Sentiment strength detection for the social web

Abstract

Sentiment analysis is concerned with the automatic extraction of sentiment-related information from text. Although most sentiment analysis addresses commercial tasks, such as extracting opinions from product reviews, there is increasing interest in the affective dimension of the social web, and Twitter in particular. Most sentiment analysis algorithms are not ideally suited for this task because they exploit indirect indicators of sentiment that can reflect genre or topic instead. Hence, such algorithms used to process social web texts can identify spurious sentiment patterns caused by topics rather than affective phenomena. This article assesses an improved version of the algorithm SentiStrength for sentiment strength detection across the social web that primarily uses direct indications of sentiment. The results from six diverse social web data sets (MySpace, Twitter, YouTube, Digg, Runners World, BBC Forums) indicate that SentiStrength 2 is successful in the sense of performing better than a baseline approach for all data sets in both supervised and unsupervised cases. SentiStrength is not always better than machine learning approaches that exploit indirect indicators of sentiment, however, and is particularly weaker for positive sentiment in news-related discussions. Overall, the results suggest that, even unsupervised, SentiStrength is robust enough to be applied to a wide variety of different social web contexts.

Lexicon based feature extraction for emotion text classification

“…Max Emotion Intensity (MEI): Research in Sentiment analysis suggest that high sentiment-bearing terms is indicative of sentiment class of the document regardless of the average score for the document [38]. We expect it to be true for emotion analysis as well.…”

Semantic Patterns for Sentiment Analysis of Twitter

“…Lexicon Features: these features are formed from the opinionated words in tweets along with their prior sentiment labels (e.g., "good positive", "bad negative", "nice positive", etc.). We assign words with their prior sentiments using both Thelwall [23] and MPQA [27] sentiment lexicons.…”

SentiCircles for Contextual and Conceptual Semantic Sentiment Analysis of Twitter

“…Most current approaches for identifying the sentiment of tweets can be categorised into one of two main groups: supervised approaches [15,4,12], which use a wide range of features and labelled data for training sentiment classifiers, and lexicon-based approaches [22,14,6], which make use of pre-built lexicons of words weighted with their sentiment orientations to determine the overall sentiment of a given text. Some of these methods tend to achieve good and consistent level of accuracy when applied to well known domains and datasets, where labelled data is available for training, or when the analysed text is well covered by the used sentiment lexicon.…”

“…Popularity of lexicon-based approaches is rapidly increasing since they require no training data, and hence are more suited to a wider range of domains than supervised approaches [22]. Nevertheless, lexicon-based approaches have two main limitations.…”

“…Our results show that our SentiCircle approach outperforms the other lexicon methods by nearly 20% in accuracy and 30-40% in F-measure. We also compare our approach against SentiStrength [22], which, to our knowledge, is the leading lexicon-based sentiment detection approach for social media. Our approach outperformed SentiStrength in accuracy in 2 datasets, and in F-measure in one dataset only (detailed later).…”