Product reviews possess critical information regarding customers' concerns and their experience with the product. Such information is considered essential to firms' business intelligence which can be utilized for the purpose of conceptual design, personalization, product recommendation, better customer understanding, and finally attract more loyal customers. Previous studies of deriving useful information from customer reviews focused mainly on numerical and categorical data. Textual data have been somewhat ignored although they are deemed valuable. Existing methods of opinion mining in processing customer reviews concentrates on counting positive and negative comments of review writers, which is not enough to cover all important topics and concerns across different review articles. Instead, we propose an automatic summarization approach based on the analysis of review articles' internal topic structure to assemble customer concerns. Different from the existing summarization approaches centered on sentence ranking and clustering, our approach discovers and extracts salient topics from a set of online reviews and further ranks these topics. The final summary is then generated based on the ranked topics. The experimental study and evaluation show that the proposed approach outperforms the peer approaches, i.e. opinion mining and clustering-summarization, in terms of users' responsiveness and its ability to discover the most important topics. ; Department of Industrial and Systems Engineering