🤖 AI Summary
This study addresses the lack of systematic understanding regarding how large language models (LLMs) respond to subtle semantic perturbations in document-level semantic similarity scoring. The authors construct a multifactorial experimental framework, embedding controlled semantic perturbations—such as negation, conjunction substitution, and named entity replacement—across varying positions, topical relevance levels, and document lengths. Through tens of thousands of comparative evaluations across five prominent LLMs, the work uncovers several novel phenomena: a positional bias wherein perturbations near document beginnings elicit stronger sensitivity, a polarization effect induced by topically irrelevant context, and consistent “scoring fingerprints” unique to each model despite a shared hierarchy in tolerance to perturbations. Building on these insights, the study introduces the first LLM-agnostic auditing toolkit for semantic similarity scoring.
📝 Abstract
We propose a scalable, multifactorial experimental framework that systematically probes LLM sensitivity to subtle semantic changes in pairwise document comparison. We analogize this as a needle-in-a-haystack problem: a single semantically altered sentence (the needle) is embedded within surrounding context (the hay), and we vary the perturbation type (negation, conjunction swap, named entity replacement), context type (original vs. topically unrelated), needle position, and document length across all combinations, testing five LLMs on tens of thousands of document pairs. Our analysis reveals several striking findings. First, LLMs exhibit a within-document positional bias distinct from previously studied candidate-order effects: most models penalize semantic differences more harshly when they occur earlier in a document. Second, when the altered sentence is surrounded by topically unrelated context, it systematically lowers similarity scores and induces bipolarized scores that indicate either very low or very high similarity. This is consistent with an interpretive frame account in which topically-related context may allow models to contextualize and downweight the alterations. Third, each LLM produces a qualitatively distinct scoring distribution, a stable "fingerprint" that is invariant to perturbation type, yet all models share a universal hierarchy in how leniently they treat different perturbation types. Together, these results demonstrate that LLM semantic similarity scores are sensitive to document structure, context coherence, and model identity in ways that go beyond the semantic change itself, and that the proposed framework offers a practical, LLM-agnostic toolkit for auditing and comparing scoring behavior across current and future models.