Temporal patterns of happiness and information in a global social network: Hedonometrics and Twitter (1101.5120v5)

Abstract: Individual happiness is a fundamental societal metric. Normally measured through self-report, happiness has often been indirectly characterized and overshadowed by more readily quantifiable economic indicators such as gross domestic product. Here, we examine expressions made on the online, global microblog and social networking service Twitter, uncovering and explaining temporal variations in happiness and information levels over timescales ranging from hours to years. Our data set comprises over 46 billion words contained in nearly 4.6 billion expressions posted over a 33 month span by over 63 million unique users. In measuring happiness, we use a real-time, remote-sensing, non-invasive, text-based approach---a kind of hedonometer. In building our metric, made available with this paper, we conducted a survey to obtain happiness evaluations of over 10,000 individual words, representing a tenfold size improvement over similar existing word sets. Rather than being ad hoc, our word list is chosen solely by frequency of usage and we show how a highly robust metric can be constructed and defended.

• The paper introduces a novel text-based hedonometer that quantifies happiness from massive Twitter data using word frequency distributions and crowdsourced word ratings.
• The paper uncovers clear temporal patterns showing happiness peaks on weekends and early mornings, with significant shifts during major societal events.
• The paper also measures increasing lexical diversity on Twitter, offering a complementary perspective on public sentiment for real-time societal monitoring.

Temporal Patterns of Happiness and Information in a Global Social Network: Hedonometrics and Twitter

Overview

The paper "Temporal patterns of happiness and information in a global social network: Hedonometrics and Twitter" introduces a novel approach to measure societal happiness using data from Twitter. The paper is significant due to its methodological contributions and observational findings. The authors, Peter Sheridan Dodds, Kameron Decker Harris, Isabel M. Kloumann, Catherine A. Bliss, and Christopher M. Danforth, formulate a text-based hedonometer to quantify happiness, designed to be real-time, remote-sensing, and non-invasive.

Methodology

The central innovation of the paper is the development of a "hedonometer," a metric for quantifying happiness from massive textual datasets. The hedonometer relies on large-scale word frequency distributions from Twitter, enriched by happiness scores of individual words obtained through Amazon's Mechanical Turk. The dataset comprises over 46 billion words from nearly 4.6 billion tweets spanning 33 months posted by over 63 million users. The paper uses a subset of 10,222 words identified by frequency for comprehensive analysis.

The hedonometer uses an algorithm that scales up individual word happiness scores to measure the happiness of a given text. By segmenting words around a neutral happiness score and excluding these "stop words," the instrument shows tunability and robustness. This methodological approach allows for various analyses, from daily and weekly cycles to reactions to significant societal events.

Observational Findings

The authors uncover several temporal patterns of happiness, revealing insights across various timescales such as hours, days, months, and years. The results show that happiness levels exhibit clear weekly and daily cycles. Notably, happiness peaks on weekends, especially Saturdays, and dips during weekdays, with Tuesday being the lowest point. Daily cycles indicate high happiness early in the morning (5-6 AM), decreasing through the day to a low in the late evening.

Significant societal events that led to deviations in happiness levels include the 2008 US financial bailout, the death of Michael Jackson, natural disasters like the 2010 Chilean earthquake, and the 2011 Tōhoku earthquake and tsunami. Conversely, positive spikes are observed during culturally significant events such as Christmas, New Year's Eve, and the Royal Wedding in 2011.

Through word shift graphs, the paper dissects how changes in word frequencies delineate these happiness variations. This detailed analysis reveals the primary drivers behind the shifts, offering a meaningful narrative around the temporal data.

Information Content

In addition to happiness, the paper measures information content using Simpson's lexical diversity. The findings indicate a significant increase in lexical diversity over the studied period, primarily due to the rising presence of non-English languages on Twitter. This measure complements the happiness index, providing a fuller picture of societal trends as seen through Twitter.

Keyword Analysis

The authors extend their analysis to tweets containing specific keywords and phrases, developing "ambient happiness" measures that remove the effect of the keyword itself. This analysis presents valuable insights into public sentiment on varied topics, from everyday expressions like "happy" or ":)" to political figures and events such as "Obama" and "BP."

Implications

By illustrating robust patterns in societal happiness and information diversity, the paper offers practical tools for real-time societal monitoring. The hedonometer can be instrumental for researchers and policymakers, providing an empirical basis for understanding public sentiment, potentially aiding in more informed decision-making.

Future Directions

Future research could explore geographic variations, individual user behavior over time, demographic influences, and more refined measures accounting for context and negations in text. Additionally, extending the hedonometer to other languages and incorporating predictive models to anticipate societal trends based on real-time data analysis would be valuable.

Conclusion

The paper contributes significantly to computational social science, particularly in sentiment analysis using big data. With its methodological rigor and broad array of findings, it sets the stage for future explorations into the digital footprints of human emotion at an unprecedented scale. The developed tools not only capture the pulse of societies but also offer a framework adaptable to various other datasets and social media platforms.