The messages by Alan can be attached to the idea,
and saved as an inspiration or a hint to improve the
idea (see Figure 2, S5). Since the user is actively
involved in this system and the requests are relatively
short, we work with relevance feedback. We use the
relevance feedback, when the user marks and saves a
message from Alan to his or her idea, as our positive
feedback to determine changes in the weights of the
request [34]. In this way, the attached messages are
used and re-analyzed in conjunction with the initial
idea. This implementation represents the interaction
and learning ability of the AI-like system. Whenever a
user marks content as important or related to the idea,
the system adapts to the input and includes it into its
analysis.