Development of a Prediction Engine for Scientific Technology Trends

Abstract: The Delphi method is typically used to predict promising technologies that may be invented in the future, but this approach is a difficult process that entails considerable time and cost. An alternative is latent Dirichlet allocation (LDA), which is a topic modeling method. In this research, LDA was used to develop a fast and low-cost prediction engine for scientific technology trends. The proposed engine features the use of text mining that is targeted toward the abstracts of patent documents. Aside from LDA, required fundamentals were used as bases for the design of the proposed engine. The results are expected to contribute to future research on scientific technology trends. The developed engine can also be used as an economical, automated prediction system in other related fields.