[4Xin2] Poster session 2

[4Xin2-15] Revealing the Nature of Wikipedia Articles from Time-of-Day Viewing Trends

〇Taketoshi Yoshii¹, Mochihashi Daichi² (1.D2C Inc., 2.The Institute of Statistical Mathematics)

Keywords:Access Log, Clustering, Web Mining

In the digital ecosystem, the internet is overflowing with a myriad of content, from blogs to engaging videos, compelling media outlets to devise strategies that maximize viewership and revenue. It becomes essential to understand the patterns of user engagement, which are shaped by daily routines and timings.
Our study leverages Wikipedia's access logs to unveil two innovative methods for dissecting the temporal dynamics of content consumption. The first method utilizes natural language processing to categorize articles into genres, pinpointing those that garner peak interest at specific moments. The second approach employs Independent Component Analysis (ICA) to delve into viewing trends, thereby uncovering user lifestyle patterns and elucidating when and how content is consumed. These methodologies equip content creators and media strategists with the tools to fine-tune publication schedules, boost engagement, and customize content according to audience preferences, offering a strategic edge in the competitive digital sphere.