TY - GEN
T1 - Segmenting and characterizing adopters of E-books and paper books based on Amazon book reviews
AU - Guan, Lu
AU - Zhang , Yafei
AU - Zhu, Jonathan
PY - 2016/10
Y1 - 2016/10
N2 - Online product reviews through which consumers express their opinions and experiences with products are extremely valuable for both potential buyers to make informed purchase decisions and retailers to improve their products/services and adjust existing marketing strategies. One of the key challenges for mining product reviews is how to obtain a “ground truth” to guide the segmentation of reviewers properly. We propose a behavior-to-opinion approach, in which users are first categorized based on some unambiguous behavioral patterns (if available) and their online reviews are then classified to reveal unique and detailed characteristics of each user category. In this paper, we identify four categories of book consumers (i.e., kindle-only, print-only, printto- kindle, and kindle-to-print) based on the long-term patterns of their review behavior. Their review posts are then clustered through word2vec and K-means, and four categories of adopters are matched with their concerned word topics. Finally, we find that print-only adopters show significantly different patterns on content-oriented topics as compared to other three groups. Kindle-to-print adopters pay more attention on portability whereas print-to-kindle adopters stress more on money and user experience. Taken together, our work indicates a diversity of characteristics among four categories of book reviewers.
AB - Online product reviews through which consumers express their opinions and experiences with products are extremely valuable for both potential buyers to make informed purchase decisions and retailers to improve their products/services and adjust existing marketing strategies. One of the key challenges for mining product reviews is how to obtain a “ground truth” to guide the segmentation of reviewers properly. We propose a behavior-to-opinion approach, in which users are first categorized based on some unambiguous behavioral patterns (if available) and their online reviews are then classified to reveal unique and detailed characteristics of each user category. In this paper, we identify four categories of book consumers (i.e., kindle-only, print-only, printto- kindle, and kindle-to-print) based on the long-term patterns of their review behavior. Their review posts are then clustered through word2vec and K-means, and four categories of adopters are matched with their concerned word topics. Finally, we find that print-only adopters show significantly different patterns on content-oriented topics as compared to other three groups. Kindle-to-print adopters pay more attention on portability whereas print-to-kindle adopters stress more on money and user experience. Taken together, our work indicates a diversity of characteristics among four categories of book reviewers.
KW - Behavioral patterns
KW - E-books
KW - Product reviews
KW - Text analytics
UR - http://www.scopus.com/inward/record.url?scp=84994500762&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-84994500762&origin=recordpage
U2 - 10.1007/978-981-10-2993-6_7
DO - 10.1007/978-981-10-2993-6_7
M3 - RGC 32 - Refereed conference paper (with host publication)
SN - 978-981-10-2992-9
T3 - Communications in Computer and Information Science
SP - 85
EP - 97
BT - Social Media Processing
PB - Springer
T2 - 5th National Conference on Social Media Processing, SMP 2016
Y2 - 29 October 2016 through 30 October 2016
ER -