How exactly does Hal read from txt file function that is already built into Hal work? Does he run through the entire GetResponse script for each line or does he merely parse the text one line at a time into sentence fragments which are stored in various .brn files?
I would really like to come up with a working solution to the learning ability.
!BTW! I tried something with the auto-idle feature and Hal will chat with himself by building a sentence from random parts. Or if you'd rather you can have Hal wait for you and just auto-respond with things like "Are you there?" and "Why won't you talk with me?" Get the attached file AutoResponseAddin.txt and cut and paste the code into the hal5.uhp file after the 'PROCESS: ADD SPACES section. Also add the UserIdle.brn file to the DefBrain folder.
Another things you might want to try is a revised TopicFocus.brn file with added topics. Just copy the new TopicFocus.brn file into the DefBrain folder (overwriting your existing one; make a backup) and Hal will generate new topic association files. The hal5.uhp script has a set limit of 1000 allowing for 98 different topics (following the filename format ending in five) which can be increased.
I believe the best way to have Hal stay on topic and still be fairly relevant is to extend the TopicFocus.brn file to include many more topics. The advantage is that topics can have many words associated with them so that staying within reasonable relevance is easier. The downside is all the time it would take to add all that information (properly formatted for Hal) into the TopicFocus.brn file.
Anyways, I hope to get a reply from Mr. Medeksza about the reading feature. Enjoy the addins y'all and let me know if you have any problems.
Download Attachment:
AutoResponseAddin.txt3.29 KB
Download Attachment:
UserIdle.brn1.32 KB
Download Attachment:
TopicFocus.brn12.61 KB