Sentiment analysis in tickets for IT support

A benchmark study on sentiment analysis for software engineering research

“…Panichella et al [30] trained their own classifier on 2,000 manually-annotated reviews in Google Play and Apple Store, using a bag-of-word approach and Naïve Bayes for training. Blaz and Becker [3] developed a polarity classifier for IT tickets. They implemented an approach based on a domain dictionary created using semiautomatic bootstrapping to expand an initial set of affectively-loaded words used as seeds.…”

“…However, off-the-shelf sentiment analysis tools have been trained on non-technical domains and have been demonstrated to produce unreliable results in software engineering [16]. Trying to overcome the limitations posed by using off-the-shelf sentiment analysis tools, researchers recently started to develop their own tools specifically customized for the software engineering domain [1] [3][5] [6][18] [20] [30].…”

Sentiment Polarity Detection for Software Development

“…To overcome the limitations and threats to validity derived from the use of off-the-shelf sentiment analysis tools in empirical software engineering studies (Blaz and Becker 2016, Jongeling et al 2017, we train an emotion polarity classifier in a supervised machine learning setting by leveraging a gold standard of technical texts contributed by developers in Stack Overflow.…”

“…With a notable few exceptions (Blaz andBecker 2016, Panichella et al 2015), empirical software engineering studies have exploited off-the-shelf sentiment analysis tools that have been trained on non-software engineering documents, such as movie reviews (Socher et al 2013), or posts crawled from general-purpose social media, such as Twitter and YouTube (Thelwall et al 2012). Jongeling et al (2017) show how the choice of the sentiment analysis tool may impact the conclusion validity of empirical studies by performing a benchmarking study on seven datasets, including discussions and comments from Stack Overflow and issue trackers.…”

Bootstrapping a Lexicon for Emotional Arousal in Software Engineering

“…Recent research has raised concerns on the use of publicly available sentiment analysis tools for empirical software engineering [17], [18], which have been trained on nonsoftware engineering documents, such as movie reviews or posts crawled from general-purpose social media (e.g., Twitter and YouTube). In particular, Jongeling et al [18] assessed the predictions of popular sentiment analysis tools showing that not only these tools do not agree with human annotation of developers' communication channels, but they also disagree between themselves.…”