, the authors propose Multi-Experiment Analysis (MEA), a method grounded in statistical modeling and causal inference that consistently estimates joint effects under arbitrary partial or full overlap and multi-variant settings—without requiring predefined factorial designs or constrained traffic allocation. MEA uniquely enables, without coordination overhead, the simultaneous modeling of bias-corrected individual effects, joint effects for any combination of variants, and conditional effects. Simulations confirm the estimator’s consistency and nominal confidence interval coverage, and the approach has been successfully deployed in large-scale production systems across multiple real-world business applications.