Design And Implementation Of Corpus Annotation System Based On Crowdsourcing

In the field of Affective Computing, annotations on corpus and revision on information have been considered difficult problems for researchers. Since the work costs a lot of labor and time without large amounts of professional skills, researchers cannot focus on it all the time. Crowdsourcing is a strategic model, which utilizes personal interest and motivation to attract the public to provide solutions. It can gather talent, and take full advantage of the public ingenious creativity, thereby improving the working efficiency. We hope to take advantage of shining points of crowdsourcing, and solve the problem by collective wisdom based on the online platform. It utilizes external resources on the web to free researchers from the labor intensive and mechanical job and pay more their attention to research.Firstly, this paper overviews systematically and comprehensively the concept of crowdsourcing, What has been achieved at home and abroad, the existing problems as well as new trends based on reading the literature related to crowdsourcing at home and abroad. Then, we propose a research programs based on crowdsourcing, which is based on the theory of crowdsourcing. We propose annotation scheme and annotation specifications based on three annotation tasks within laboratory, and propose system-oriented solutions for two aspects of the most critical and most difficult in crowdsourcing system, incentive programs and quality control. It utilizes external resources on the web to resolve the problem of corpus annotation in the field of affective computing research.The system uses ASP.NET(C#) and SQL SERVER2008in the implementation and exploit AJAX and JAVASCRIPT to promote the user interface. Statistics based on initial operation of our system shows that our system can meet the annotation demand in our lab with friendly interface and much convenience.