ICLR 2024 Stats

iclr_2024_stats.py

import collections
import json
import os
import openreview

with open('creds.json') as f:
    creds = json.loads(f.read())

cache_file = 'iclr_2024_submissions.json'

# API V2
client = openreview.api.OpenReviewClient(
    baseurl='https://api2.openreview.net',
    username=creds['username'],
    password=creds['password'],
)

if not os.path.exists(cache_file):
    venue_id = 'ICLR.cc/2024/Conference'
    venue_group = client.get_group(f'{venue_id}')
    submission_name = venue_group.content['submission_name']['value']
    submissions = client.get_all_notes(invitation=f'{venue_id}/-/{submission_name}')
    data = [x.to_json() for x in submissions]
    with open(cache_file, 'w') as f:
        f.write(json.dumps(data))
else:
    with open(cache_file) as f:
        data = json.loads(f.read())

c_author = collections.Counter()
c_first_author = collections.Counter()
c_venue = collections.Counter()
author_to_paper = collections.defaultdict(list)

def fix_venue(venue):
    # ICLR 2024 Conference Withdrawn Submission: 1615
    # ICLR 2024 poster: 1807
    # Submitted to ICLR 2024: 3476
    # ICLR 2024 Conference Desk Rejected Submission: 53
    # ICLR 2024 spotlight: 367
    # ICLR 2024 oral: 86
    if venue == 'ICLR 2024 Conference Withdrawn Submission':
        return 'Withdrawn'
    elif venue in ('ICLR 2024 poster', 'ICLR 2024 spotlight', 'ICLR 2024 oral'):
        return 'Accepted'
    elif venue == 'Submitted to ICLR 2024':
        return 'Rejected'
    elif venue == 'ICLR 2024 Conference Desk Rejected Submission':
        return 'Desk Rejected'
    else:
        raise ValueError(f'Unknown venue: {venue}')

for x in data:
    authors = x['content']['authors']['value']
    venue = x['content']['venue']['value']
    for author in authors:
        author_to_paper[author].append(x)
    c_author.update(authors)
    c_first_author[authors[0]] += 1
    venue = fix_venue(venue)
    c_venue[venue] += 1

def get_author_paper_stats(author, author_to_paper):
    papers = author_to_paper[author]
    c_venues = collections.Counter([fix_venue(x['content']['venue']['value']) for x in papers])
    return c_venues

# Summarize author stats.
print('Top 30 authors:')
for author, count in c_author.most_common(30):
    this_c_venues = get_author_paper_stats(author, author_to_paper)
    print(f'{author}: {count} --- {this_c_venues}')

# Summarize first author stats.
print('Top 30 first authors:')
for author, count in c_first_author.most_common(30):
    this_c_venues = get_author_paper_stats(author, author_to_paper)
    print(f'{author}: {count} --- {this_c_venues}')

# Summarize venue stats.
print('Venues:')
for venue, count in c_venue.items():
    print(f'{venue}: {count}')