UCT CS Research Document Archive

Crowdsourcing a Text Corpus is not a Game

Packham, Sean and Hussein Suleman (2015) Crowdsourcing a Text Corpus is not a Game. In Allen, Robert B, Jane Hunter and Marcia L. Zeng, Eds. Proceedings 17th International Conference on Asia-Pacific Digital Libraries, pages 225-234, Seoul, South Korea.

Full text available as:
PDF - Requires Adobe Acrobat Reader or other PDF viewer.

Abstract

Building language corpora for low resource languages such as South Africa’s isiXhosa is challenging because of limited digitized texts. Language corpora are needed for building information retrieval services such as search and translation and to support further online content creation. A novel solution was proposed to source original and relevant multilingual content by crowdsourcing translations via an online competitive game where participants would be paid for their contributions. Four experiments were conducted and the results support the idea that gamification by itself does not yield the widely expected benefits of increased motivation and engagement. We found that people do not volunteer without financial incentives, the form of payment does not matter, they would not continue contributing if the money is taken away and people preferred direct incentives and the possibility of incentives was not as strong a motivator.

EPrint Type:Conference Paper
Keywords:Crowdsourcing, Gamification, Translation, Language corpora, Information retrieval
Subjects:H Information Systems: H.4 INFORMATION SYSTEMS APPLICATIONS
J Computer Applications: J.5 ARTS AND HUMANITIES
H Information Systems: H.3 INFORMATION STORAGE AND RETRIEVAL
ID Code:1064
Deposited By:Suleman, Hussein
Deposited On:26 January 2016