Khatua, Aparup, Cambria, Erik, Khatua, Apalak, and Chaturvedi, Iti (2017) Let's chat about Brexit! A politically-sensitive dialog system based on Twitter data. In: Proceedings of the IEEE International Conference on Data Mining Workshops. pp. 393-398. From: ICDMW 2017: 17th IEEE International Conference on Data Mining Workshops, 18-21 November 2017, New Orleans, LA, USA.

PDF (Published Version) - Published Version Restricted to Repository staff only

4

Abstract

Data scientists are exploring various semi-supervised learning methods to build conversational agents - commonly known as chatterbot. This paper investigates various issues related to a political chatterbot where human agents are politically opinionated. Here, understanding the latent intent of human agent is crucial for developing an efficient political chatterbot. We set our study in the context of 2016 Brexit referendum. We argue that employing a subjectivity detector and an emotion analyzer, in addition to the keyword based topic detector, enhances the intent detection process. Next, we discuss the importance of maintaining political neutrality. To maintain its neutrality, a chatterbot needs to disassociate itself from a politically opinionated response. This can be achieved by associating a response with a user or a set of users. Nowadays, the Twitter platform provides an enormous amount of user-generated contents for various socio-economic events. Hence, we have considered tweet feeds for developing the overall chatterbot architecture in the political domain.

Actions (Repository Staff Only)

Item Control Page