This Came Up In A Previous Hobby.

Anyway, there's this hobby that some people get into where they write these computer scripts that pretend to be a human conversant (e.g. they "type" back and forth with a real human)... Back when I used to do this, the two major software platforms for this worked the same way:  1) you supplied complex patterns to be matched against whatever the user typed, and 2) for each of those you provide a list of random responses to choose from and serve back...  (And then you usually had a list of conversation starters to pick from when -nothing- matches)

In other words, you could match for questions about where the character lives and you could have responses like "I live in the Nation's capital.", "I live near Obama", "I grew up and Washington, DC, and still live there", etc...  (You have more than 1 response to pick from so if the same pattern gets matched multiple times in a conversation, doesn't look too "canned").

It's not artificial intelligence (like the authors of both platforms claim), but rather a clever farse best suited for automated FAQs, or believe it or not, art (cause you are creating a new "personality" portrayal, as you decide "the tone" of the responses).

Anyway, it sounds simple enough, but you'd be darned surprised how many ways there are to ask a person the same question (even something simple like how old someone is)...

I got to the point where I was coding these complex "include files" which would provide the patterns for all the ways a single question could be asked and then convert this into a unique tag which means that particular question has been asked.  Then I could include these in all my botscripts and only had to recognize the tag (cause the included code handled everything else).  Then if I found a new way to ask that question, I fixed this in one place and all my bots got the benefit...

ILoveMarie ILoveMarie
4 Responses Mar 10, 2010

Please do :-)<br />
<br />
It is strange to see artists doing things with technology now. There was an exhibition in Melbourne that WG went to and they were doing some pretty advanced stuff - things which would have been the preserve (and delight) of geeky types 10 years ago.<br />
<br />
I hope the use of technology by artistic types is only for the best ;)

Basically all of this stuff grew out of Eliza... it's all variations on the same theme... The difference is that people have made this easier for non-programmers to do... You literally have websites with web-UIs for making these now.. I used to be a user on one, but then I wrote my own such interface (A web app I call "Bella", and no this wasn't named after anyone as is a several year old project :-) ) and I do all mine from my own domain now...<br />
<br />
Anyway, everyone tries to call this AI, science, etc and really isn't.<br />
<br />
What I noticed is that when the technology got simpler for the average user (no longer required writing procedural code), this became more of an "artist" thing... Creative "linguist"/"writer" types became interested instead of geeks. And that's what I use this form mostly: art... An attempt at personality portrayals. I may be sending you a link to one of mine soon...

How did I miss this story??? I always wondered what happened to Eliza....<br />
<br />
Text parsing was probably easier back when people didn't really type that often. I have a feeling that since we've all started typing to communicate more, the ways of abbreviating and shortening sentences to communicate meaning have flourished. <br />
<br />
I wonder if your approach is used in language translation engines - would just about have to be, I think?

You got it exactly :-)