Building Intelligent Chatbots from Scratch  

Introduction

 
  • Industrial application of bots

  • Business use cases

  • Types of chatbot: Retrieval based/Generative/ Open-domain/Closed-domain

 

Chatbot Mechanics: Concepts and Definitions.

 
  • Intent Detection

  • Entity Extraction

  • State Management

  • Information Retrieval

Chatbot Architecture: In this section, we will talk about standing up the bot architecture and framework. We will cover, in detail, the different tools we employed to create a web-based chatbot service.


  • Dialogflow (Google API for building simple chatbots): software review & how we hacked it and used it for our specific use case.
  • Flask: Creating simple apps in Python
  • Docker: Containerizing all the dependencies for running in an independent environment
  • Kubernetes Engine: Deploying the application on cloud to handle web traffic
  • App Engine: To route user traffic from Hangouts Chat.

Data Science Algorithms: In this section, we will talk about ML/AI models and how we deployed them as independent APIs, easily callable by any other service. We will focus on basic how-tos in Python, if required.


  • Basic NLP algorithms: NER, Google NLP API (used for intent detection)

  • FAQ model: Using text vectorization to answer simple FAQs

  • QA model: Using deep learning based Machine Comprehension model to answer question related to a particular document

  • API Creation: Converting models into API using Flask, Docker and Kubernetes for best production-level usage

 

Speakers



Amitoj Singh

Amitoj works with the AI Accelerator team which caters to clients’ analytically advanced requirements with reference to Machine Learning, Artificial Intelligence & Natural Language Processing. His project experience includes: building ML/AI enabled chatbots, bayesian networks, movie recommender systems and predictive maintenance for aviation. Currently he is working with PwC teams from multiple countries to build a Responsible AI platform for AI Fairness.

Amitoj has a background in Statistics and Econometrics and he holds a Masters in Economics from Indira Gandhi Institute of Development Research, an advanced research institute established by the Reserve Bank of India; and Bachelors in Economics from Hansraj College, University of Delhi.





Rishi Shah

Rishi is a data scientist at PwC US Advisory. He works with the AI Accelerator team which caters to clients’ analytically advanced requirements with reference to Machine Learning, Artificial Intelligence & Natural Language Processing. His projects in the past have included building machine-learning enabled chat-bots, developing predictive modeling and process-modeling analysis tools for a major multinational financial services corporation, and building advanced NLP engine to monitor mergers and acquisitions in global markets.

He is currently working on deep learning enabled simulation modeling for scenario analysis of the US healthcare industry.

Rishi is a graduate from IIT-Bombay.

 
Buy Ticket
Social media & sharing icons powered by UltimatelySocial