Research

Affect Regulation via Technology Affordances

Affect regulation helps us to enhance social acceptability and desirability, whereas Affect dysregulation can lead to the development of mental-health disorders such as anxiety or depression. Therefore, “effective” Affect regulation is of paramount importance to a person’s well-being and successful function in society. As, all of us, one way or another, fail to regulate our stress and emotions, we wonder how deploying technology affordances, can facilitate effective Affect regulation.

More material will be posted after the submitted publications are accepted. 

 


 

What Causes an Emotionally Charged Comment on a Reddit Post?

With continuously growing use of social media websites, analyzing the sentiment of comments people post became invaluable in understanding users behavior and the impact the social media has on their sentiment. In this paper we investigated whether an individual's pre- and post-sentiment would significantly differ from one another, after exposure to the group sentiment on a post in Reddit social media. Our research question was formed based on a previous study by Goldenberg et.al, that exposure to average group sentiment, referred to as Collective Sentiment (CS) through the rest of the paper, has an effect on an individual’s sentiment. Given that in practice, mental computation of CS becomes a challenge, as the number of people who comment on post increases, and people are unlikely to reads all the existing comments, we hypothesized that a smaller subset for CS could be responsible for the sentiment change. Our results support such an observation as we found that sentiment change of an individual at time t is significantly affected by the sentiment of the commenter at time t-1 but not the CS. In this study, we focused on using Stanford NLP tool to extract sentence-level sentiments of the comments and then we applied our comment-level sentiment analyzer algorithm to identify the comment-level sentiment. Our attempt to improve the causal inference resulted in resulted in application of LDA topic model and Fourier Transformations to identify similar patterns in posts.

pardis-sriram-final-presentation1pardis-sriram-final-presentation3

pardis-sriram-final-presentation  pardis-sriram-final-presentation2  pardis-sriram-final-presentation4

Summary of Findings

  • Sentiment of threads in different sub-reddits behave differently.
  • Running LDA to get topics for comments is not a good idea.
  • The average sentiment of “group” in Reddit mostly does not affect the sentiment of a person commenting.
  • The cause of sentiment of a person’s comment most likely depends on the previous comment.

 

c

Effective Affective Communication

Affective Communication

A sample of 110 tech-savvy smartphone users were surveyed about their choice of communication medium (email, phone calling, text messaging, instant messaging, and video calling) and how effectively it communicated two primary emotions: happiness and anger. Results suggest the communicated emotion influenced both the choice of a medium and its perceived effectiveness. For example, email was preferred to video calling for communicating anger, and video was preferred to email for communicating happiness. In addition, happy conversations were rated as being more effective than angry conversations. These findings suggest that communication technology plays a larger role than just being a channel or a facilitator in ``affective communication'' -- it may be more or less conducive to expressing the communicated emotion.

With the barriers posed by current lifestyles and working conditions, social regulation of emotions through traditional face-to-face communication is not always an option. For example, compared to a generation ago, more of our close friends live in a different city, and our trusted colleagues are regularly located in a different office. As a consequence, people often need to choose among the various media options available today to socially regulate their emotions.

Yet, today's technologies are not designed with the primary goal of accommodating social regulation of emotion. There- fore, it is unknown which technologies help and which ones hinder communications about emotions. After choosing a medium, people still have the challenge to produce and maximize “readability”(easy-to-understand representation of emotional information). In aural and visual communications, readability may come from clear, verbal and salient social cues (e.g., facial expression, voice tone, and body gestures). In text-based communications, readability may be achieved through use of emoticons, capital letters, letter repetition, multiplication of exclamation marks, etc. In our opinion, medium choice and “readability” are interrelated and equally important to (1) reflect one's emotional state, (2) enable that state to be accurately assessed by receivers, and (3) communicate both affect and non-affect oriented information.

Summary of findings:

–  Emotion is a valid factor to consider when choosing a communication medium. But, currently available communication technologies are not “very effective” for affective communication.

–  Approximately 80% of longest interpersonal conversations are affective related. The rest of 20% is about planning.

–Techs-savvy population preference for  expression positive emotions and suppression of negative ones when using technology.

–The content of interpersonal conversations differ from one medium to another; more intimate over phone whereas more small talks over text-based mediums.

–The choice of a medium differs from one emotional state to another; Email is preferred to video calling when communicating anger, but when communicating happiness, video is preferred to email.