Scholarlink

Abstract <p>Improving human–machine interfaces for intelligent systems in industries with high decision-making responsibility, such as the electric power industry, is considered. A method developed by the author based on the Shapley additive explanation method is described. A comprehensive modification of this method is proposed featuring normalization of the feature contribution vector, semantic grouping of interrelated features, visualization of contributions that differs from the currently used visualization in the element arrangement and in the generation of text explanations using a language model. The presented method aims to reduce the cognitive load on users when analyzing recommendations and forecasts generated by intelligent decision support systems and to increase the confidence of industry specialists in such systems. As an example, the problem of short-term forecasting of electricity consumption at an industrial enterprise is considered.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Method for Explaining Forecasts of Artificial Intelligence Models Based on the Shapley Algorithm and a Generative Language Model

  • 期刊论文

  • 发表日期：

    2026年2月1日

查看全文
  • P. V. Matrenin

摘要

Abstract

Improving human–machine interfaces for intelligent systems in industries with high decision-making responsibility, such as the electric power industry, is considered. A method developed by the author based on the Shapley additive explanation method is described. A comprehensive modification of this method is proposed featuring normalization of the feature contribution vector, semantic grouping of interrelated features, visualization of contributions that differs from the currently used visualization in the element arrangement and in the generation of text explanations using a language model. The presented method aims to reduce the cognitive load on users when analyzing recommendations and forecasts generated by intelligent decision support systems and to increase the confidence of industry specialists in such systems. As an example, the problem of short-term forecasting of electricity consumption at an industrial enterprise is considered.