Entity Level QA Pairs Dataset for Sentiment Analysis

Abstract

In this paper, we present a named entity based sentiment analysis dataset OTTQA V1.0 that aims to detect primary TV series character opinions along with their sentiment polarities from the tweets generated from answer span extraction. The dataset, named OTTQA V1.0, contains 5237 unique question answer pairs from "Game Of Thrones" TV series . Along with the dataset, supportive tweets are extracted according to their relevancy with answer span keyword which is used to gauge opinion changes of OTT series characters over a given time period. The primary goal of proposing this sentiment analysis task is to provide the users with an utility dataset that calculates the sentiments of primary TV series characters from the tweets.