Retrieve accepted paper metadata from ML/DL/NLP/CV/Robotics/Security/SE conferences. Uses the OpenReview API, web scraping, CVF Open Access, conf.researchr.org, Crossref, roboticsproceedings.org, and DBLP's SPARQL endpoint.
uv is a fast Python package manager used to manage dependencies.
# macOS / Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows (PowerShell)
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"After installing, restart your terminal so the uv command is available.
git clone https://github.com/brightjade/paper-explorer.git
cd paper-explorer
uv sync
source .venv/bin/activateThis creates a virtual environment and installs all required packages automatically. Activating it once lets you run ppr directly; re-run source .venv/bin/activate in each new shell.
The paper data is hosted as a GitHub Release asset. Run the setup script to download and extract it:
./setup.shThis downloads the latest data snapshot (~95 MB) and extracts it to data/.
Note: Requires either the GitHub CLI (
gh) orcurl. If you don't havegh, the script falls back tocurlautomatically.
Only needed if you want to crawl OpenReview conferences (ICLR, NeurIPS, ICML, COLM, CoRL). Other conferences are scraped from public websites and don't require authentication.
- Create a free account at https://openreview.net/signup
- Copy the example env file and fill in your credentials:
cp .env.example .env- Edit
.envwith your OpenReview username and password.
# Crawl one or more conferences
ppr crawl iclr_2025
ppr crawl iclr_2025 neurips_2025 icml_2025
# Enrich with citation counts and abstracts
ppr enrich iclr_2025
ppr enrich iclr_2025 neurips_2025 icml_2025
ppr enrich --all # refresh data/; archive years (4+ old) follow a monthly cohort, batch-only
# Every run also re-attempts papers no earlier run could match, and re-checks
# the papers it already matched against the record they are bound to.
# Validate paper counts against DBLP
ppr validate iclr_2025
ppr validate iclr_2025 neurips_2025 --tolerance 0.15
# Check tracked venues for accepted-paper lists we haven't registered yet
ppr discover
ppr discover --venue cvpr --venue iclr # limit the sweep to some venues
ppr discover --json # machine-readable, for the monthly workflow
# Build static JSON for web app
./build.shConference ID format: <venue>_<year> (e.g., iclr_2025). Selections indicate available paper tracks.
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| ICLR | oral, poster | oral, spotlight, poster | oral, spotlight, poster | oral, spotlight, poster |
| NeurIPS | oral, spotlight, poster | oral, spotlight, poster | oral, spotlight, poster | |
| ICML | main, spotlight | oral, spotlight, poster | oral, spotlight, poster | oral, poster |
| TMLR | main | main | main | main |
| AAAI | main | main | main | main |
| IJCAI | main | main | main | main |
NeurIPS also includes datasets_oral, datasets_spotlight, datasets_poster tracks.
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| ACL | main, findings, industry | main, findings, industry | main, findings | main, findings, industry |
| EMNLP | main, findings, industry | main, findings, industry | main, findings, industry | |
| NAACL | main, findings, industry | main, findings, industry | ||
| COLM | main | main | main | |
| EACL | main, findings, industry | main, findings | main, findings | |
| COLING | main, industry | main |
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| CVPR | main | main | main | main |
| ICCV | main | main | ||
| ECCV | main | |||
| WACV | main | main | main | main |
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| CoRL | oral, poster | main | oral, poster | |
| ICRA | main | main | main | main |
| IROS | main | main | main | |
| RSS | main | main | main | main |
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| USENIX Security | main | main | main | main |
| Conference | 2026 | 2025 | 2024 | 2023 |
|---|---|---|---|---|
| ICSE | main | main | main | main |
| FSE | main | main | main | main |
| ASE | main | main | main | main |
| ISSTA | main | main | main | main |
Earlier years come from DBLP, except where OpenReview keeps the presentation type. DBLP's proceedings volumes separate main, findings and industry papers but not orals.
| Venue | Years | Source | Labels |
|---|---|---|---|
| ICLR | 2013–2022 | DBLP 2013–2016, OpenReview 2017–22 | main (2013–2016); oral, poster 2017–19; oral, spotlight, poster 2020–22 |
| NeurIPS | 2013–2022 | DBLP 2013–2020, OpenReview 2021–22 | main (2013–2020); oral, spotlight, poster 2021; main 2022; datasets 2021–22 |
| ICML | 2013–2022 | DBLP 2013–2022 | main |
| AAAI | 2013–2022 | DBLP 2013–2022 | main |
| IJCAI | 2013, 2015–2022 | DBLP 2013, 2015–2022 | main |
| ACL | 2013–2022 | DBLP 2013–2022 | main; findings 2021–2022 |
| EMNLP | 2013–2022 | DBLP 2013–2022 | main; findings 2020–2022; industry 2022 |
| NAACL | 2013, 2015–2016, 2018–2019, 2021–2022 | DBLP 2013, 2015–2016, 2018–2019, 2021–2022 | main; findings 2022; industry 2018–2019, 2021–2022 |
| EACL | 2014, 2017, 2021 | DBLP 2014, 2017, 2021 | main |
| COLING | 2014, 2016, 2018, 2020, 2022 | DBLP 2014, 2016, 2018, 2020, 2022 | main; industry 2020 |
| CVPR | 2013–2022 | DBLP 2013–2022 | main |
| ICCV | 2013, 2015, 2017, 2019, 2021 | DBLP 2013, 2015, 2017, 2019, 2021 | main |
| ECCV | 2014, 2016, 2018, 2020, 2022 | DBLP 2014, 2016, 2018, 2020, 2022 | main |
| WACV | 2013–2022 | DBLP 2013–2022 | main |
| CoRL | 2017–2022 | DBLP 2017–2020, OpenReview 2021–22 | main (2017–2020); oral, poster 2021–22 |
| ICRA | 2013–2022 | DBLP 2013–2022 | main |
| IROS | 2013–2022 | DBLP 2013–2022 | main |
| RSS | 2013–2022 | DBLP 2013–2022 | main |
| USENIX Security | 2013–2022 | DBLP 2013–2022 | main |
| ICSE | 2013–2022 | DBLP 2013–2022 | main |
| FSE | 2013–2022 | DBLP 2013–2022 | main |
| ASE | 2013–2022 | DBLP 2013–2022 | main |
| ISSTA | 2013–2022 | DBLP 2013–2022 | main |
Four venue-years differ in scope from 2023 on. ICSE 2013 is one DBLP volume of 242 papers (its research track had about 85), so it also carries other tracks; from 2014 DBLP splits a companion volume off and the backfill excludes it. AAAI's DBLP volumes include student abstracts and demos, which the 2023+ crawl leaves out. FSE and ASE publish every track in one volume, so their 2013–2022 counts are higher than a research-track-only crawl would give.
Use the /survey command in Claude Code to generate a grounded literature survey from the accepted papers in this repository. Requires enriched data (papers_enriched.jsonl) for the target conferences.
# Specify conferences and year range
/survey I'm exploring efficient inference methods for large language models,
like speculative decoding and early exit strategies. Search NLP and ML
conferences from 2023-2025.
# Let it ask you for scope
/survey What papers exist on 3D object detection from point clouds?
# Target specific venues
/survey Find related work on code generation with LLMs. Search ICSE, FSE,
ASE, ACL, and EMNLP from 2023-2025.
The survey searches through real accepted papers, ranks by citation count, identifies datasets and benchmarks, and highlights research gaps. Output is saved to outputs/<topic>_<timestamp>.md.