TwMiner: Mining Relevant Tweets of News Articles

Abstract

In this paper, we propose a pseudo-relevance feedback-based automated approach that utilizes both the content and context attributes of a news article to determine the tweets relevant to that news article. Extensive empirical validation on a set of 1000 news articles highlights that the proposed approach can ensure high precision (0.942) in comparison to the current research works and can successfully extract relevant tweets for a majority of the news articles, around 95% of the total news articles.