Professor Jonathan ZHU, Dr Tetsuro KOBAYASHI, and Dr LIU Xiaofan of the Department of Media and Communication presented their latest research at the First International Workshop on Computational Humanities and Social Science at Tohoku University, Sendai, Japan on 19 January 2020.
Professor Zhu has been working with fellow researchers to develop methods for measuring mobile use sessions. While the amount of time users spend on their devices is the most widely studied characteristic of mobile phone use, other user behaviours such as the frequency, timing, and sequence of use are also worth to be examined. For instance, users A, B and C use their devices for the same amount of time in a given day, but A uses it mainly in the morning, B in the afternoon and C throughout the day. The different timing and frequency in their mobile use sessions tell valuable information. In addition, most people alternate between mobile-off time and mobile sessions. Devising methods to quantify various user behaviours will help address many unsolved theoretical or practical questions about mobile phone use, and be useful for understanding other mobile-off behaviours.
The collaboration project of Dr Kobayashi employs deep learning algorithms, a type of machine learning which is useful for classifying a large amount of information, to detect politicians’ faces from TV news. Tracking the appearance of politicians in TV news is important from the perspectives of political neutrality, comparing public and commercial broadcastings, and so-called presidentialisation (which means political leaders gain more power and public attention than political parties and other institutions). However, such tracking is almost impossible to achieve manually, especially when the news archive covers over a decade. In view of the shortcoming of the tedious operations, the research team has applied the deep learning algorithms against samples of NHK News 7 (Japanese evening news programme broadcast domestically on NHK General TV and internationally on NHK World Premium) and evaluated two other tracking methods at the same time. The performance of their deep learning algorithms was promising, and the team is endeavouring to improve it further.
Dr Liu presented the results of a recent scientometric study of communication journal publications. Scientometrics is a study of measuring and analysing the importance of scientists and their works, the construct and trend of research realms and the relationships between different research realms, through qualitative, quantitative and computational analysis of the contents and citations of scientific literature. In the study, Dr Liu and his team tried to look into the long precepted “fragmentation” problem of the communication research field. Communication research papers in the past 50 years to political science and sociology literatures in the same period were being compared using machine learning and network science algorithms. It is found that these social sciences realms are probably rather integrated than fragmented in terms of their research topics and methods, but fragmented from scientific collaboration’s perspective. This work will also be presented in the annual conference of the International Communication Association in May 2020.
Apart from demonstrating a wide range of topics and the advanced level of computational methods in the faculty research of the Department, they also had a fruitful exchange with other participants of the workshop from Japan and the Netherlands. As well, they met with Dr Hiroki TAKIKAWA, Director of Division of Computational Humanities and Social Science at Tohoku University, to pave the way for further collaboration.
Cover photo: (From the left) Dr Xiaofan Liu, Professor Jonathan Zhu, Dr Hiroki Takikawa, and Dr Tetsuro Kobayashi meet to discuss future collaboration opportunities.