hoannv / twitter-researcher

Identifying and Analyzing Researchers on Twitter

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Researchers on Twitter

This repository contains the data used in the paper Identifying and Analyzing Researchers on Twitter [PDF]. At the moment, this includes computer scientists, though an extension to other disciplines is planned.

  • The data can be cited as follows: Asmelash Teka Hadgu and Robert Jäschke. 2014. Identifying and Analyzing Researchers on Twitter. In Proceedings of the 6th Annual ACM Web Science Conference (WebSci '14). 23-30. ACM, New York, NY, USA. DOI: 10.1145/2615569.2615676
  • The data is also available on Zenodo: DOI
  • The data is licensed under a Creative Commons Attribution 4.0 International License Creative Commons License

Please help us verifying the data by participating in this survey and forwarding the link to other computer scientists. It's just two questions! We will publish the results in this repository.

Overview

This is the processing pipeline. The data for each step is provided below.

Overview of the processing pipeline

Seeds

We could find 170 Twitter accounts for 98 of the 268 conferences from the Wikipedia list of computer science conferences:

Area Acronym Area Conference Acronym Conference Name Twitter Screen Names
AI Artificial intelligence AAMAS International Conference on Autonomous Agents and Multiagent Systems aamas2012
AI Artificial intelligence BMVC British Machine Vision Conference bmvc2013, bmvc2014
AI Artificial intelligence COLT Conference on Learning Theory colt2010a
AI Artificial intelligence CVPR Conference on Computer Vision and Pattern Recognition cvpr
AI Artificial intelligence EACL Annual Conference of the European Chapter of the Association for Computational Linguistics eacl2014
... ... ... ... ...

Candidates

We could find 52678 candidates, i.e., users which follow, retweeted, or are followed by at least one of the seeds.

Matched Candidates

We could find a matching DBLP author profile for 9191 of the 52678 candidates.

ID Screen Name Real Name DBLP URL
1043459150 01Myers Michael Myers http://dblp.uni-trier.de/pers/hd/m/Myers:Michael.html
65845659 0xcharlie Charlie Miller http://dblp.uni-trier.de/pers/hd/m/Miller:Charlie.html
55857505 10705013 Nina Jeliazkova http://dblp.uni-trier.de/pers/hd/j/Jeliazkova:Nina.html
293990161 113134180 Wang Yao http://dblp.uni-trier.de/pers/hd/y/Yao:Wang.html
171917048 123456qwertz Anna Averbakh http://dblp.uni-trier.de/pers/hd/a/Averbakh:Anna.html
... ... ... ...

In a manual check of a random sample of 150 matched candidates, 109 (73%) of them were correct matches, 21 (14%) of them were wrong, and for 20 (13%) of them the experts could neither confirm nor confute the match. Still, many of the 41 non-positive candidates turned out to be computer scientists.

Screen Name Real Name DBLP URL Expert Judgement
777VJ Junghong Choi http://dblp.uni-trier.de/pers/hd/c/Choi:Junghong.html ✔️ true
afsanehdoryab Afsaneh Doryab http://dblp.uni-trier.de/pers/hd/d/Doryab:Afsaneh.html nan
akira Akira Yokokawa http://dblp.uni-trier.de/pers/hd/y/Yokokawa:Akira.html ✔️ true
albertobeta Alberto Betella http://dblp.uni-trier.de/pers/hd/b/Betella:Alberto.html ✔️ true
alexander_usoro Alexander James http://dblp.uni-trier.de/pers/hd/j/James:Alexander.html nan
... ... ... ...

Negative Examples

Screen Name Real Name Type
0127Kikki あきひろー random
0127Syuu 岩崎柊 random
02Ykss 瀬野尚子 random
0308Penny どぅー@たいち・8・ random
0555068668 ♥عبدالرحمن بن سعود♥ random
... ... ...

Features

Feature ID Type Description
location.exists Boolean true if location exists in user profile
has.url Boolean true if URL exists in user profile
url.has.tilde Boolean true if URL in user profile contains ~ (the tilde character)..
url.has.academic.domain Boolean true if URL in user profile contains academic top-level domain .edu or country code second-level domain .ac (e.g., .ac.uk)
has.bio Boolean true if user profile has bio
bio.has.pattern Boolean true if user bio contains at least one of the keywords: architect, assistant, associate, author, candidate, co-founder, cs, designer, developer, director, engineer, fellow, founder, geek, graduate, lecturer, manager, phd, ph.d, prof, professor, programmer, researcher, scientist, senior
language Boolean true if user language is English
default.profile.image Boolean true if user has a non-default profile picture
nof.tweets Numeric number of tweets in profile
nof.followers Numeric number of followers in profile
nof.friends Numeric number of friends in profile
friend.follower.ratio Numeric nof.friends/(nof.followers ratio + 0.01)
nof.self.tweets Numeric number of original tweets from the latest 3200 tweets at the time of crawl
nof.retweets Numeric number of retweets of a user from the latest 3200 tweets at the time of crawl
retweet.tweet.ratio Numeric nof.retweets to (nof.self.tweets + nof.retweets) ratio
nof.tweets.with.url Numeric number of tweets with at least one URL from the latest 3200 tweets at the time of crawl
fraction.of.tweets.with.url Numeric nof.tweets.with.url to (nof.self.tweets + nof.tweets.with.url) ratio
nof.conf.hashtags Numeric number of tweets with at least one bootstraped hashtag from the latest 3200 tweets at the time of crawl
nof.distinct.hashtags Numeric number of distinct hashtags used by the user in the latest 3200 tweets
nof.overall.hashtags Numeric number of tweets with hashtag(s) from the latest 3200 tweets at the time of crawl
fraction.of.distinct.hashtags Numeric nof.distinct.hashtags to nof.overall.hashtags ratio
nof.conf.mentions Numeric number of bootstrapped users mentioned in the latest 3200 tweets

Classification

Screen Name Real Name class
00002163 maria guerrero parra researcher
000Foofn999 Fofo000 000 non-researcher
001101110011001 tyo researcher
005nd andyr researcher
007Natty007 Nat researcher
... ... ...

About

Identifying and Analyzing Researchers on Twitter