Development of a sentiment corpus from a gamified approach

College

College of Computer Studies

Department/Unit

Software Technology

Document Type

Archival Material/Manuscript

Publication Date

2010

Abstract

This paper discusses a sentiment corpus that was built from a gamified application. Each entry in the corpus contains a topic, a statement, and an annotation which tells if the statement bears a positive or negative sentiment towards the topic. These were generated from the Polarity gamified web application. Manual evaluation of the corpus yields an 82.86% accuracy and a 90.99% inter-rater agreement, showing that there is a potential for corpora built from gamified approaches to be used in natural language processing tasks.

html

Disciplines

Computer Sciences

Note

Publication/creation date supplied

Keywords

Sentiment analysis; Natural language generation (Computer science)

Upload File

wf_no

This document is currently not available here.

Share

COinS