SOTAVerified

SciCap: Generating Captions for Scientific Figures

2021-10-22Findings (EMNLP) 2021Code Available1· sign in to hype

Ting-Yao Hsu, C. Lee Giles, Ting-Hao 'Kenneth' Huang

Code Available — Be the first to reproduce this paper.

Reproduce

Code

Abstract

Researchers use figures to communicate rich, complex information in scientific papers. The captions of these figures are critical to conveying effective messages. However, low-quality figure captions commonly occur in scientific articles and may decrease understanding. In this paper, we propose an end-to-end neural framework to automatically generate informative, high-quality captions for scientific figures. To this end, we introduce SCICAP, a large-scale figure-caption dataset based on computer science arXiv papers published between 2010 and 2020. After pre-processing - including figure-type classification, sub-figure identification, text normalization, and caption text selection - SCICAP contained more than two million figures extracted from over 290,000 papers. We then established baseline models that caption graph plots, the dominant (19.2%) figure type. The experimental results showed both opportunities and steep challenges of generating captions for scientific figures.

Tasks

Benchmark Results

DatasetModelMetricClaimedVerifiedStatus
SCICAPCNN+LSTM (Vision only, First sentence)BLEU-40.02Unverified
SCICAPCNN+LSTM (Text only, First sentence)BLEU-40.02Unverified
SCICAPCNN+LSTM (Text only, Single-Sent Caption)BLEU-40.02Unverified
SCICAPCNN+LSTM (Vision only, Single-Sent Caption)BLEU-40.02Unverified
SCICAPCNN+LSTM (Vision + Text, First sentence)BLEU-40.02Unverified
SCICAPCNN+LSTM (Vision + Text, Single-Sent Caption)BLEU-40.02Unverified
SCICAPCNN+LSTM (Vision only, Caption w/ <=100 words)BLEU-40.02Unverified
SCICAPCNN+LSTM (Vision + Text, Caption w/ <=100 words)BLEU-40.02Unverified
SCICAPCNN+LSTM (Text only, Caption w/ <=100 words)BLEU-40.02Unverified

Reproductions