soda-lmu/tweet-annotation-sensitivity-1

该数据集来源于一个关于推文标注敏感性的实验。实验从Davidson等人的研究中选取了20条推文，并招募了1000名Prolific工人对这些推文进行标注。标注者在六种不同的实验条件下进行标注，并收集了标注者的人口统计信息和任务相关数据。数据集包含了推文的标注结果、标注者的背景信息以及任务完成的时间等详细数据。

原始信息汇总

数据集概述

数据集描述

该数据集包含20条推文，这些推文是从Davidson et al. (2017)的研究中抽取的，并根据多数投票的类别和分歧程度进行了分层抽样。数据集用于情感分类和仇恨言论检测任务。

数据收集

招募了1000名Prolific工作者对每条推文进行标注，标注者被随机分配到六种实验条件之一，并被要求对推文进行仇恨言论、冒犯性语言或两者都不是的标注。此外，还收集了标注者的多种人口统计变量（如年龄和性别）以及一些辅助数据（如任务总时长、每个屏幕的时长）。

推文示例

数据集包含以下20条推文，用户名已进行哈希处理：

Tweet ID	Tweets Hashed
1	Jihadi John has struck again. Prayers with the victims families. What is the #bounty on His head? Time 4 him 2 meet allah
2	RT @#####: Seriously, Florida. A vote for Charlie Christ is a vote for taxes and ebola.
3	Islamic Jihadis run away to Pakistan @##### @##### #UniteBlue
4	At this rate, Id cheer for the awful New York Yankees over the St. Louis Cardinals.
5	Allergies have my eyes all red and chink eyed, I look like Im stoned
6	Not protected in the state of Utah. There is no fear of queer couples forcing churches to marry them. They do not legally have to do so.
7	RT @#####: Boss Niggas Fuk wit Boss bitches Vice versa..
8	RT @#####: ever wanna text somebody and be like bitch die I hope you dont wake up you low down dirty muthafxcka&#128520
9	@##### @##### @##### shes a right twat, Id tell you, honestly Sky! Honestly &#128520
10	So yall give farwells to Lulu yet did all but shit on Kris? Ok bitches... ok
11	RT @#####: Dis bitch wanna smoke all our weed
12	Shittin on all my old hoes
13	Getting hit off a cunt thats about 6 foot sideways and a cunt that was in the army lol, i dini care, yous are fuckin laughable
14	When it rains its good "go to the mall weather" all the coons be at home
15	Dont mind black people.....hate niggers!
16	My eyes feel so chink eyed
17	"From now on, I will call all radical MUSLIMS niggers!It is very fitting and it is racist! I AM RACIST AGAINST ANYONE WHO HATES AMERICA!"""
18	RT @#####: Fucking chinks in Clash of Clans always fuck me up. Bitch ass kamikaze faggots.
19	@##### ya I know all the slang Im racist I h8 porch monkies
20	RT @#####: Stupid fucking nigger LeBron. You flopping stupid jungle bunny monkey faggot.

数据字典

数据集包含以下列：

Column Name	Description	Type
id	annotator ID	integer
age	Age	integer
gender	Gender<br> 1: Female<br>2: Male<br> 3: Something Else<br> 4: Prefer not to say<br>	factor
afam	African-American<br> 0: No<br> 1: Yes	binary
asian	Asian-American<br> 0: No<br> 1: Yes	binary
hispanic	Hispanic<br> 0: No<br> 1: Yes	binary
white	White<br> 0: No<br> 1: Yes	binary
race_other	Other race/ethnicity<br> 0: No<br> 1: Yes	binary
race_not_say	Prefer not to say race/ethnicity<br> 0: No<br> 1: Yes	binary
education	Highest educational attainment<br> 1: Less than high school<br>2: High school<br> 3: Some college<br> 4: College graduate<br> 5: Masters degree or professional degree (Law, Medicine, MPH, etc.) <br> 6: Doctoral degree (PhD, DPH, EdD, etc.)	factor
sexuality	Sexuality<br> 1: Gay or Lesbian<br>2: Bisexual<br> 3: Straight<br> 4: Something Else<br>	factor
english	English first language? <br> 0: No<br> 1: Yes	binary
tw_use	Twitter Use <br> 1: Most days<br>2: Most weeks, but not every day<br> 3: A few times a month<br> 4: A few times a year<br> 5: Less often <br> 6: Never	factor
social_media_use	Social Media Use<br> 1: Most days<br>2: Most weeks, but not every day<br> 3: A few times a month<br> 4: A few times a year<br> 5: Less often <br> 0: Never	factor
prolific_hours	Prolific hours worked last month	integer
task_fun	Coding work was: fun<br> 0: No<br> 1: Yes	binary
task_interesting	Coding work was: interesting<br> 0: No<br> 1: Yes	binary
task_boring	Coding work was: boring<br> 0: No<br> 1: Yes	binary
task_repetitive	Coding work was: repetitive<br> 0: No<br> 1: Yes	binary
task_important	Coding work was: important<br> 0: No<br> 1: Yes	binary
task_depressing	Coding work was: depressing<br> 0: No<br> 1: Yes	binary
task_offensive	Coding work was: offensive<br> 0: No<br> 1: Yes	binary
another_tweettask	Likelihood to do another Tweet related task<br> not at all: Not at all likely<br> somewhat: Somewhat likely<br> very: Very likely	factor
another_hatetask	Likelihood to do another Hate Speech related task<br> not at all: Not at all likely<br> somewhat: Somewhat likely<br> very: Very likely	factor
page_history	Order in which annotator saw pages	character
date_of_first_access	Datetime of first access	datetime
date_of_last_access	Datetime of last access	datetime
duration_sec	Task duration in seconds	integer
version	Version of annotation task <br> A: Version A<br>B: Version B<br> C: Version C<br> D: Version D<br> E: Version E<br> F: Version F	factor
tw1-20	Label assigned to Tweet 1-20<br> hate speech: Hate Speech<br> offensive language: Offensive Language<br> neither: Neither HS nor OL <br> NA: Missing or "dont know"	factor
tw_duration_1-20	Annotation duration in milliseconds Tweet 1-20	numerical
num_approvals	Prolific data: number of previous task approvals of annotator	integer
num_rejections	Prolific data: number of previous task rejections of annotator	integer
prolific_score	Annotator quality score by Prolific	numerical
countryofbirth	Prolific data: Annotator country of birth	character
currentcountryofresidence	Prolific data: Annotator country of residence	character
employmentstatus	Prolific data: Annotator Employment Status<br> Full-timePart-time<br> Unemployed (and job-seeking)<br> Due to start a new job within the next month<br> Not in paid work (e.g. homemaker, retired or disabled)<br> Other<br> DATA EXPIRED	factor
firstlanguage	Prolific data: Annotator first language	character
nationality	Prolific data: Nationality	character
studentstatus	Prolific data: Student status<br> Yes<br> No <br> DATA EXPIRED	factor