Truncation—the use of wildcard symbols to capture morphological variants of word stems—is a foundational skill in information literacy, essential for comprehensive, reproducible scholarly searching. In sport and exercise sciences, where terms like sport* , athlete* , and exercis* require high-recall searching, reliable wildcard processing is critical for evidence synthesis. Yet as academic platforms introduce “smart” search interfaces that automate query interpretation, the reliability of this core function can no longer be taken for granted. This study presents a systematic diagnostic evaluation of wildcard processing in Web of Science Smart Search, examining right truncation, left truncation, internal masking, and their combinations across a diagnostic test suite. Grounded in the Three-Tier Model of Retrieval Intelligence and employing quantitative comparison with the deterministic Advanced Search benchmark alongside qualitative deep case analysis, the findings reveal four distinct failure modes: omission, undocumented expansion, syntactic restructuring, and multi-wildcard compounding. These failures stem from an undocumented dual-parser architecture: a flawed parser activated by simple keyword queries and a correct parser triggered only when Boolean operators are present. The study translates these findings into actionable resources: the SIFT-S framework for critical technological literacy, a diagnostic guide linking user symptoms to failure patterns, teachable cases derived from the four failure modes, and empirical evidence for advocating vendor transparency. By bridging technical evaluation with library practice, this research equips librarians and educators to foster critical technological literacy in an era where the “intelligence” of search systems can no longer be assumed reliable.
Yang et al. (Mon,) studied this question.