To prevent wasting excessive amount of time and effort on an innovative project, their owners need an automated system to evaluate if it can reach their required amount of money in crowdfunding initiatives such as Kickstarter. For this purpose, we are develop­ing a decision support system based on a project description text, to assess if a new project can reach their target amount of money. The originality of this research is that it extracts sub-categories (e.g., “mobile battery” out of general technology category) and observes their popularity change in time. Other project features include target amount of money, SNS mention counts, etc. Our algorithm observes a notable increase in accuracy of a project success prediction and can have high implication for measuring a new project’s popularity in other areas where crowds are involved, i.e., crowdsourcing.