Reproduce “Towards Natural Language Interfaces for Data Visualization: A Survey” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Leixian Shen, Enya Shen, Yuyu Luo, Xiaocong Yang, Xuming Hu, Xiongshuai Zhang, Zhiwei Tai, Jianmin Wang

Researchers from Tsinghua University provide a comprehensive survey of Visualization-oriented Natural Language Interfaces (V-NLIs), systematically categorizing existing systems across an extended seven-stage visualization pipeline. The review consolidates two decades of development, identifying current limitations and outlining future research directions, especially considering advancements in natural language processing.

Key insights

  • V-NLI research has seen significant progress, particularly in an "outbreak stage" since 2018, driven by advancements in NLP, yet current systems still largely rely on general NLP toolkits and rule-based approaches for query interpretation and task inference.
  • Key challenges lie in enhancing human interaction (e.g., ambiguity widgets, multimodal input) and dialogue management (e.g., co-reference resolution, conversational context) to improve usability and analytical flow.
  • Future research should prioritize integrating domain knowledge, leveraging advanced NLP models for deeper semantic understanding, developing innovative presentation methods, and creating large-scale, high-quality V-NLI datasets.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.