Adaptive Text Extraction for New Product Development

The first step in product design and development involves concept generation. Concept generation involves identifying customer needs and then mapping those needs onto a set of product attributes (specifications). Traditional methods for concept generation involve focus groups, surveys, and anthropological studies to assess user needs. Techniques, like Quality Function Deployment (QFD), then guide designers in relating needs to explicit product specifications. In this paper, we propose to augment traditional methods for concept generation by automatically processing user generated online product reviews. We apply adaptive text extraction methods to automatically learn user needs and product attributes. Association rule mining is used to learn the mapping between needs and attributes. We summarize results from prior work for independently learning user needs and attribute specifications from product reviews and then discuss the application of these methods to concept generation for new product development.