It is a well-known fact that humans have an enormous capacity for storing knowledge in memory. With the help of modern technology, you can replicate this feature by creating systems which act as a huge database of information. One particular type of system that has been gaining popularity, especially in recent years, is FAQ bots. FAQ bots are self-learning chatbots that are designed to learn and store information by interacting with a large number of people. It is this ability to learn over time which makes them popular among businesses and executives.
When you want to write a FAQ bot, it’s important to understand how this technology works and the various opportunities it presents for your business. You should also be aware of the various obstacles you might face, perhaps even before you finish building the system.
This is a great way to build an in-house support system because your FAQ bot can handle more complex requests than a typical search engine. It responds to questions rather than looks for keywords, and you can customize the bot’s language and tone of voice based on the user’s needs. It can also deal with complaints or concerns, and it doesn’t mind if you don’t know how to answer a query, because it’ll find someone who does.
To launch your FAQ bot, you need a plan and strategies that are tailored to your business goals. What is the purpose of your FAQ bot? How will you go about handling customer queries? And what if your FAQ bot promotes a product or service? Once you know what the goal of your FAQ bot is, you are ready to begin the design.
Getting your hands dirty
Today, we're going to build a FAQ bot for the popular messaging app, Telegram. We're going to start by learning how to use the natural language processing (NLP) API from Wonop. In these videos, we build on what we learned in the previous two tutorials. We introduce the concept of a knowledge graph and extend the logic to navigate this graph. While this is a beginner level tutorial, it is recommended that you see the two previous videos if you have not.
In the second half of the tutorial, we focus on make phrases sound natural and add relations between objects. At the end of the tutorial, you can extend the bot with additional knowledge.
As an additional exercise, try to see if you can extend the bot to cover other than “what is” questions.