A week has passed since the opening of public access to Steve. The first dozens of users have already communicated with him. We have a little debugged the server part of the dialog interface.
Steve can already identify words or knowledge that he doesn't know. For about three next weeks, we will use the following learning strategy:
- Users communicate with Steve.
- Forming a vector of learning which is based on questions that Steve was unable to answer. It is passed to the Intelligent Dictionary knowledge extraction module.
- The found knowledge is checked by an expert, and if they are recognized as true, they are added to Steve's knowledge.
We launched this model in the middle of the week and the first results have already been obtained. New knowledge has been added and users in a couple of days were able to appreciate it.