Conversational AI:
Building clever chatbots
Tom Bocklisch, Lead ML Engineer @ LASTMILE
Example of a live Skill:
A customer can change
her address via
Facebook Messenger
Conversational AI will
dramatically change
how your users
interact with you.
An open source, highly scalable ML
framework to build
conversational software
is a technology company developing conversational AI.
Goal: next-generation intelligent bots
Team: tight-knit, fast-moving team of researchers,
engineers, designers and product people
Location: everywhere (honestly: Berlin, Edinburgh, Beijing)
We work on the core technology for next-generation conversational AI
Founders:
Dr. Alan Nichol (CTO)
Alexander Weidauer (CEO)
Advisory Board:
Chad Fowler (MD & CTO @ Wunderlist)
Matthaus Krzykowski (former Co-Founder @ Xyo)
Cat Noone (Designer & Founder @ Iris)
Investors: Reference customers:
Introduction
Architectural Overview
Introduction
AI
(Natural Dialogue
Management)
(Natural Language
Understanding)
(Natural Language
Generation)
(Conversational
Platform, e.g.
Facebook
Messenger)
“What’s the weather
like tomorrow?”
(User Request via text or voice)
“It will be sunny and
20°C.”
(AI response via text or voice)
(Your backend,
database or API)
Under The Hood
Natural Language Understanding
Natural Language
Understanding
What’s the
weather like
tomorrow?
Example Entity Extraction Pipeline
”What’s the weather like tomorrow?” { “date”: “tomorrow” }
Tokenizer
Part of Speech
Tagger
Chunker
Named Entity
Recognition
Entity Extraction
Example Intent Classification Pipeline
”What’s the weather like tomorrow?” { “intent”: “request_weather” }
Vectorization Intent Classification
Under The Hood
Demo
1. Create your training data
Demo
E.g. using the contributed rasa NLU gui at
https://golastmile.github.io/rasa-nlu-trainer/
2. Configure the model
Configure the
model
Demo
3. Train
Training the
model
Demo
4. Use Model
Playing around
with the trained
model
Demo
Under The Hood
Dialogue Handling
Under The Hood
“What’s the weather
like tomorrow?”
Intent
Entities
next
ActionState
previous
Action
“Thanks.”
after next
Action
updated
State
“It will be sunny
and 20°C.”
SVM
Recurrent NN
...
Detailed Dialogue Handling
Under The Hood
“What’s the weather
like tomorrow?”
“It will be sunny and
20°C.”
Entity
Input
Action Mask
Renormal-
ization
Sample
action
Action
type?
Response
API Call
Recurrent
NN
API Call
Entity
Output
Intent
Classification
Entity
Extraction
Similar to LSTM-dialogue prediction paper: https://arxiv.org/abs/1606.01269
Final Thoughts
Open Challenges
Final Thoughts
Challenges for curious minds:
●
● Unsupervised multi-language entity recognition
● Dialogue generalisation (e.g. optional questions)
Chit-Chat
model
Task-Oriented
model
I want to travel
to Spain.
?
● Combination of different dialogue models
Current Research
Final Thoughts
Good reads for a rainy day:
● Last Words: Computational Linguistics and Deep Learning (blog)
https://goo.gl/lGSRuj
● Memory Networks (paper)
https://arxiv.org/pdf/1410.3916
● End-to-End dialogue system using RNN (paper)
https://arxiv.org/pdf/1604.04562.pdf
● MemN2N in python (github)
https://github.com/vinhkhuc/MemN2N-babi-python
● Conversational AI is a big part of the future
● Deep ML techniques help advance state of the art NLU
and conversational AI
● Open source is strategically important for enterprises
implementing AI
Summary
Final Thoughts
3 take home thoughts:
Get in
touch!
Tom Bocklisch
Lead ML Engineer
tom@golastmile.com

Rasa AI: Building clever chatbots

  • 1.
    Conversational AI: Building cleverchatbots Tom Bocklisch, Lead ML Engineer @ LASTMILE
  • 2.
    Example of alive Skill: A customer can change her address via Facebook Messenger Conversational AI will dramatically change how your users interact with you.
  • 3.
    An open source,highly scalable ML framework to build conversational software
  • 4.
    is a technologycompany developing conversational AI. Goal: next-generation intelligent bots Team: tight-knit, fast-moving team of researchers, engineers, designers and product people Location: everywhere (honestly: Berlin, Edinburgh, Beijing) We work on the core technology for next-generation conversational AI Founders: Dr. Alan Nichol (CTO) Alexander Weidauer (CEO) Advisory Board: Chad Fowler (MD & CTO @ Wunderlist) Matthaus Krzykowski (former Co-Founder @ Xyo) Cat Noone (Designer & Founder @ Iris) Investors: Reference customers: Introduction
  • 5.
    Architectural Overview Introduction AI (Natural Dialogue Management) (NaturalLanguage Understanding) (Natural Language Generation) (Conversational Platform, e.g. Facebook Messenger) “What’s the weather like tomorrow?” (User Request via text or voice) “It will be sunny and 20°C.” (AI response via text or voice) (Your backend, database or API)
  • 6.
  • 7.
    Natural Language Understanding NaturalLanguage Understanding What’s the weather like tomorrow? Example Entity Extraction Pipeline ”What’s the weather like tomorrow?” { “date”: “tomorrow” } Tokenizer Part of Speech Tagger Chunker Named Entity Recognition Entity Extraction Example Intent Classification Pipeline ”What’s the weather like tomorrow?” { “intent”: “request_weather” } Vectorization Intent Classification Under The Hood
  • 8.
  • 9.
    1. Create yourtraining data Demo E.g. using the contributed rasa NLU gui at https://golastmile.github.io/rasa-nlu-trainer/
  • 10.
    2. Configure themodel Configure the model Demo
  • 11.
  • 12.
    4. Use Model Playingaround with the trained model Demo
  • 13.
  • 14.
    Dialogue Handling Under TheHood “What’s the weather like tomorrow?” Intent Entities next ActionState previous Action “Thanks.” after next Action updated State “It will be sunny and 20°C.” SVM Recurrent NN ...
  • 15.
    Detailed Dialogue Handling UnderThe Hood “What’s the weather like tomorrow?” “It will be sunny and 20°C.” Entity Input Action Mask Renormal- ization Sample action Action type? Response API Call Recurrent NN API Call Entity Output Intent Classification Entity Extraction Similar to LSTM-dialogue prediction paper: https://arxiv.org/abs/1606.01269
  • 16.
  • 17.
    Open Challenges Final Thoughts Challengesfor curious minds: ● ● Unsupervised multi-language entity recognition ● Dialogue generalisation (e.g. optional questions) Chit-Chat model Task-Oriented model I want to travel to Spain. ? ● Combination of different dialogue models
  • 18.
    Current Research Final Thoughts Goodreads for a rainy day: ● Last Words: Computational Linguistics and Deep Learning (blog) https://goo.gl/lGSRuj ● Memory Networks (paper) https://arxiv.org/pdf/1410.3916 ● End-to-End dialogue system using RNN (paper) https://arxiv.org/pdf/1604.04562.pdf ● MemN2N in python (github) https://github.com/vinhkhuc/MemN2N-babi-python
  • 19.
    ● Conversational AIis a big part of the future ● Deep ML techniques help advance state of the art NLU and conversational AI ● Open source is strategically important for enterprises implementing AI Summary Final Thoughts 3 take home thoughts:
  • 20.
    Get in touch! Tom Bocklisch LeadML Engineer tom@golastmile.com