Searching to Measure the Novelty of Collected Ideas

Visual representations of ideas are valuable for creative thinking and expression. Prior research on design and informationbased ideation has assessed novelty in creative products as the inverse of the frequency that an idea or visual element occurs in the complete space of responses. In controlled experiments, frequency has previously been calculated in reference to the set of ideas collected by all participants (corpus). Experimental conditions restricting the space of possible elements resulted in overlap between participant responses, yielding a range of frequencies. Alas, in field investigations the space of possible elements is unrestricted, resulting in little overlap of ideas, and thus mostly a single frequency (1/N ) of collected elements. We introduce a new method that uses web search to measure the novelty of individual ideas. Instead of using the local corpus directly to calculate frequency, we use the number of results from web searches generated by the elements in the corpus. Our implementation uses Googles reverse image lookup to determine the popularity of images. We compare results with those derived via prior methods. Author