Utility-Aware Data Pricing: Token-Level Quality and Empirical Training Gain for LLMs
For LLM data market participants, this provides a fairer, auditable pricing method that captures nonlinear data contributions, though the framework is domain-specific and incremental in combining existing techniques.
This paper proposes a dynamic data valuation framework for LLMs that moves from static row-count pricing to utility-based pricing using token-level information density, empirical training gain (via influence functions and Data Shapley), and cryptographic verifiability. Experiments on three domains show proxy-based empirical gain achieves near-perfect ranking alignment with realized utility, outperforming row-count and token-count baselines.
Traditional data valuation methods based on ``row-count $\times$ quality coefficient'' paradigms fail to capture the nuanced, nonlinear contributions that data makes to Large Language Model (LLM) capabilities. This paper presents a dynamic data valuation framework that transitions from static accounting to utility-based pricing. Our approach operates on three layers: (1) token-level information density metrics using Shannon entropy and Data Quality Scores; (2) empirical training gain measurement through influence functions, proxy model strategies, and Data Shapley values; and (3) cryptographic verifiability through hash-based commitments, Merkle trees, and a tamper-evident training ledger. We provide comprehensive experimental validation on three real domains (instruction following, mathematical reasoning, and code summarization), demonstrating that proxy-based empirical gain achieves near-perfect ranking alignment with realized utility, substantially outperforming row-count and token-count baselines. This framework enables a fair Data-as-a-Service economy where high-reasoning data is priced according to its actual contribution to model intelligence, while providing the transparency and auditability necessary for trustworthy data markets.