Conversation mock data - where to get or how to generate

Question

I'm writing a special sort of chat/forum software and need a source of mock conversations to use in screenshots, demos and tests.

They should contain no real names or other potentially harmful information, should be 100% correct English and should make at least some sense in relation to each other.

Example:

Message #1 Subject: Hello John, how about a pint of lemonade? Text: Hi John, just wanted to check if you have a free evening sometime next week. We could go out and have some lemonade. What do you think? Best, Sue

Message #2 Text: Hi Sue, what about Tuesday.

etc., but hundreds of those...

Is there an online source for this sort of thing? I thought about scraping my own email and removing names, but that's a lot of work. Also thought about putting two chat bots against each other, but don't know where to start. Any clues?

Answer 1

Have your friends and siblings use your chat software.

Answer 2

You could always go for the standard "Lorem ipsum..." placeholder text.

http://en.wikipedia.org/wiki/Lorem%5Fipsum

http://www.lipsum.com/

Answer 3

You can download one of the readily available chat bot scripts and adapt them to use the HTTP chat interface, one of the examples is Chatbot-Elisa perl module, which even includes the example that has two bots talking:

Sally:  Do you think coming here will help you not to be sad?
Harry:  We were discussing you -- not me.
Sally:  Does someone else believe I -- not you?
Harry:  You're not really talking about me -- are you?
Sally:  Do you sometimes wish you were not really talking about you -- are me?
Harry:  Why do you think I -- are you?
Sally:  Oh, I?
Harry:  Do you say you for some special reason?
Sally:  You're not really talking about me -- are you?
Harry:  What makes you think I am not really talking about you -- are me?

So you will just need to add the code that will glue them to your chat engine and look what they can come up with :) (Or, if all you are looking is the "content", you can just use that "twobots" example.

Answer 4

You want:

• A play, split into (speaker,words) tuples. It should be simple enough to use (f'rinstance) Macbeth, from Gutenberg, and a little bit of regexery to get the text into this format. Either use the scenes directly as chat threads, or...
• A Markov Chain text generator, trained on each character's speech, which will give you an infinite amount of data to play with.