Hello forum. I’ve just joined the chatbots community and I’m excited to dive straight in. Wasn’t sure the most locally appropriate place to introduce myself so I apologize if there is already a ‘hello world’ thread.

Meanwhile, on to my reason for coming. My metaquestion is as follows: Where is the best place to ask about -

I’ve heard of chatbots that read documents for linguistic bases and met some very interesting bots on the web that build their conversational bases from user feedback. Can anyone point me towards one that does both?

Finally, should such a thing be findable, does anyone think it feasible to try an run such a thing as an app for mobile ereaders? I would love my kindle so much more if it could love me back. :-)

Welcome to the forums Phineas! We’ll be happy to have NLP people on board.

In order to understand your questions: can you mention the example chatbots you’ve seen that read documents for linguistic based and those who build from conversational bases?

I’ve not sure, but I think most popular chatbots nowadays get information from external knowledge bases like Wikipedia and they always use their converstional history with other users.

If you check out our awards tab (http://www.chatbots.org/awards) you’ll see many example of winners in contests and as far as I know they all make use of knowledge bases and conversational history.

Thanks for getting back to me Erwin.

To answer your question, I can only be vague in my reference to those who read local documents for language base. Thought I saw mention of them while searching out terms like ‘ai’ and ‘chat with robots’ into google ~ 9 years ago. As for user feedback driven chatbots, I’ve had some intriguing conversations with jabberwacky and alice (plus spinoffs), though I really know virtually nothing of how theyre coded, which means my clarification below will probably horrify coders of NL parsers.

What I had imagineered was a personal agent styled bot which could talk with the owner of an ereader and also read the things they read. Perhaps such things could all be themselves part of a cloud like book club, swapping notes on individual quirks of language encountered (something I’m confident is regularly done).

I think such an endeavor would bring leagues of nuance into NL data stores, since it could simultaneously follow along with natural users of natural language (people) through different types of linguistic experience ie chat and reading. Not to mention that you’d be selecting users invested in language as a hobby for a perhaps more positive mob therapy

There are a couple of us here on the forum that are building NL parsers designed to do exactly what you propose: contextually parse volumes of text. The parser I’m building has the ability to store a complex sentence and ask the user to break the sentence down into simpler sentences that confer all the same information. The program then uses the complex sentence, simple sentences, and some grammar rules to build a template for analyzing similar sentences in the future. With this ability, combined with many hard-coded grammar rules, I’m hoping to develop a parser that can read simple English wikipedia and use the text to build a knowledge base.

I think it will be a while before any NLP program could read volumes of text and learn grammar from them unguided. But perhaps it won’t be so long before they come up with good grammar questions to ask. Your idea of using e-reader enthusiasts as language teachers is quite brilliant.

I think it will be a while before any NLP program could read volumes of text and learn grammar from them unguided.

I think so a well.