D2C-Routing: Dimension-to-Composition Evidence Routing for Mixed-Origin AI-Generated Text Detection
2026-08-27 • Computation and Language
Computation and Language
AI summaryⓘ
The authors studied how to detect texts that are partly written by humans and partly by AI, instead of just labeling entire texts as human or AI-made. They developed a method called Dimension-to-Composition Routing (D2C-Routing) that looks separately at the content and the style of writing to decide who wrote each part. Their approach can better identify mixed-origin texts by combining these clues into four types of collaboration. They tested their method on a special dataset and found it performed better than previous tools. However, they note it is still hard to tell apart texts with AI content but human style from those fully AI-generated.
AI-generated text detectionmixed-origin writingcontent originexpression originDimension-to-Composition Routingsource attributionHART benchmarkcollaboration typestrue positive ratefalse positive rate
Authors
Xin Chen, Fuwei Zhang, Yiqi Tong, Wei Guo, Yutian Xiao, Fuzhen Zhuang
Abstract
AI-generated text detection is commonly framed as a binary document-level judgment about whether a text is human-written or machine-generated. This framing breaks down for mixed-origin writing, where content origin and expression origin may differ. We cast mixed-origin detection as dimension-to-composition source attribution, inferring content origin and expression origin before composing them into four collaboration types. We propose Dimension-to-Composition Routing (D2C-Routing), which routes content-side and expression-side evidence to supervised dimension heads before a learned gated composition layer predicts the final label. On MixD2C, a reconstructed split derived from the HART mixed-origin benchmark, our disclosed D2C-Routing-based detector system reaches 0.8603 four-way Avg TPR@1%FPR, 6.5 points above the same-split RACE-local rerun. Core ablations support the routing design, while error analysis shows that distinguishing AI-content/human-expression from fully AI-generated text remains the hardest boundary. Code is available at https://github.com/bystander563/d2c-routing-artifact.