Opinion Label: A Gamified Crowdsourcing System for Sentiment

• Douglas Cirqueira UFPA
• Lucas Vinícius UFPA
• Márcia Pinheiro UFPA
• Antônio Jacob Junior UEMA
• Fábio Lobato UFOPA
• Ádamo Santana UFPA

Abstract

Online Social Networks are more than a source of large amounts of data, but a potential treasure for companies and institutions that seek information about how they are perceived by their audience. Sentiment Analysis (SA) is a technique that allows the automatic mining of opinions, which can be applied in this context. However, such approach faces many challenges reported in the state of art. Among those, there is the lack of labeled datasets for sentiment classification algorithms. This work presents a platform for SA annotation, with crowdsourcing (CS) and gamification principles, aiming to solve this challenge. Differential aspects of this tool are the possibilities of labeling data for fields related to SA, such as emotion and subjectivity tagging, besides other common tasks in the area, planned to be added in a future version of the tool.

Keywords: Sentiment Analysis, crowdsourcing, gamification, annotation, text tagging, text mining