mycroft can only answer a few questions in the world.
Alexa or google always have an answer (sometimes funny). in my opinion that could extend the spread of mycroft.
many answers can not be answered locally and land at mycroft.i wish that the community will take care of these requests and thus create a personality. my thoughts on this:
- Everyone should be able to learn simple sentences. (why is the banna grumpy). currently this already works for the learning skill what is in the exam.
- these requests should be uploaded to mycroft. now the community should have the opportunity to examine the requests as precice.
- now everyone can profit. if only everyone learns such a sentence, the result will be wonderfully learned. and enrich the feeling with mycroft
- The data could also be used to train a chattbot or other things.
- a large part of this software already exists
I retired from Google last year, and while I was there question answering was one of the things I worked on. Most of the question answering done by Google is pulled from knowledge graph which was started by an acquisition of metaweb. That database is now ENORMOUS and basically consists of a bunch of entity-attribute pairs (e.g., “barack obama” -> height). it would be hard for mycroft to compete against that, but it’s completely possible that such databases could be developed within limited domains. The reality is that data is now a currency, and it’s not easy to find good sources of data for question answering. Even finding something to replace IMDB is nontrivial.
I’m not just interested in knowing. Also related like a monkey is an animal. I think I have no idea what such an option could achieve. everyone has their own ideas for each funktion a own skill is not a solution.
Mycroft tells me that “Barack Obama is 1.85 meters tall”
It is a challenge for anyone wanting to compete with Google, however there is the open source wikidata which appears to follow a similar structure to what you describe - entity: attribute: value. Google themselves give around $3M to wikipedia I believe. We also use services like Wolfram Alpha to answer general Q&A type questions.
As for the more personality side of things I know @gras64 that you have contributed to the Persona-Fallback previously. It hasn’t had a lot of attention lately but hopefully will have some more time to put into it this year.
you could set up persona-fallback with upload function. then you could then upload via the learning skill. and edit the data via mycroft translate.
tell me what I have to do.
As far as wikipedia is concerned lately I have been intensely busy referring to import for mimic recording tool, there is code to extract such an answer “Barack Obama is 1.85 meters tall”