Towards an efficient R&D theme prediction with machine learning

This paper proposes an efficient method for R&D theme selection. There are various methods for the R&D theme selection such as patent analysis, paper survey and delphi investigation and so on. Patents and peer reviewed papers are easy to obtain, and are frequently used as materials for the selection method. In addition, making use of the national projects such as the social infrastructures feasibility project is one of the efficient theme selection method. A survey shows that the R&D theme selection is one of the biggest challenge for the private company. Generally, short term R&D theme selection is aiming at implementation within 2 years such as commercialization and sales expansion. Long-term theme is used for the R&D investment budget plan. On the other hand, Medium-term R&D theme is often aimed implementation within 5 years such as an exploratory technology theme. Since it relies on the heuristics knowledge for the theme selection with technology trends, an efficient selection method of the medium-term theme is required. Many technological analyses with the intellectual properties are performed so far. In this paper we propose a method of selecting the R&D theme using a machine learning based on public information.