ar7hur

Also: https://github.com/facebook/duckling

The current Clojure version is quite stable, we used it at wit.ai/Facebook for several years before moving to Haskell.

I'd love to see somebody taking over and resuscitate it! One interesting direction could be to remove Java dependencies (mostly to Date) so that it's usable in ClojureScript. It would make a great JS library.

It's exactly how we used it internally for wit.ai

(Very small correction: Duckling is rule-based but uses a super simple Naive Bayes classifier to prioritize between the many potential parses produced by the rules -- we see it as a hybrid approach)

Yes all the training data is in the repo.

Duckling is relevant to parse very structured language, typically temporal expressions (dates and times...). It relies on a mix of rules and machine learning. Rules and datasets for many (human) languages are available in the repo. You don't need a lot of data to add support for what you need, owing to this hybrid rules+ML approach (as opposed to just ML).

Hello HN, I'm the original author of Duckling (with @blandinw). As usual, always happy to get feedback and suggestions.

a16z ("our" partner was Chris Dixon) led our seed round at Wit.ai. Working with them on a day-to-day basis was an incredible experience. They never asked for anything, but were always ready to jump in instantly when we needed anything from sales intros to offices in Palo Alto. When we got an offer from Facebook we founders wanted to accept it. They initially disagreed with us, but once we confirmed our decision they supported us 100% in the negotiation process. That's what I call class.

... or so says Durov, CEO of Telegram, without providing any way to fact-check his affirmation.

I admire the PR stunt.

OP was quoting Alex.

[Shameless plug] My own take on the birth, evolution and Mao's simplification of Chinese writing: https://al3x.svbtle.com/on-chinese-writing-1

Self-promoting our own Duckling https://duckling.wit.ai/ which has rules, but "prioritized" by ML -- and BSD license.

+1 for Fastmail with my own domain, I've used it for 4 years.

From the terms:

   By applying to Yellow, you agree to irrevocably waive any legal claim you 
   may have against us under any theory of law in any territory, including, without 
    limitation, copyright infringement or breach of implied in fact contract 
    (idea submission), that your rights were infringed due to any similarity between 
    your Content and any other content that is or may be developed by Snap

I would do something like this:

1. Get a few thousands book titles that are "long enough" so that the probability they appear in a sentence without meaning the book is low

2. Search these spans in HN comments and use this corpus to train Stanford's CoreNLP NER

3. Run the NER on all comments

4. Check on Openlibrary or another book DB that the extracted spans are real books titles

Author here: you're absolutely right. My tone when I wrote "weirdness" was supposed to be 2nd-degree :)

Author here, I just discovered this was posted on HN. Please note that it's episode 2 of a "series", the first episode is https://al3x.svbtle.com/on-chinese-writing-1

Thanks for your interesting comments!

"M.C. Dean charged $795 to $1,095 for access to the Wi-Fi it provided"

That's so outrageous.

I think they missed the most of important of all, DeepMind.

There's a jack in the mic port, so that cuts the internal microphone.

"Lambdas are relegated to relative obscurity until Java makes them popular by not having them." hahaha

> "how hard is it for someone to put true artificial intelligence into a chatbot? (...) The answer is: it requires 50 lines of code."

Well, it all depends on your definition of true intelligence...

Thanks Nick. No name yet, do you have a suggestion? Uberfitting? :)

I believe we share the same goal: make it easy to build smart bots. This is a very hard problem, all of us are far from solving it. It's good that different approaches are tried.

Wit.ai (including Bot Engine) is 100% Clojure. Our Grand Master is Rich Hickey.

1. None, AFAIK their API is a way to send and receive messages, but does not provide automation. Am I wrong?

2. We use a combination of traditional NLP (CRF NER, maxent classifiers) and deep learning. The "old" stuff needs less data to be trained. We'll continue to support duckling, we always love PRs, please help us!

You can use Bot Engine to build a Slack bot too, it's not tied to Messenger in any way, there is a HTTPS API.

Use cases on Messenger will probably be more around customer care or entertainment, news, etc.

Hi everyone! I'm one of the cofounders. This is an early beta of our Bot Engine. We would love to get the community's feedback. Thanks!

Wit.ai cofounder here. We would love to get the community's feedback for our new Bot Engine. It's still early beta. Thanks!

"For unpaid web users, MetaMind's products will be discontinued on May 4. For our monthly recurring users, MetaMind's products will be discontinued on June 4."

Our service Wit.ai (YC W14) does just that.

Demo: https://labs.wit.ai/demo/index.html