Newest papers

Charts as Epistemic Actions for Experience-Driven Agentic Data Analysis

Xinzhe Wang, Xudong Weng, Zang Li, Shaolei Zhang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.30072 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Os agentes de dados desenhados construídos em loops de estilo ReAct agora executam fluxos de trabalho analíticos complexos sobre dados estruturados, mas seu raciocínio intermediário ainda corre quase inteiramente através do texto. Os analistas humanos, em vez disso, se movem entre números, tabelas e gráficos, lendo evidências visuais para expor tendências e anomalias que os resumos de texto caem. Perguntamos se os gráficos podem servir como ações epistemológicas internas dentro do loop de raciocínio de um agente de dados, e como ensinar um agente a usá-los. Primeiro construímos \probe{}, uma série de 200 tarefas controladas sob quatro condições de interface correspondentes, e encontramos uma lacuna persistente entre o que os agentes podem ler e o que escolhem usar. Os agentes

Machine translation (Meta M2M100); the English original follows.

Abstract

Data agents built on ReAct-style loops now execute complex analytical workflows over structured data, yet their intermediate reasoning still runs almost entirely through text. Human analysts instead move between numbers, tables, and charts, reading visual evidence to expose trends and anomalies that text summaries drop. We ask whether charts can serve as internal epistemic actions inside a data agent's reasoning loop, and how to teach an agent to use them. We first build \probe{}, a suite of 200 controlled tasks under four matched interface conditions, and find a persistent gap between what agents can read and what they choose to use. Agents score as well from self-generated charts as from numeric output, yet plotting access alone does not translate into consistent use. Task slices show that charts lead on temporal, pattern, and anomaly reasoning, while text leads on precise reading and conditional verification. We then propose \method{}, which distils two families of experience from conditioned trajectory rollouts. One family records how to select and verify analytical operations, and the other records when to invoke text, charts, or both. A trajectory judge scores outcomes and labels intermediate steps, and a validation gate admits only experience that improves held-out performance. On three held-out benchmarks with four LLM backbones, \method{} raises analytical accuracy and visual evidence quality over matched baselines, and backbones that never drew charts begin using them, all without touching model parameters.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Task (project management), Workflow, Suite, Reading (process), Quality (philosophy)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange