Reflection: Analysing Emotions of Social Writing

by Marina Santini

A few days ago, I attended a fascinating session organized by the Quantified Self Stockholm (QS) MeetuUp, in a venue with an inspiring name, Psykologifabriken (The Psychology Factory), in center Stockholm.

This QS session – Adding Power to body and soul… – included two presentations: one about adding power to the body through a robotic glove that adds gripping energy to the hand of those who have lost strength in this limb; the other one about methods to enable self-development through digital tools.

Since I am not into robotics, I will only say that the empowering glove shown by Johan Ingvast from Bioservo is simply amazing…

I am not a psychologist either, but I found the presentation about empowring the “soul” very relevant to some of my interests, namely sentiment analysis, mood tracking, and the impact of emotion on individual behaviours and social actions. I should also say that my interests in these areas are linguistically- and textually-oriented since I am a computational linguist. This means that, from my perspective, sentiment, mood, emotions, behaviours, and social actions are information that can be automatically extracted from texts, i.e. any written piece of communication, from tweets, to blog posts or emails or web comments.
This presentation was centred around some experiments on monitoring self-development through a mobile app called Viary. This app does not include any text analysis at the moment, but it is mainly based on check-boxes or radio buttons. Therefore, the statistics about personal development is computed through self-monitoring and self-recording one’s actions through a number of lists, similar to any online survey or questionnaire. Monitoring questions include topics like “how do you work with personal developent at work?” or “ways to improve relationships“; statistics includes “rescue time“, “mood rating“, etc. The presenter also talked about the key aspects of positive psychology, namely the benefit of regular practice, the importance of cultivating relationships, the advantage of remembering happy moments and so on. Many additional suggestions were put forward by the audience. My attention was drawn by some suggestions centred on language, for instance the fun and pleasure of learning together in language drop-ins, like Språkcafeet. Interestingly, my sitting neighbour suggested also a language-oriented practice, such as writing a few lines about more or less abstract topics like a colour or a picture.

The importance of writing in psychology is not new. For instance, Sonja Lyubomirsky (2008) states: “writing is inherently a structured process that forces a person to organize and integrate his thoughts, to reflect on what causes what, to create a coherent narrative about himself, and to consider systematic, step-by-step solutions.” (p.203). Importantly,Sonja Lyubomirsky emphises that: “writing is an effective strategy when one needs to cope with negative experiences because it helps a person make sense of them and to get past them” (p. 203). However, when it comes to the best experience in life she remarks that writing “may well prompt you systematically to analyze that event — for example, by breaking it down into its component parts — resulting in your probably reducing the pleasure associated with it and perhaps even evoking negative emotions, such as guilt or worry. So don’t savor through writing, but instead reflect, relish and share with others.” (p. 203). It seems that this latter suggestion underlies the picture-based mobile app called Timehop, which aggregates the best personal pictures from happy moments from social networks.

During this conversation between the audience and the presenter, I suddenly had the clear perception of the potential that the analysis of (written) language could have on positive psychology practices. What about creating an app that aggregates written posts and comments published by a user and automatically analysing the actual feelings the come out by this spontaneous written interaction with other people on social networks? This would provide more self-awareness about one’s spontaneous feelings and emotions hidden in our writing style. While self-monitoring application like Viary are based on the awareness that we are consciously making an effort towards self-development, the assessment of spontaneous writing would give hints about the actual, but unconscious, emotional state of a user in a point in time.

At present, several companies are focussing on the automatic identification of collective emotions, sentiment, mood or attitudes in the open web. The main approach is to extract information about important entities such as politicians, companies, events, etc. from the web, the blogshpere and social networks For instance, Recorded Future, Gavagai and Augify chew big quantities of data to detect mood trends, attitudes, opinions, general feelings of web communities. A recent European Project — “CYBEREMOTION” — has created SentiStrength, an application that “estimates the strength of positive and negative sentiment in short texts, even for informal language”. Language technology is then ready to help out! Not only for self-development, but also to get out from depressive moods, detrimental overthinking and all negative habits that hinder a more positive attitude towards life. The popular practice of reformulating negative thoughts in positive form whenever the level negativity is assessed to be self-destructive could be included in a new app…

Your thoughts are welcome…

2 comments for “Reflection: Analysing Emotions of Social Writing

  1. Jay
    14 December, 2012 at 11:40

    Hi Marina. Great post about the meetup! Thanks for the mention of Augify.


  2. 15 December, 2012 at 14:03

    Thanks, Jay. Hope to discuss more about this in the next meetup.

Leave a Reply

Your email address will not be published. Required fields are marked *