Identifying long-term periodic cycles and memories of collective emotion in online social media
arXiv:1903.06410 · doi:10.1371/journal.pone.0213843
Abstract
Collective emotion has been traditionally evaluated by questionnaire survey on a limited number of people. Recently, big data of written texts on the Internet has been available for analyzing collective emotion for very large scales. Although short-term reflection between collective emotion and real social phenomena has been widely studied, long-term dynamics of collective emotion has not been studied so far due to the lack of long persistent data sets. In this study, we extracted collective emotion over a 10-year period from 3.6 billion Japanese blog articles. Firstly, we find that collective emotion shows clear periodic cycles, i.e., weekly and seasonal behaviors, accompanied with pulses caused by natural disasters. For example, April is represented by high Tension, probably due to starting school in Japan. We also identified long-term memory in the collective emotion that is characterized by the power-law decay of the autocorrelation function over several months.
arXiv comment19 pages, 5 figures, 2 tables, accepted PLOS ONE
References in corpus (14)
- Distributed Representations of Words and Phrases and their Compositionality
- Twitter mood predicts the stock market
- Distributed Representations of Sentences and Documents
- Temporal patterns of happiness and information in a global social network: Hedonometrics and Twitter
- Measuring Emotional Contagion in Social Media
- Early Prediction of Movie Box Office Success based on Wikipedia Activity Big Data
- Measuring the happiness of large-scale written expression: Songs, Blogs, and Presidents
- Circadian patterns of Wikipedia editorial activity: A demographic analysis
- Competing for Attention in Social Media under Information Overload Conditions
- PANAS-t: A Pychometric Scale for Measuring Sentiments on Twitter
- Analysing Mood Patterns in the United Kingdom through Twitter Content
- Self-organization on social media: endo-exo bursts and baseline fluctuations
- Empirical observations of ultraslow diffusion driven by the fractional dynamics in languages: Dynamical statistical properties of word counts of already popular words
- Detecting Events and Patterns in Large-Scale User Generated Textual Streams with Statistical Learning Methods