Trial Balance Transactions and Income Statement Analysis Dataset (LLM-Generated Prompts and Outputs)

This dataset contains ChatGPT prompts and the corresponding model-generated outputs used to create income statements from trial balance transactions. It was generated as part of the study “Evaluating Large Language Models’ Emergent Capabilities in Generating Income Statements from Accounting Transactions.” The dataset includes: Structured prompt sets used with ChatGPT Generated outputs for each trial balance scenario Notes for comparing outputs with expected income statements This dataset is intended to support reproducibility, analysis, and further research on AI-assisted accounting and financial data generation. Citation suggestion: Ndlazi, Trevor. (2025). Trial Balance Transactions and Income Statement Analysis Dataset (LLM-Generated Prompts and Outputs). Zenodo. DOI: https://doi.org/10.5281/zenodo.17095629本数据集包含用于从试算平衡表（trial balance）交易中生成利润表（income statements）的ChatGPT (ChatGPT)提示词，以及对应的模型生成结果。本数据集是研究"评估大语言模型（Large Language Model）从会计交易中生成利润表的涌现能力"的配套研究成果。 本数据集涵盖以下内容： 1. 与ChatGPT配套使用的结构化提示词集 2. 各试算平衡表场景对应的模型生成结果 3. 用于将生成结果与预期利润表进行比对的说明文档 本数据集旨在为人工智能（Artificial Intelligence）辅助会计与财务数据生成领域的可复现性研究、分析及后续拓展研究提供支撑。 引用建议： 恩德拉齐，特雷弗. (2025). 试算平衡交易与利润表分析数据集（大语言模型生成的提示词与输出结果）. Zenodo. DOI: https://doi.org/10.5281/zenodo.17095629