Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2507.02212 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914447102050304 |
|---|---|
| author | Kawada, Takuro Kitada, Shunsuke Nemoto, Sota Iyatomi, Hitoshi |
| author_facet | Kawada, Takuro Kitada, Shunsuke Nemoto, Sota Iyatomi, Hitoshi |
| contents | Graphical Abstracts (GAs) play a crucial role in visually conveying the key findings of scientific papers. Although recent research increasingly incorporates visual materials such as Figure 1 as de facto GAs, their potential to enhance scientific communication remains largely unexplored. Designing effective GAs requires advanced visualization skills, hindering their widespread adoption. To tackle these challenges, we introduce SciGA-145k, a large-scale dataset comprising approximately 145,000 scientific papers and 1.14 million figures, specifically designed to support GA selection and recommendation, and to facilitate research in automated GA generation. As a preliminary step toward GA design support, we define two tasks: 1) Intra-GA Recommendation, identifying figures within a given paper well-suited as GAs, and 2) Inter-GA Recommendation, retrieving GAs from other papers to inspire new GA designs. Furthermore, we propose Confidence Adjusted top-1 ground truth Ratio (CAR), a novel recommendation metric for fine-grained analysis of model behavior. CAR addresses limitations of traditional rank-based metrics by considering that not only an explicitly labeled GA but also other in-paper figures may plausibly serve as GAs. Benchmark results demonstrate the viability of our tasks and the effectiveness of CAR. Collectively, these establish a foundation for advancing scientific communication within AI for Science. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_02212 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SciGA: A Comprehensive Dataset for Designing Graphical Abstracts in Academic Papers Kawada, Takuro Kitada, Shunsuke Nemoto, Sota Iyatomi, Hitoshi Computer Vision and Pattern Recognition Computation and Language Machine Learning Graphical Abstracts (GAs) play a crucial role in visually conveying the key findings of scientific papers. Although recent research increasingly incorporates visual materials such as Figure 1 as de facto GAs, their potential to enhance scientific communication remains largely unexplored. Designing effective GAs requires advanced visualization skills, hindering their widespread adoption. To tackle these challenges, we introduce SciGA-145k, a large-scale dataset comprising approximately 145,000 scientific papers and 1.14 million figures, specifically designed to support GA selection and recommendation, and to facilitate research in automated GA generation. As a preliminary step toward GA design support, we define two tasks: 1) Intra-GA Recommendation, identifying figures within a given paper well-suited as GAs, and 2) Inter-GA Recommendation, retrieving GAs from other papers to inspire new GA designs. Furthermore, we propose Confidence Adjusted top-1 ground truth Ratio (CAR), a novel recommendation metric for fine-grained analysis of model behavior. CAR addresses limitations of traditional rank-based metrics by considering that not only an explicitly labeled GA but also other in-paper figures may plausibly serve as GAs. Benchmark results demonstrate the viability of our tasks and the effectiveness of CAR. Collectively, these establish a foundation for advancing scientific communication within AI for Science. |
| title | SciGA: A Comprehensive Dataset for Designing Graphical Abstracts in Academic Papers |
| topic | Computer Vision and Pattern Recognition Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2507.02212 |