Innovation in China's construction industry depends on the interplay among universities, firms, and government, yet existing Triple Helix studies rarely provide dynamic, data‐driven measures suited to this project‐based, policy‐led context. To develop and validate a computational framework that quantifies and visualizes university–industry–government interactions, enabling timely diagnosis of innovation drivers, barriers, and coordination patterns. We build a large‐scale text corpus of 1.2 million records from news, government documents, corporate reports, and academic papers (2020–2025). After domain‐specific segmentation and named entity recognition, we derive four indices—attention, trend, sentiment, and actor relevance—to map signals onto strengths, weaknesses, opportunities, and threats. Topic modeling identifies thematic clusters, and time‐decayed cooccurrence networks track evolving relationships among actors and strategic elements. A longitudinal case study demonstrates practical use. The framework reveals a structurally asymmetric configuration characterized as policy‐led, industry‐responsive, and weak university–industry coupling. Interaction dynamics follow a stage‐based pattern—policy initiation, industry adaptation, and academic alignment—showing time‐lagged coordination rather than continuous synergy. The indices detect emerging topics, expose sentiment misalignment across stakeholders, and identify engagement gaps. Cooccurrence networks distinguish short‐lived attention spikes from persistent collaborations and help track policy impacts on technology adoption. The proposed framework converts heterogeneous texts into actionable evidence for monitoring innovation dynamics in near real time. It advances Triple Helix research with quantifiable, replicable metrics tailored to the construction sector and provides decision support for aligning policies, industry implementation, and academic contribution.
Xiang et al. (2026) studied this question.