Paper abstract: We present a method of detecting Influential blog entries, defined as entries which are 1)
persuasive/dissuasive or by which 2) readers can obtain new information. Although Influential is one
of the key usefulness factors, existing approaches fail to detect Influential blog entries reliably. To
address this issue, we utilize differences between the communication acts triggered by Influential blog
entries and by Non-influential blog entries; that is, our method of detecting Influential blog entries takes
the communication-based approach. In particular, twenty comment and trackback attributes are analyzed.
Compared to the baseline method (word-only approach), the proposed method offers better performance
in multiple/unknown domains.
Keywords:
Blog entry, Influential, Response, Communication, Behavior, Machine learning
