Bruce Wilcox - Jan 1, 2014:
New documentation on bots Harry, Stockpile, and NLTK (new bot for natural language utility processing).
New paper “Winning the Loebner’s”
new functionality to allow you to associate 32 bit values with any dictionary word and a walkdictionary function to browse the dictionary for values (used in natural language toolkit functionality).
Below is a table for Ben (ESL bot for SpeakGlobal), Suzette (for <a href="http://www.chatbots.org/avatar" class="term">Avatar</a> Reality), Tom Loves Angela
(for Outfit7) and Rose (for ourselves).
Ben: 10K rules 25 hours 30%gambits 37% responders 33% rejoinders
Suzette: 16K rules 32 hour 24% gambits 52% responder 14% rejoinders
TLA: 16K rules 20 hours 15% gambits 49% responders 36% rejoinders
Rose: 11.3K rules 10 hours 11% gambits 71% responders 18% rejoinders
Interesting reading. Can you estimate how many manhours went into the creation of each bot?