AI4SciSci 2025

About the 2^nd Workshop on AI4SciSci About AI4SciSci

Over the past years, the field of artificial intelligence (AI) has witnessed significant advancements in various fields, which have further accelerated discoveries, innovations, and breakthroughs in many interdisciplinary fields, including the Science of Science, which relies on big data to unveil the reproducible patterns that govern individual scientific careers and the workings of science. Recently, the Science of Science has made promising progress by adopting AI techniques into big data, providing deep and interesting insights into new and long-standing Science of Science problems. Here, we present the second International Workshop on Artificial Intelligence for the Science of Science (AI4SciSci) to engage the AI and Science of Science communities in further promoting research and discussion on frontier topics in this field.

Call for Papers

This workshop is aimed at bringing together researchers from the areas of data mining (DM), artificial intelligence (AI), and Science of Science (SciSci). We expect to encourage an exchange of ideas and perceptions through the workshop, focusing on novel research directions, challenges, and techniques in the intersectional areas of AI and DM for SciSci. We welcome papers on topics of interest that include, but are not limited to:

Scholarly Graph Analytics
- Citation networks and their applications
- Social networks and their applications
- Knowledge graphs and their applications
Scholarly Data Mining
- Innovation/novelty detection and prediction
- Scholarly collaboration and mobility prediction
- Career trajectory modeling
- Numerical modeling of scientific development
- Causal inference and reasoning for SciSci
- Streaming data mining and applications for SciSci
Scientific Document Processing and Understanding
- Scientific document parsing and applications
- Scientific metadata extraction and applications
- Scientific claim extraction and verification
- Scientific tabular data structure recognition and understanding
- Scientific figure/chart extraction and understanding
- Scientific document summarization and topic modeling
Computational Social Sciences
- Scientific reproducibility, replicability, repeatability, and generalizability
- Innovation automation with AI
- Scientific goods for AI and ethical implications
- Socio-economic effects on scientific progress and innovation
- Scientific misinformation and disinformation
- Scientific impact quantification using metrics outside of academia (e.g., patents, technology transfer to industry and revenue)

Originality

The workshop welcomes original AND published work. The authors of published work are not obligated to rewrite it using the templates and there is not any page limit. However, only original works will be published.

Double-Blind Review and Anonymity Guidelines

Ai4SciSci adopts the same double-blind review policy as JCDL. Submissions must not include any information that could identify the authors, such as names, institutional affiliations, acknowledgments, or references to prior work written in the first person. Authors should refer to their own prior work in the third person (e.g., “Prior work by Smith et al. (2020)” rather than “In our previous work…”). If authors share supplementary materials (e.g., code, data, models), these must be anonymized to avoid revealing author identities. Authors are encouraged to use anonymous hosting platforms such as anonymous.4open.science.

Submissions

All submissions must be written in English, following the CEUR workshop proceedings style.

Paper submission link: EasyChair

Submission Length

Extended Abstracts (up to 2 pages)
Short Papers (up to 4 pages)
Full Papers (up to 8 pages)

** References and appendices are not counted towards the page limits.

All accepted papers that report original work will be published at the EPiC series by easychair.

Important Dates

Call for Papers: Friday, October 03^rd, 2025
Paper Submission Deadline: ~~October 24^th, 2025~~ Friday, November 07^th, 2025
Author Notification: ~~November 14^th, 2025~~ Friday, November 21^st, 2025
Camera-Ready Deadline: ~~November 28^th, 2025~~ Friday, December 07^th, 2025
Registration: Link
Workshop date: Monday, December 15^th, 2025

Program

Monday, December 15^th, 2025 (EST, New York)

EST (New York)	MST (Boulder)	CET (Berlin)	Beijing (Dec. 16^th)	Duration (minutes)	Type	PaperID.Title	Presenter
11:00 AM	9:00 AM	5:00 PM	12:00 AM	15	Opening	Opening remark	Workshop chairs
11:15 AM	9:15 AM	5:15 PM	12:15 AM	60	Keynote	Keynote speech	Dr. Daniel Acuña
12:15 PM	10:15 AM	6:15 PM	1:15 AM	20	Long	8. Tracing Research Inequality in NLP: How Resource Disparities Shape Topic Trends and Diffusion	Lizhen Liang
12:35 PM	10:35 AM	6:35 PM	1:35 AM	20	Published	10. Scientific productivity and practice in the era of Large Language Models	Keigo Kusumegi
12:55 PM	10:55 AM	6:55 PM	1:55 AM	20	Long	1. Humans vs. LLMs on Open Domain Scientific Claim Verification: A Baseline Study	Benjamin Curtis
1:15 PM	11:15 AM	7:15 PM	2:15 AM	15	Short	13. Quantifying Contextual Hallucinations in NLP Research Papers Before and After the LLM Era	Adiba Ibnat Hossain
Break
4:00 PM	2:00 PM	10:00 PM	5:00 AM	20	Short	9. A Case For Clarity: Knowledge Engineering And Its Evolving Role In Medical Sciences	Henri Van Overmeire
4:20 PM	2:20 PM	10:20 PM	5:20 AM	15	Abstract	5. Applying LLM to Library Metadata: Mapping Geography and Language in the Library of Congress Collection	Hongyu Zhou
4:35 PM	2:35 PM	10:35 PM	5:35 AM	15	Abstract	3. Geography of Medical Knowledge: Scientific Focus, Disease Burden, and Research Response	Hongyu Zhou
4:50 PM	2:50 PM	10:50 PM	5:50 AM	20	Long	7. A Gradio-Based Toolkit for Remote Sensing Data Fusion Literature	Jiaxin Du
5:10 PM	3:10 PM	11:10 PM	6:10 AM	20	Published	4. Transforming Role Classification in Scientific Teams using LLMs	Wonduk Seo
5:30 PM	3:30 PM	11:30 PM	6:30 AM	15		Ending Remark	Workshop chairs
5:45 PM	3:45 PM	11:45 PM	6:45 AM			Workshop Ends

Keynote Speaker

Dr. Daniel Acuña

Associate Professor

University of Colorado Boulder

Title: Estimating the predictability of questionable open-access journals

Questionable journals threaten global research integrity, yet manual vetting can be slow and inflexible. Here, we explore the potential of artificial intelligence (AI) to systematically identify such venues by analyzing website design, content, and publication metadata. Evaluated against extensive human-annotated datasets, our method achieves practical accuracy and uncovers previously overlooked indicators of journal legitimacy. By adjusting the decision threshold, our method can prioritize either comprehensive screening or precise, low-noise identification. At a balanced threshold, we flag over 1000 suspect journals, which collectively publish hundreds of thousands of articles, receive millions of citations, acknowledge funding from major agencies, and attract authors from developing countries. Error analysis reveals challenges involving discontinued titles, book series misclassified as journals, and small society outlets with limited online presence, which are issues addressable with improved data quality. Our findings demonstrate AI’s potential for scalable integrity checks, while also highlighting the need to pair automated triage with expert review.