Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略

Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略

ID:18440539

大小:1.97 MB

页数:10页

时间:2018-09-18

Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略_第1页
Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略_第2页
Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略_第3页
Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略_第4页
Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略_第5页
资源描述:

《Strategies for Crowdsourcing Social Data Analysis众包社会数据分析策略》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、StrategiesforCrowdsourcingSocialDataAnalysisWesleyWillett?,JeffreyHeer†,ManeeshAgrawala??ComputerScienceDivision,UniversityofCalifornia,Berkeleyfwillettw,maneeshg@cs.berkeley.edu†ComputerScienceDepartment,StanfordUniversityjheer@cs.stanford.eduABSTRACTelicitinghigh

2、-qualityexplanationsofthedatarequiresseed-Web-basedsocialdataanalysistoolsthatrelyonpublicdis-ingthediscussionwithprompts,examples,andotherstartingcussiontoproducehypothesesorexplanationsofpatternspointstoencouragecontributions[8,32].andtrendsindatararelyyieldhigh-

3、qualityresultsinprac-Outsidethelab,inreal-worldweb-baseddeployments,thetice.Crowdsourcingoffersanalternativeapproachinwhichvastmajorityofthevisualizationsinthesesocialdataanal-ananalystpaysworkerstogeneratesuchexplanations.Yet,ysistoolsyieldverylittlediscussion.Eve

4、nfewervisualiza-askingworkerswithvaryingskills,backgroundsandmoti-tionselicithigh-qualityanalyticalexplanationsthatareclear,vationstosimply“Explainwhyachartisinteresting”canplausible,andrelevanttoaparticularanalysisquestion.resultinirrelevant,unclearorspeculativeex

5、planationsofvariablequality.Toaddresstheseproblems,wecontributeWerecentlysurveyedtheManyEyeswebsiteandfoundthatsevenstrategiesforimprovingthequalityanddiversityoffrom2006to2010,userspublished162,282datasetsbutworker-generatedexplanations.Ourexperimentsshowthatgener

6、atedonly77,984visualizationsandleftjust15,464com-using(S1)feature-orientedprompts,providing(S2)goodex-ments.Wethenrandomlysampled100ofthevisualizationsamples,andincluding(S3)referencegathering,(S4)chartcontainingcommentsandfoundthatjust11%ofthecom-reading,and(S5)an

7、notationsubtasksincreasesthequalitymentsincludedaplausiblehypothesisorexplanationfortheofresponsesby28%forUSworkersand196%fornon-datainthechart.Thelowlevelofcommentingmayrepre-USworkers.Feature-orientedpromptsimproveexplanationsentashortageofviewersormaybeduetolurk

8、ing–acom-qualityby69%to236%dependingontheprompt.Wealsomonwebphenomenoninwhichvisitorsexploreandreadshowthat(S6)pre-annotatingchartscanfocusworker

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。