Get openness statistics of conferences from DBLP

calc_openness.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""
Get openness statistics of top conferences, motivated by
http://s3.eurecom.fr/~balzarot/notes/inbreeding/inbreeding.html

install: pip install requests matplotlib

usage example:

python calc_openness.py MobiCom MobiSys SenSys --plot --save_dir mobile

python calc_openness.py SOSP OSDI EuroSys USENIX FAST ASPLOS NSDI --plot

python calc_openness.py ACL EMNLP NAACL --plot --save_dir nlp

python calc_openness.py CVPR ICCV ECCV --plot --save_dir cv

"""

__author__ = "Qingqing Cao, https://awk.ai/, Twitter@sysnlp"
__copyright__ = "Copyright 2020, MIT LICENSE"

import argparse
import json
import os
from collections import defaultdict

import requests

_API_BASE = 'https://dblp.org/search/publ/api?'
_API_TEMPLATE = _API_BASE + 'q=/conf/{}/{}&format=json&h=1000'
author_record = defaultdict(dict)


# conference names can be found at https://dblp.org/db/conf/

def gen_author_flags(conf, year_str=''):
    years = set()
    api = _API_TEMPLATE.format(conf.lower(), year_str)
    response = requests.get(api)
    if not response.ok:
        print('request failed, try again or check api endpoint!')
        return -1

    data = json.loads(response.content)
    hits = data['result']['hits']
    all_hits = hits['hit']
    paper_authors_map = dict()
    for item in all_hits:
        info = item['info']
        authors = info.get('authors', None)
        if authors is None:
            # skip no author record
            continue
        venue = info.get('venue', None)
        if venue is None:
            continue
        year = int(info['year'])
        years.add(year)
        all_authors = authors['author']
        if isinstance(all_authors, dict):
            record_authors = [all_authors]
        else:
            assert isinstance(all_authors, list)
            record_authors = all_authors
        paper_key = info['key']
        paper_authors = set()
        for author in record_authors:
            author_id = author['@pid']
            paper_authors.add(author_id)
            author_years = author_record[author_id].get(conf, set())
            author_years.add(year)
            author_record[author_id][conf] = author_years
        if isinstance(venue, list):
            venue = venue[0]
        if conf in venue:  # only consider conf venue
            paper_authors_map[paper_key] = paper_authors
    return years, paper_authors_map


def main(args):
    conferences = args.conferences
    conf_old_papers = defaultdict(dict)
    conf_new_papers = defaultdict(dict)
    all_old = defaultdict(dict)
    all_new = defaultdict(dict)
    for conf in conferences:
        years, _ = gen_author_flags(conf)
        conf_years = sorted(list(years), reverse=True)
        for conf_year in conf_years:
            if conf_year < args.after_years:
                continue
            _, paper_authors_map = gen_author_flags(
                conf, conf_year)
            num_year_papers = len(paper_authors_map)
            if num_year_papers == 0:
                continue
            num_old_author_paper = 0
            for paper_key, paper_authors in paper_authors_map.items():
                is_old_author = False
                for paper_author in paper_authors:
                    # if one of the author published in previous years,
                    # it is old-author paper
                    author_years = author_record[paper_author][conf]
                    is_old_author = min(author_years) < conf_year
                    if is_old_author:
                        break
                num_old_author_paper += 1 if is_old_author else 0
            num_new_author_papers = num_year_papers - num_old_author_paper
            conf_old_papers[conf][conf_year] = num_old_author_paper
            conf_new_papers[conf][conf_year] = num_new_author_papers
            print(conf, conf_year, num_old_author_paper, num_new_author_papers)
            all_old[conf_year][conf] = num_old_author_paper
            all_new[conf_year][conf] = num_new_author_papers
    # conf stats for all years
    print('all combined:')
    x = []
    y = []
    for year in sorted(all_old.keys(), reverse=True):
        x.append(year)
        num_old = sum(all_old[year].values())
        num_new = sum(all_new[year].values())
        frac = num_new / (num_old + num_new)
        y.append(frac)
        print(year, frac, num_new, num_old)

    if args.plot:
        import matplotlib as mpl
        mpl.use('Agg')
        import matplotlib.pyplot as plt
        cnt = 0
        save_dir = args.save_dir
        os.makedirs(save_dir, exist_ok=True)
        for key in conf_new_papers.keys():
            old_data = conf_old_papers[key]
            new_data = conf_new_papers[key]
            x = []
            y = []
            for year in sorted(old_data.keys(), reverse=True):
                x.append(year)
                y.append(new_data[year] / (new_data[year] + old_data[year]))
            plt.figure(cnt)
            cnt += 1
            plt.plot(x, y, '-o')
            plt.xticks(x, rotation=45)
            plt.yticks([i / 10 for i in range(0, 11)])
            plt.ylim(0, 1)
            plt.grid(True)
            plt.title("{} Fraction of Papers from New Authors".format(key))
            plt.savefig(os.path.join(save_dir, '{}.png'.format(key)))
            plt.show()
        plt.figure(cnt)
        plt.plot(x, y, '-o')
        plt.xticks(x, rotation=45)
        plt.yticks([i / 10 for i in range(0, 11)])
        plt.ylim(0, 1)
        plt.grid(True)
        plt.title("Fraction of Papers from New Authors")
        plt.savefig(os.path.join(save_dir, 'combined.png'))
        plt.show()


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('conferences', type=str, nargs='+',
                        help='list of conferences separate by space')
    parser.add_argument("--plot", action="store_true",
                        help="if true, plot stats")
    parser.add_argument("--after_years", type=int, default=2000,
                        help="get data after the year, set to 0"
                             "to get all avaiable dblp data")
    parser.add_argument("--save_dir", type=str, default='.',
                        help="dir to save plot image")

    main(parser.parse_args())