Extracting Meaning from Cell Phone Improvement Ideas
Jenine Turner, Raimondas Lencevicius, Mark Adler · 2009
Companies have recently begun gathering product improvement ideas via web tools. Resulting data collections are too large to be effectively dealt with by human users. But natural language processing and machine learning techniques are well suited for this type of problem. We explore several ways to organize such data in the cell phone domain: supervised classification, unsupervised clustering, and time-based analysis. Numerous companies nowadays gather product improvement ideas. Reviewing all of the resulting thousands of ideas without tools would require a great deal of time and resources. Automatic tools can help these reviewers in a number of ways. The questions we address here are categorization, finding common ideas, and finding idea trends over time. We explore techniques to answer these questions using suggestions from the cell phone domain. Each idea is presented to us as a title along with free text.