Challenges of Sentiment Analysis for Dynamic Events

Abstract

With the proliferation of social media over the last decade, determining people's attitude with respect to a specific topic, document, interaction or events has fueled research interest in natural language processing and introduced a new channel called "sentiment and emotion analysis" [1]. For instance, businesses routinely look to develop systems to automatically understand their customer conversations by identifying the relevant content to enhance marketing their products and managing their reputations [2]. Previous efforts to assess people's sentiment on Twitter have suggested that Twitter may be a valuable resource for studying political sentiment and that it reflects the offline political landscape. According to a Pew Research Center report, in January 2016 44% of US adults stated having learned about the presidential election through social media. Furthermore, 24% reported use of social media posts of the two candidates as a source of news and information, which is more than the 15% who have used both candidates' websites or emails combined (http://j.mp/PewSocM). The first presidential debate between Trump and Hillary was the most tweeted debate ever with 17.1 million tweets.

Multimodal sentiment analysis using hierarchical fusion with context modeling

“…While most works approach it as a simple categorization problem, sentiment analysis is actually a suitcase research problem [17] that requires tackling many NLP tasks, including word polarity disambiguation [18], subjectivity detection [19], personality recognition [20], microtext normalization [21], concept extraction [22], time tagging [23], and aspect extraction [24]. Sentiment analysis has raised growing interest both within the scientific community, leading to many exciting open challenges, as well as in the business world, due to the remarkable benefits to be had from financial [25] and political [26] forecasting, e-health [27] and e-tourism [28], user profiling [29] and community detection [30], manufacturing and supply chain applications [31], human communication comprehension [32] and dialogue systems [33], etc.…”

Semi-Supervised Approach to Monitoring Clinical Depressive Symptoms in Social Media

“…For the above examples, the term “exhausted” would be a seed for the first user while “drained” and “tired” would be the seeds for the second user. To address the second challenge, given the recent advances in sentiment analysis techniques [28; 29; 30], we disambiguate a polysemous word based on the sentiment polarity of its enclosing sentence. We include a term as a seed only if the enclosing context has negative sentiment.…”

Predictive Analysis on Twitter: Techniques and Applications

“…Sentiment, emotions, volume, topics, and interactions between Twitter users can be considered as signals, while the importance and informativeness of each of these parameters may vary depending on the event and its domain. For instance, gauging the sentiment of a populace towards an electoral candidate would be very significant to predict the outcome of an election [56], but the same kind of information may not be as critical in the context of disaster management because, in the latter case, the sentiment may be largely negative. Further, for reliable decision making, the sentiment must be interpreted in a broader context.…”

“…Importance of the signals in some domains and their related events may vary, and sole use of these signals would not be sufficient to make a reliable judgment call, although these signals are essential parameters in a real-world event context. For instance, an election usually whips up discussions on various sub-topics, such as unemployment, foreign policy; and necessitates proper cultivation of a diverse variety of signals following contextual knowledge of the domain [56]. We provide two use cases in this subsection to illustrate how a coarse-grained or high-level predictive analysis can be conducted.…”