Chuck Bolin - Jul 15, 2010:
My plan is quite lofty and most likely ‘unobtainable’...but it represents my long-term objectives.
Chuck, I can very much appreciate that statement ! I have taken a very pragmatic approach in the development of my bot’s engine. I began this ‘quest’ at age 15 in 1986, and from that year, up to and including the end of 2008, my design was pretty much the same. I experimented with many, many approaches. It seems my latest algorithm approach in Feb of 2009 is the most promising. Right now, July 2010, I have spent countless hours (thank God I have a great wife, lol), and it seems very promising. I’m not 100% sure, but I think there will be something like six levels of code by the time I am completely done - 1st level, completed about 5 months ago, and 2 levels (grammer and meaning) which are being worked on concurrently right now, and probably until the end of the year at least, and then ‘reactor’ code hopefully in early 2011. Reactor code is where, once the bot has parsed, and chosen what it believes is the correct interpretation of what you said, goes about finding a rule or rules to apply it against, and generate an internal result. Then, the layer after that, will be the final one, which takes that internal state and generates a sentence, picking from a set of output templates perhaps.
So I can really relate to what you said, perhaps unobtainable. But we don’t know if it is unobtainable until we know exactly what and how we are going to do it… once we know that, we can make a better estimate of the time and effort. For example, when I code a grammer rule, it may take a bit of time, but that same grammer rule can be used in many, many other combinations of other sentence structures.