Virtual Coaches, also known as e-coaches, are a disruptive technology in healthcare. Indeed, among other usages, they might provide cost-effective solutions for increasing human wellbeing in different domains, such as physical, nutritional, cognitive, social, and emotional. This paper presents a systematic review of virtual coaches specifically aimed at improving or maintaining older adults' health in the aforementioned domains. Such digital systems assume various forms, from classic apps, to more advanced conversational agents or robots. Fifty-six articles describing a virtual coach for older adults and aimed at improving their wellbeing were identified and further analyzed. In particular, we presented how previous studies defined their virtual coaches, which behavioral change models and techniques they adopted and the overall system architecture, in terms of monitoring solutions, processing methods and modalities for intervention delivery. Our results show that few thorough evaluations of e-coaching systems have been conducted, especially regarding multi-domain coaching approaches. Through our analysis, we identified the wellbeing domains that should be addressed in future studies as well as the most promising behavior change models and techniques and coaching interfaces. Previous work illustrates that older adults often appreciate conversational agents and robots. However, the lack of a multidomain intervention approach in the current literature motivates us to seek to define future solutions.

No abstract

As the use of automated social media analysis tools surges, concerns over accuracy of analytics have increased. Some tentative evidence suggests that sarcasm alone could account for as much as a 50% drop in accuracy when automatically detecting sentiment. This paper assesses and outlines the prevalence of sarcastic and ironic language within social media posts. Several past studies proposed models for automatic sarcasm and irony detection for sentiment analysis; however, these approaches result in models trained on training data of highly questionable quality, with little qualitative appreciation of the underlying data. To understand the issues and scale of the problem, we are the first to conduct and present results of a focused manual semantic annotation analysis of two datasets of Twitter messages (in total 4334 tweets), associated with; (i) hashtags commonly employed in automated sarcasm and irony detection approaches, and (ii) tweets relating to 25 distinct events, including, scandals, product releases, cultural events, accidents, terror incidents, etc. We also highlight the contextualised use of multi-word hashtags in the communication of humour, sarcasm and irony, pointing out that many sentiment analysis tools simply fail to recognise such hashtag-based expressions. Our findings also offer indicative evidence regarding the quality of training data used for automated machine learning models in sarcasm, irony and sentiment detection. Worryingly only 15% of tweets labelled as sarcastic were truly sarcastic. We highlight the need for future research studies to rethink their approach to data preparation and a more careful interpretation of sentiment analysis.

Applying simple natural language processing methods on social media data have shown to be able to reveal insights of specific mental disorders. However, few studies have employed fine-grained sentiment or emotion related analysis approaches in the detection of mental health conditions from social media messages. This work, for the first time, employed fine-grained emotions as features and examined five popular machine learning classifiers in the task of identifying users with selfreported mental health conditions (i.e. Bipolar, Depression, PTSD, and SAD) from the general public. We demonstrated that the support vector machines and the random forests classifiers with emotion-based features and combined features showed promising improvements to the performance on this task.

No abstract