Document Type



This paper demonstrates a quick and efficient method of assessing the ideation structure for a group of people via the Internet and text mining. Data collection using the Internet is increasing; although the Internet has made access easier figuring out what people think is still a challenge. Email was used to contact and then direct survey respondents to a web site. At the web site, openended questions requiring text (type written) responses were asked. Conceptual (ideation) structure was obtained via an algorithm similar to that suggested by Quillian [7] and Bonnet [1]. To discover ideation structure, a modified Hopfield neural network based text-mining algorithm was used to obtain the statistical weights of ideas and concepts and the weights of the joint occurrences of the ideas and concepts with other ideas and concepts. Applying neural network technologies to text allows the analysis of open-ended responses without incurring the expensive, timeconsuming and error-prone task of manually reading the open-ended comments. The Internet via email contact and web-based survey input dramatically speeds the process.