ChatBots introduction and use cases Eugen Busoiu , ChatBots - - PowerPoint PPT Presentation
ChatBots introduction and use cases Eugen Busoiu , ChatBots introduction and use cases Well be talking about... 1. What is a chatbot and how it all started 2. How can a chatbot help your organization (possible use cases) 3. Building a
,
We’ll be talking about...
1. What is a chatbot and how it all started 2. How can a chatbot help your organization (possible use cases) 3. Building a chatbot in minutes 4. How do you monetize a chatbot?
What is a chatbot and how it all started
A chatbot can be described as a form of automated service, powered by a predefined set of rules and sometimes by machine learning algorithms and NLPs (Natural Language Processing), from where they can learn and optimize next interactions.
✓
Functions based on a set of rules This type of chatbot is very limited, it can only respond to specific commands
✓
Functions based on ML and NLPs Based on artificial intelligence, it can understands language and continuously gets smarter as it learns from conversations
What is a chatbot and how it all started
Early NLP computer program created by Joseph Weizenbaum in 1966. ELIZA imitated the language of a therapist and it was using a script that would recognize certain patterns and keywords, and generate a response accordingly.
ELIZA (1966)
Written in 1972 by Kenneth Colby, PARRY simulated the thinking of a paranoid individual.
PARRY (1972)
PARRY encounters ELIZA (September 1972)
ELIZA: Tell me your problems. Please terminate input with a period or a question mark. PARRY: People get on my nerves sometimes. ELIZA: I am not sure I understand you fully. PARRY: You should pay more attention. ….
What is a chatbot and how it all started
One of the earliest attempts at creating an AI through human interaction.
Jabberwacky (1988)
AI speech synthesis program created for MS-DOS based PCs. Designed to showcase a digitized voice, assuming the role of a psychologist when interacting with users.
What is a chatbot and how it all started
Stands for Artificial Linguistic Internet Computer Entity, natural language processing bot, applying heuristic pattern matching rules to human input. Created in AIML (Artificial Intelligence Markup Language), open, minimalist stimulus-response language for creating bot personalities.
A.L.I.C.E (1995)
First chatbot for popular messaging platforms, joined MSN Messenger and AIM early 2000s. Precursor to Apple's Siri and Samsung's S Voice.
Smarterchild (2001)
Uses natural language processing and ML to reveal insights from large amounts of data.
IBM Watson (2006)
What is a chatbot and how it all started
Personal assistants features natural language UI to answer questions, make recommendations and perform web service requests.
Siri, Google Now and Cortana (2010-2015)
Intelligent personal assistant, capable of voice interaction, music playback, providing weather and other real time information.
Amazon Alexa (2014) Messenger Bots (2016)
Launched in April 2016, Facebook Messenger Platform allows developers to build bots that can interact with Facebook users.
What is a chatbot and how it all started
Developed by Microsoft and released on Twitter. Designed to mimic the speach and habits of a teenage American girl. Shutdown after 16 hours because she rapidly developed a vicios paranoia and started tweeting all sorts of misogynistic, racist remarks.
Tay (2016) Next awesome chatbot? (2017)
How can a chatbot help your organization (possible use cases)
News/Publishing
✓
Personalized breaking news or specific stories
✓
Curated lists of stories and news, from multiple sources
✓
Sport news, alerts, live scores and stats
e-commerce
✓
Help people discover products and guide them through order process
✓
Gifts suggestions
✓
Send shipping updates
✓
Price alerts
How can a chatbot help your organization (possible use cases)
Entertainment
✓
Movie reservation - get access to all information about movies, show times and locations ✓ Event ticketing - browse and buy tickets to favorite events ✓ Image processing tools ✓ Help choose games, movies from a larger range based on personal preferences
How can a chatbot help your organization (possible use cases)
Customer service
✓
Integrate with companies ticketing/client support services ✓ Get ticket status updates
Business/Productivity
✓
Live stock quotes ✓ Updates or alerts about credit cards or bank account balance ✓ Legal assistance, price quotes and other advices for those opening a business or already have one ✓ Integrate with business tools and enable access to meetings scheduling service, HR, company updates and general information
How can a chatbot help your organization (possible use cases)
Travel
✓
Flight assistant - get flight documentation, booking details, check-in notification, status updates
✓
Hotel booking - search, browse and make reservations
✓
Railway info - railway company information and updates
✓
Order a ride from Uber or favorite taxi provider
How can a chatbot help your organization (possible use cases)
Food/Restaurants
✓
Cooking recipes based on user queries and preferences ✓ Food ordering services ✓ Restaurant suggestions according to user location and preferences
How to build a chatbot in minutes
Dividing the building process into steps:
✓
Prototyping Planning to create the next cleverbot? First, define scope by answering questions like: what will by chatbot do? will it provide information? will it be fun? how will the flow conversation be?
✓
Development
○
Using a web service for building and deploying chatbots (no coding skills required) examples: PandoraBots, Motion.AI, Converse.AI, Octane.AI, Chatfuel
○
Building a chatbot from scratch ✓ Testing Test the user conversational flow and UX examples: BotTesting.co, BotFlair.com, Dimon.co ✓ Promoting
How to build a chatbot in minutes
Factors to take in consideration when defining scope / plan for your chatbot:
✓ Platform(s) ✓ Where the users are
2 billion messages each month
How to build a chatbot in minutes for Facebook Messenger
Why Facebook Messenger? 1.2 billion users
100.000 active chatbots
* Source: Facebook data, April 2017 ✓
Massive adoption among users Reaching 1.2 billion users monthly (April 2017 stats) and more than 2 billion messages between users / businesses. ✓ Familiarity with the interface ✓ Simple API and BOT elements are well-defined Predefined templates available (quick replies, call-to-action buttons, lists and more) ✓ Updates directly in the Messenger
How to build a chatbot in minutes for Facebook Messenger
How to build a chatbot in minutes for Facebook Messenger
How to build a chatbot in minutes for Facebook Messenger
How to build a chatbot in minutes for Facebook Messenger
How to build a chatbot in minutes for Facebook Messenger
How to build a chatbot in minutes for Facebook Messenger
Monetization Strategies
How do you monetize a chatbot?
✓
Affiliation One of the oldest tricks. Affilition means connecting users with products/businesses. ✓ Leads generation ✓ Collaborations If you think that your bot can generate income for another company, collaborate with them and share revenue. ✓ e-commerce Drive sales by suggesting relevant products to users.