Nation-wide Mood: Large-scale Estimation of People's Mood from Web Search Query and Mobile Sensor Data
This work addresses the challenge of estimating user moods for user-centric services, though it is incremental as it builds on existing data sources without a new paradigm.
The authors tackled the problem of estimating people's affective statuses from web data by proposing a method that combines web search queries and mobile sensor data, resulting in a system deployed with over 11 million users that revealed daily rhythms, pandemic-related mood fluctuations, and news impacts.
The ability to estimate the current affective statuses of web users has considerable potential for the realization of user-centric services in the society. However, in real-world web services, it is difficult to determine the type of data to be used for such estimation, as well as collecting the ground truths of such affective statuses. We propose a novel method of such estimation based on the combined use of user web search queries and mobile sensor data. The system was deployed in our product server stack, and a large-scale data analysis with more than 11,000,000 users was conducted. Interestingly, our proposed "Nation-wide Mood Score," which bundles the mood values of users across the country, (1) shows the daily and weekly rhythm of people's moods, (2) explains the ups and downs of people's moods in the COVID-19 pandemic, which is inversely synchronized to the number of new COVID-19 cases, and (3) detects the linkage with big news, which may affect many user's mood states simultaneously, even in a fine-grained time resolution, such as the order of hours.