Real R&D Options Under Sentimental Information

The problem of understanding how to modify the probability of success for a stage in an R&D project is still open. Primarily in cases where it is impossible to compare a project with other competitors, the probability of passing a certain phase of the experimentation is determined by taking into account only information from within the company and not from external information. In this paper, we propose to use Natural Language Processing techniques to obtain a sentiment score for the news from the outside world. In this way, we can transform sentences expressed in natural language into a numerical value which, in addition to the internal information, allows us to better "direct" the probabilities of success in a stage.