license: odc-by
pretty_name: Asta AI Citation Logs
configs:
- config_name: default
data_files:
- split: all_time
path: 2025-10-27/sqa_citation_ranking_all_time.parquet
- split: last_week
path: 2025-10-27/sqa_citation_ranking_last_week.parquet
- split: last_month
path: 2025-10-27/sqa_citation_ranking_last_month.parquet
- config_name: '2025-10-07'
data_files:
- split: all_time
path: 2025-10-07/sqa_citation_ranking_all_time.parquet
- split: last_week
path: 2025-10-07/sqa_citation_ranking_last_week.parquet
- split: last_month
path: 2025-10-07/sqa_citation_ranking_last_month.parquet
- config_name: '2025-10-20'
data_files:
- split: all_time
path: 2025-10-20/sqa_citation_ranking_all_time.parquet
- split: last_week
path: 2025-10-20/sqa_citation_ranking_last_week.parquet
- split: last_month
path: 2025-10-20/sqa_citation_ranking_last_month.parquet
- config_name: '2025-10-27'
data_files:
- split: all_time
path: 2025-10-27/sqa_citation_ranking_all_time.parquet
- split: last_week
path: 2025-10-27/sqa_citation_ranking_last_week.parquet
- split: last_month
path: 2025-10-27/sqa_citation_ranking_last_month.parquet
Dataset Summary
This dataset tracks which scientific papers are most often cited by Asta, an agentic research platform that uses retrieval-augmented generation (RAG) to answer scientific questions. Each record is a paper cited by Asta's Summarize Literature tool, ranked by the number of times the system cited that paper. Across more than 113,000 user queries, we track 4M citations to over 2M distinct papers. By making this data public, we aim to create a transparent, trackable measure of which research most directly powers AI-generated answers—helping ensure that scientific contributions are visible and credited in the AI era.
Weekly updates reflect ongoing usage patterns as Asta continues to evolve. We invite researchers, bibliometricians, and AI developers to explore citation dynamics across fields, assess how AI systems surface influential work, and help build a future where credit and accountability are integral to AI-assisted discovery.
The most recent update to the data can always be retrieved using the 'latest' config:
dataset = load_dataset("allenai/asta-summary-citation-counts", "latest")
Older checkpoints can be retrieved by date. Eg:
dataset = load_dataset("allenai/asta-summary-citation-counts", "2025-10-07")
Column Descriptions
| Field Name | Description |
|---|---|
corpus_id |
Unique identifier for the paper from Semantic Scholar |
title |
Title of the paper |
sqa_citation_rank |
Overall rank of the paper in terms of unique citation counts across queries on Asta Literature Summarizer |
sqa_citation_count_queries |
Unique citation counts of the paper across queries that powers its sqa_citation_rank |
sqa_citation_count_total_citations |
Total citation counts of the paper across queries (A paper can be cited multiple times in the answer report to a query) |
authors |
Comma separated string of paper authors |
venue |
Publishing venue/conference/journal of the paper |
year |
Year of publishing of the paper |
s2FieldsOfStudy |
Academic field of study categories assigned to the paper in Semantic Scholar by their classifier. The possible fields are: Computer Science, Medicine, Chemistry, Biology, Materials Science, Physics, Geology, Psychology, Art, History, Geography, Sociology, Business, Political Science, Economics, Philosophy, Mathematics, Engineering, Environmental Science, Agricultural and Food Sciences, Education, Law, and Linguistics. |
Dataset Details
- Dataset name: Asta Summary Citation Counts
- Maintainer: Allen Institute for AI (AI2)
- License and Use: This dataset is licensed under ODC-BY. It is intended for research and educational use in accordance with Ai2's Responsible Use Guidelines.
- Update frequency: Weekly
- Source platform: Asta (https://asta.ai)
- System Paper: Ai2 Scholar QA: Organized Literature Synthesis with Attribution
- System Code: ai2-scholarqa-lib
- Primary use cases: bibliometrics, AI transparency, citation dynamics, evaluation of retrieval-augmented generation systems