Intelligent Retrieval of Corresponding Engineering Standard Clauses for Cross-Border Construction Projects
Year: 2026
Type: article
Abstract: Cross-border construction projects increasingly involve heterogeneous engineering standards from different national systems, creating challenges for standards comparison and compliance assessment. Existing approaches largely rely on textual similarity or syntactic matching, which may be insufficient for identifying semantically corresponding technical provisions. This study proposes an engineering-aware intelligent retrieval framework for identifying potentially corresponding clauses across national engineering standards. The framework integrates multilingual semantic embeddings with engineering-specific terminology, numerical information, and topic-based search-space constraints to retrieve and rank candidate clauses for subsequent expert comparison. The methodology was evaluated using a controlled benchmark comprising 105 manually aligned Chinese–Uzbek engineering clause pairs across 11 engineering topics. The results achieved a Top-1 Accuracy of 62.9% and a Recall@5 of 91.4%, demonstrating the feasibility of automated retrieval for identifying semantically corresponding engineering provisions. The findings suggest that engineering-specific information can provide constraints beyond general semantic similarity and substantially narrow the candidate search space in cross-standard clause retrieval. The proposed framework is intended as a decision-support layer preceding expert interpretation rather than as a substitute for technical judgment or regulatory compliance assessment. It provides an initial methodological basis for improving the efficiency and consistency of cross-standard comparison and for supporting the digital management of heterogeneous engineering standards in cross-border construction projects.
Source: Buildings
Institutions: Chongqing University, Beijing City University
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,137
OpenAlex ID: W7222460004