

臺大管理論叢
第
27
卷第
2S
期
21
Mining Consumer Knowledge from Social Media: Development of
an Opinion Mining Technique
1. Purpose/Objective
Social media and user-generated content have been considered as vital assets for
supporting crucial business intelligence applications. The knowledge gained from social
media can potentially lead to the development of novel services or products that are tailored
more effectively to users’ needs, and can also be used to meet the objectives of the
businesses offered to them. Online consumer reviews are a crucial type of user-generated
content. Appropriate analysis of consumer reviews not only provides valuable information to
facilitate the purchase decisions of consumers, but also helps merchants or product
manufacturers understand the general responses of customers regarding their services or
products for marketing campaign improvement. However, the rapidly increasing number of
consumer reviews hinders both consumers and businesses from obtaining a comprehensive
view of consumer opinions pertaining to products of interest when manual analysis
techniques are used. Therefore, developing techniques for analyzing the exceedingly high
number of online consumer reviews and summarizing their sentiments pertaining to specific
products is desirable and essential. The objective of this study was to design a novel and
effective opinion sentence identification technique, which can analyze and summarize
consumer sentiments pertaining to specific products from online consumer reviews.
2. Design/Methodology/Approach
We propose a rule-based opinion sentence identification (R-OSI) technique, which can
retrieve relevant review sentences to a specific product feature of interest from a large
volume of consumer reviews. Assume that we are interested in the “battery” product feature
of a digital camera; the R-OSI technique finds review sentences that discuss the battery
product feature from the selected digital camera consumer review corpus. Specifically, the
R-OSI technique adopts a semi-supervised learning approach by requesting a user to provide
keywords (e.g., battery, rechargeable, lithium) to describe the target product feature (e.g.,
battery). In addition, a set of unannotated consumer reviews are retrieved from various social
media websites. On the basis of the user-provided keywords and the set of unannotated
Chin-Sheng Yang
, Assistant Professor, Department of Information Management, Yuan Ze University
Pei-Yun Xie
, Master, Department of Information Management, Yuan Ze University
Hsiao-Ping Shih
, Master, Department of Information Management, Yuan Ze University