KScope: A Framework for Characterizing the Knowledge Status of Language Models
This work addresses the challenge of systematically evaluating LLM knowledge for researchers and practitioners, though it is incremental in building on prior knowledge conflict studies.
The authors tackled the problem of characterizing language models' knowledge status by introducing KScope, a hierarchical framework that classifies knowledge into five statuses based on consistency and correctness, and applied it to nine LLMs across four datasets, finding that supporting context narrows knowledge gaps and specific context features improve update effectiveness.
Characterizing a large language model's (LLM's) knowledge of a given question is challenging. As a result, prior work has primarily examined LLM behavior under knowledge conflicts, where the model's internal parametric memory contradicts information in the external context. However, this does not fully reflect how well the model knows the answer to the question. In this paper, we first introduce a taxonomy of five knowledge statuses based on the consistency and correctness of LLM knowledge modes. We then propose KScope, a hierarchical framework of statistical tests that progressively refines hypotheses about knowledge modes and characterizes LLM knowledge into one of these five statuses. We apply KScope to nine LLMs across four datasets and systematically establish: (1) Supporting context narrows knowledge gaps across models. (2) Context features related to difficulty, relevance, and familiarity drive successful knowledge updates. (3) LLMs exhibit similar feature preferences when partially correct or conflicted, but diverge sharply when consistently wrong. (4) Context summarization constrained by our feature analysis, together with enhanced credibility, further improves update effectiveness and generalizes across LLMs.