Fig. 6. (Color online) Similarity measurement using a k-nearest neighbor confusion matrix for the “topic” of user interest. A dendrogram of the authors' topics of interest was constructed using the confusion matrix. Bloggers writing about “Science” and “Technology” have similar writing styles. Even if there is not a significant connection between the topics, a cluster can still form if the authors have similar writing styles. This is evident in the topics of “Arts”, “Education”, “Non-Profit”, and “Student” (school life).
© NPSM