The repository for the paper "A Formative Study on Designing Accurate and Natural Figure Captioning Systems" (CHI LBW '20).
- Resources
- Contacts
corpus is the directory that contains a directory named figures and a file named IELTS-crawled-corpus.json.
The figures directory contains figures in the .gif or .png format.
The IELTS-crawled-corpus.json contains 95 model answers, where the last 20 answers are commonly annotated by all annotators.
annotations is the directory that contains the annotation .json1 files from the nine annotators. Each line is the annotation for one model answer.
For any question or feedback, please reach Xin at xinq@umd.edu!