tag:blogger.com,1999:blog-18648436.post725215204942668071..comments2017-05-23T12:56:25.318+01:00Comments on Wading Through Treacle: My theorem-prover is (basically) workingNigel Seelhttp://www.blogger.com/profile/17960803873815821524noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-18648436.post-18471891096439834002017-03-16T11:40:42.104+00:002017-03-16T11:40:42.104+00:00In fact if N is the number of facts and rules in t...In fact if N is the number of facts and rules in the axioms, then I expect the chatbot's to be an extensible list of 'microtheories' so filtering on relevance solves the problem.<br /><br />If N is inference depth (eg number of relevant rules, or recursion depth) then efficiency is more of an issue but due to lack of inferential depth in normal conversation, I'm not that worried!Nigel Seelhttp://www.blogger.com/profile/17960803873815821524noreply@blogger.comtag:blogger.com,1999:blog-18648436.post-66686829938702847542017-03-16T08:10:11.980+00:002017-03-16T08:10:11.980+00:00Q&A Continued 6. What about scaling up to larg...Q&A Continued 6. What about scaling up to larger N?<br /><br />A. As I explained in (3) above I will not be using a large inference base and so do not expect the Combinatorial Explosion to be relevant to the Chatbot application. It probably scales fairly badly but I hope that wont matter.Royhttp://www.blogger.com/profile/17844610890344836500noreply@blogger.com