R U :-) or :-( ? Character- vs. Word-Gram Feature Selection for Sentiment Classification of OSN Corpora

Abstract

Binary sentiment classification, or sentiment analysis, is the task of computing the sentiment of a document, i.e. whether it contains broadly positive or negative opinions. The topic is well-studied, and the intuitive approach of using words as classification features is the basis of most techniques documented in the literature. The alternative character n-gram language model has been applied successfully to a range of NLP tasks, but its effectiveness at sentiment classification seems to be under-investigated, and results are mixed. We present an investigation of the application of the character n-gram model to text classification of corpora from online social networks, the first such documented study, where text is known to be rich in so-called unnatural language, also introducing a novel corpus of Facebook photo comments. Despite hoping that the flexibility of the character n-gram approach would be well-suited to unnatural language phenomenon, we find little improvement over the baseline algorithms employing the word n-gram language model.

Detection of Opinion Spam with Character n-grams

“…Character n-grams have been used for email spam detection [11] and sentiment classification [1] with higher effectiveness than using word n-grams. They are also considered the state-of-the-art for authorship attribution [19].…”

Incorporating Emotion and Personality-Based Analysis in User-Centered Modelling

“…Big social data offers the potential for new insights into human behaviour and development of robust models capable of describing individuals and societies [6]. Social media has been used in varying computer system approaches; in the past this has mainly been the textual information contained in blogs, status posts and photo comments [2,3], but there is also a wealth of information in the other ways of interacting with online artefacts. Research in image or video analysis includes promising studies on YouTube videos for classification of specific behaviours and indicators of personality traits [1].…”

Reprint of “Supervised sentiment analysis in Czech social media”

“…Analyzing media with a very informal language benefits from involving novel features, such as emoticons (Pak & Paroubek, 2010;Montejo-Ráez, Martínez-Cámara, Martín-Valdivia, & Ureña López, 2012), character n-grams (Blamey, Crick, & Oatley, 2012), POS and POS ratio (Ahkter & Soria, 2010;Kouloumpis et al, 2011), or word shape (Go et al, 2009;Agarwal, Xie, Vovsha, Rambow, & Passonneau, 2011).…”

“…Similarly to the word n-gram features, we added character n-gram features, as proposed by, e.g., (Blamey et al, 2012). We set the minimum occurrence of a particular character n-gram to five, in order to prune the feature space.…”