map2seq

Data

The dataset consits of 7,672 Natural Language Landmark Navigation Instructions and corresponding route paths in OpenStreetMap. The data collection is described in the paper by (Schumann & Riezler, 2021), please cite it if you use the data.
Visualization of the routes in the dataset and a demo of the model trained on it can be found here.

Download

Link: map2seq_v1-0.zip / mirror [133 MB]

Publication

  1. Raphael Schumann and Stefan Riezler
    Generating Landmark Navigation Instructions from Maps as a Graph-to-Text Problem
    Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics (ACL), Bangkok, Thailand (virtual), 2021
    @inproceedings{schumann2021,
      author = {Schumann, Raphael and Riezler, Stefan},
      title = {Generating Landmark Navigation Instructions from Maps as a Graph-to-Text Problem},
      journal = {Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics},
      journal-abbrev = {ACL},
      year = {2021},
      city = {Bangkok},
      country = {Thailand (virtual)},
      url = {https://arxiv.org/abs/2012.15329}
    }