Exploiting tweet sentiments in altmetrics large-scale data

This article aims to exploit social exchanges on scientific literature, specifically tweets, analyse media users’ sentiments towards publications within a research field. First, we employ the SentiStrength tool, extended with newly created lexicon terms, classify of 6,482,260 tweets associated 1,083,535 provided by Altmetric.com. Then, propose harmonic means-based statistical measures generate specialised lexicon, using positive and negative sentiment scores frequency metrics. Next, adopt novel article-level summarisation approach domain-level analysis gauge opinion users Twitter about literature. Last, an aspect-based analytical mine expressions relating various aspects article, such as its title, abstract, methodology, conclusion or results section. We show that communities exhibit dissimilar their respective fields. The field-wise distribution shows in Medicine, Economics, Business Decision Sciences, tweet are focused In contrast, Physics Astronomy, Materials Sciences Computer Science, these methodology Overall, study helps us understand online community Specifically, fine-grained may help improving articles disseminate findings effectively further increase societal impact.