Overview
Most of us spend minutes on Netflix but are unable to find that perfect movie for ourselves. We get suggestions based on the previously watched movies, selected genre, popularity and affinity groups. And sometimes, we just find ourselves lost in this paradox of choice. This is a bummer especially when we are craving a good watch and that’s the challenge we tackled with Botflix - a chatbot that recommends movies on the basis of users’ personality type using Myers Briggs type indicators.
Type: Group Project, 2 members
Duration: 1 month (Fall Semester 2019, Designing for Smart Systems)
My Role: Intent Planning, Conversation Flow
Tools: Google DialogFlow API
Joe is bored at home on a cold Sunday evening. She is in a happy mood and wants to watch a movie. Yay!
She looks up for movie recommendations on Netflix. After minutes of failed search and not finding anything of her interest, she is pissed.
She is reminded of BotFlix, as suggested by her friend. She installs BotFlix and waits for the magic to happen!
To design a chatbot that recommends movies on the basis of users’ personality traits, using Myer Briggs personality type indicators
Project goals
● Attempts to understand user personality to find movies, characters and themes that they can best relate to through platforms like IMDB, Netflix, Prime Video, etc
● Facilitating an easy flow of conversation
● Ease user frustration and confusion while deciding which movie to watch
● Dealing with the paradox of choice on streaming apps such as Netflix
Hence, to establish a unique movie-recommendation system and deliver a personalised experience to users looking for for variation in movies.
Personality Types
Source: Do What You Are: Discover the Perfect Career for You Through the Secrets of Personality Type
Process Flow
Intent Planning
The intents were categorised into four broad groups:
● Category 01: Allowing the users to ask about BotFlix’s background and engaging in small talk (example: Tell me about yourself, I am bored, Tell me a joke, etc.)
● Category 02: asking a variety of questions to users for each indicator category; Providing users with three movie recommendations based on their personality types
● Category 03: Asking the users to start afresh if their reply don’t match the training data
● Category 04: Allowing the user to give feedback on the recommendations and provide with more suggestions or end the conversation
Question Planning
The questions were choice-based to categorise the users in each of the subgroups, and were out-of-movie context. We performed literature review on Myer Briggs personality types to get deep insights about how the characteristics, preferences and decision-making process of different personalities.
Following are the examples of our questions based on subgroups:
● Introversion or Extroversion
Do you spend your weekends staying cozyin bed or partying with friends?
● Sensing or Intuition
How do you solve problems: rely onfacts or explore different possibilities?
● Feeling or Thinking
Do you generally plan your vacations or like to be spontaneous?
● Perceiving and Judging
Do you like to work in organised spaces or flexible spaces?
Future Scope
● The recommendations can be enhanced by building a database of movies over time and improve the variety of recommendations provided
● Links to movies on different streaming channels can be added
● Immense future possibilities for Botflix by developing anuanced personalised questionnaire and training data
● Deploy on platforms such as Netflix, Youtube, Prime Video, and smart TV’s application.