So you want to build a chatbot from scratch? (Chatbots Part-1)

Chatbot Types: Rule-Based Chatbots and AI-Based Chatbots



A Chatbot (sometimes referred to as a chatterbot) is a computer program that attempts to simulate the conversation or “chatter” of a human being via text or voice interactions.
Basically, it receives the command from the user and responds back in a satisfactory way to the user which results in the completion of the action or conversation initiated by either of them.


“Its usage is potentially endless” many sectors like E-commerce, Finances, HR, Healthcare each and every place where there is a need for certain information to be responded back or certain action to be taken based on simple text or voice query by the user. Also, Chatbots Market worth 3,172.0 Million USD by 2021 which will only increase based on the NLP improvements and the AI improvements. Also, Chatbots expected to cut business costs by $8 billion by 2022

Trending Chatbot Tutorials

1. AWS setup for Deep Learning

2. Building a Successful Chatbot

3. How to Integrate API.AI with Chatfuel

4. Build a working SMS chat-bot in 10 minutes


There are to types of bots
1. Rule-Based Chatbots
2. AI-Based Chatbots

Rule-Based Chatbots

A rule-based chatbot is provided with a list of set answers for a set of queries a user might use the chatbot for. For example, if a user asks question X, the chatbot will answer Y. But the problem arises when the chatbot is having a query asked in a different way.
An example would be
Q1: How is the weather today?
Q2: Is it hot today?
Q3: Is it cold today?
Q4: Is it a good time to go out for a walk today?

All four of these are related to the weather but the problem is that we need to make sure that all four of them to be entered there to give the same response. Imagine 100 similar ways of asking the same question then we need to enter all 100 types of question for a single response.
Here, in this case, the NLP or NLU implementation is not present

If a question is asked in a new way where it has not been programmed then it will point to default intent where it will display mostly as “Sorry I didn’t understand it” which will affect user experience.

AI-Based Chatbots(Conversational Based)

These bots learn independently on there own based on the data provided and keep on learning and improving based on previous interactions. In this method, chatbots will learn to interact properly over a period of time. It needs to be interactive as it needs to respond like a human over a period of time.

The data that is fed or the responses were given needs to be controlled otherwise it will lead to chatbots developing a negative personality as was the case wherein Twitter taught Microsoft’s AI chatbot to be a racist in less than a day where the users were asked to train the bots by chatting with them and the bots learned from the users conversations.


Q & A:

Question1: Which one to choose among them?
It all depends on the customer If a bot needs to be deployed quickly, so their preference is for a rule-based bot but this will result in the loss of uniqueness and AI related capabilities.

Question2: Do we always need to prefer AI based bots?
Sometimes a simple rule-based chatbot is preferred and it is cheaper and the AI-based bot would be overkill. If it is a simple pizza ordering chatbot deciding about the toppings and cheese needed there is no need for AI-based bot here a simple rule-based chatbot would be fine.
But consider a situation where “I need to order pizza” OR “I am hungry” OR “I am starving” OR “I want something to eat” essentially means the same but needs NLP capabilities to understand and interpret these.

Question3: How accurate are the AI-based bots?
The accuracy of AI-Based bots depends broadly on three factors:
1. Machine Learning Model
2. Training of NLP
3. Historic Data

Question4: What’s the accuracy of these bots with different languages?
It all depends on the NLP engine that you are using and the method that will be followed for the translation to a common language and using a single NLP engine or building of a new NLP engine for an each of languages.

Question5: Is there a need for the chatbot to have a personality?
My opinion is that if you are trying to build intelligence does it make sense to not have a name or a personality or have an opinion. Humanizing a chatbot would make users feel that there is someone actually listening to them and responding appropriately.

Although having multiple personalities is not preferred. It is recommended to have a personality for the chatbot and continuously monitor it so it wont change over a period of time when it gets new data.

One thing is clear though — a chatbot is only as good as how it is programmed to respond. If AI is involved, then a chatbot can learn and improve over time. It takes a conscious effort to humanize a chatbot by designing Conversational UIs so that a chatbot can become a convincing conversational partner.


Want to learn more check out the topics below.

More references:

  1. Artificial Intelligence real or is it just a hype of this decade??
  2. Artificial Intelligence: Definition, Types, Examples, Technologies
  3. Artificial Intelligence vs Machine Learning
  4. Why Machine learning for achieving Artificial Intelligence? “ The Need for Machine Learning
  5. Machine Learning Types and Algorithms
  6. Linear Regression Part -1
  7. Linear Regression Part -2(example implementation)
  8. How to get AI related jobs?

Make sure to follow me on medium, linkedin, twitter to get more updates. And also if you liked this article make sure to give a clap and share it.

So you want to build a chatbot from scratch? (Chatbots Part-1) was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

Source link

Related posts

Google CEO: A.I. is more important than fire or electricity


Building a News Bot using Twitter API in 10 mins!


Probabilistic Model Selection with AIC, BIC, and MDL


This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy