Voice of customer (VOC) is a key driving force for product development, as well as a fundamental base of marketing decision-making. The traditional approaches for VOC analysis using face-to-face customer interviews often result in high cost and long lead-time. Thus it is no longer suitable for dynamic response to changing market conditions. With fast growth and application of internet technology, extensive VOC with dynamic information can be collected and exchanged through the web media. In this paper, a new method of exploring and analyzing customer requirements on Web data source is presented. The scope and advantages of this approach are discussed. A systematic mapping framework, from the original web data to structured marketing setting, is constructed. Supporting knowledge base and rule-driven mining methods are also developed. Finally, a case study using car users data is presented to demonstrate the effectiveness of the proposed approach.