nvBench 2.0
收藏资源简介:
nvBench 2.0是由香港科技大学创建的一个自然语言到可视化任务的数据集,包含7878条自然语言查询和24076个相应的可视化结果,这些数据来源于780个表格,覆盖153个领域。该数据集通过控制模糊注入管道构建,能够生成具有多个有效解释的模糊查询。每个查询-可视化对都配有推理路径,解释模糊是如何解决的,确保了解释的可解释性。
nvBench 2.0 is a natural language to visualization task dataset developed by The Hong Kong University of Science and Technology. It consists of 7,878 natural language queries and 24,076 corresponding visualization results, which are derived from 780 tables and span 153 domains. This dataset is built through a controlled ambiguity injection pipeline, enabling the generation of ambiguous queries that admit multiple valid interpretations. Each query-visualization pair is accompanied by a reasoning path that explains how the ambiguity is resolved, thus ensuring the interpretability of the explanations.

- 1nvBench 2.0: A Benchmark for Natural Language to Visualization under Ambiguity香港科技大学(HKUST) · 2025年



