dorsal/arxiv
View SchemaInterpretable Probability Estimation with LLMs via Shapley Reconstruction
| Authors | Yang Nan, Qihao Wen, Jiahao Wang, Pengfei He, Ravi Tandon, Yong Ge, Han Xu |
|---|---|
| Categories | |
| ArXiv ID | 2601.09151vv1 |
| URL | https://arxiv.org/abs/2601.09151 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Large Language Models (LLMs) demonstrate potential to estimate the probability of uncertain events, by leveraging their extensive knowledge and reasoning capabilities. This ability can be applied to support intelligent decision-making across diverse fields, such as financial forecasting and preventive healthcare. However, directly prompting LLMs for probability estimation faces significant challenges: their outputs are often noisy, and the underlying predicting process is opaque. In this paper, we propose PRISM: Probability Reconstruction via Shapley Measures, a framework that brings transparency and precision to LLM-based probability estimation. PRISM decomposes an LLM's prediction by quantifying the marginal contribution of each input factor using Shapley values. These factor-level contributions are then aggregated to reconstruct a calibrated final estimate. In our experiments, we demonstrate PRISM improves predictive accuracy over direct prompting and other baselines, across multiple domains including finance, healthcare, and agriculture. Beyond performance, PRISM provides a transparent prediction pipeline: our case studies visualize how individual factors shape the final estimate, helping build trust in LLM-based decision support systems.
{
"annotation_id": "8a0c904a-54a1-4a76-a1d4-e4738de1f1b6",
"date_created": "2026-02-17T05:53:20.217000Z",
"date_modified": "2026-02-17T05:53:20.217000Z",
"file_hash": "3f06a5f65b0299fb2c3191d3a7bf77bbc8eef00d783ce6c2ee8ad268d50bd2e7",
"private": false,
"record": {
"abstract": "Large Language Models (LLMs) demonstrate potential to estimate the probability of uncertain events, by leveraging their extensive knowledge and reasoning capabilities. This ability can be applied to support intelligent decision-making across diverse fields, such as financial forecasting and preventive healthcare. However, directly prompting LLMs for probability estimation faces significant challenges: their outputs are often noisy, and the underlying predicting process is opaque. In this paper, we propose PRISM: Probability Reconstruction via Shapley Measures, a framework that brings transparency and precision to LLM-based probability estimation. PRISM decomposes an LLM\u0027s prediction by quantifying the marginal contribution of each input factor using Shapley values. These factor-level contributions are then aggregated to reconstruct a calibrated final estimate. In our experiments, we demonstrate PRISM improves predictive accuracy over direct prompting and other baselines, across multiple domains including finance, healthcare, and agriculture. Beyond performance, PRISM provides a transparent prediction pipeline: our case studies visualize how individual factors shape the final estimate, helping build trust in LLM-based decision support systems.",
"arxiv_id": "2601.09151",
"authors": [
"Yang Nan",
"Qihao Wen",
"Jiahao Wang",
"Pengfei He",
"Ravi Tandon",
"Yong Ge",
"Han Xu"
],
"categories": [
"cs.LG"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "Interpretable Probability Estimation with LLMs via Shapley Reconstruction",
"url": "https://arxiv.org/abs/2601.09151",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "6822aa87-b6ce-4681-8c7e-4dbbf01ae6af",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}