Building Chatbot Chennai Geeks Meetup

Here’s a quick summary of the Chennai Geeks December Meetup, this event was about “Building Chatbots”

Siddharta started with the intro talk by explaining a bit about the importance of chatbots in today’s world and continued with a hands-on tutorial, showing us how to build a slackbot from scratch using slack’s API. He also talked about python’s NLTK and Spacy library for Natural language processing, analyzing the sentiment of the received message and to make the bot reply with an apt text. He also spoke about wit.ai APIs (hosted solution) that you can use when you don’t want to do the ‘NLP’ part on your own.

Then we had lightning talks. Five people had volunteered to show a demo of the chatbots they’ve built.

Demo 1:

First one to go was Hariharan, he showed us his E-commerce bot based on telegram APIs which would display a short list of items. A user can select the items, enter the quantity and click pay button. Here’s the video link: https://www.youtube.com/watch?v=Arp28qcnLqM

He also showed us another simple valet parking bot, which can take a token number as input and ask the valet guy to get the vehicle ready so that the customer doesn’t have to wait too long for the car to arrive from the parking space. Video link: https://www.youtube.com/watch?v=SH8HgiP0LcA

Demo 2:

Bhavani showed us the interview-bot which greets the user, asks him to select a programming language and competence level to begin the test. Questions are displayed to the user based on his competence level. Users can paste code into the code editor, run it and submit to the bot for evaluation and then scores will be given to each candidate. All the results will be stored on an excel spreadsheet for the interviewer to review finally. She also mentioned another case where the bot would adjust the level of the user based on questions answered. Answering two questions correctly will take him to the next level automatically. She used a bit of luis.ai APIs for NLP.

Demo 3:

Trivikram showed us a Python tutor bot which can answer basic questions on Python programming language. It fetches answer from the database matching the question and sends it to the user. If the answer is not available in the DB, it would open a browser and display google search results matching the question. T he bot can also challenge the user with random question from the database upon user’s request and it has the ability to compile code with user-defined inpu(using hackerrank’s api). It can also store user data (like levels or course he took last time) and continues from where he left, the next time he logs in.

Finally, as a bonus he showed us his usual bot to bot conversation demo, where a waiter bot displays food menu to the user, he then selects the items to be ordered, waiter bot confirms the order and commands the chef-bot to prepare the food. He used api.ai for NLP and configuring the intent and responses.

Demo 4:

Ashish showed us the code he used to build a bot for his college’s computer science club which answers a bunch of questions about the club and also gives the student a programming challenge (question). The bot can also answer basic technology questions that the user asks, if the answer is not known, the bot would open stack overflow search results matching the question in a browser.

Demo 5:

Praneeth from Paypal showed us a demo of his PayPal customer support agent bot which can read customers questions and suggest solutions or create a dispute ticket in their dispute management portal. They usually get these kinds of questions from customers all the time so this chatbot helps automate the process, saving a lot of time for support engineers.

Hope everyone had a good time watching all these demos.

Thanks to Pravin Shekar and Ramesh for hosting us at Krea’s Office. Thank you Siddharta for the introduction tutorial. Thanks, Bhavani, Trivikram, Ashish, Hariharan and Praneeth for volunteering to do the lightning talks/demos.

Take a look at the photo album for more pics: https://www.facebook.com/iMagesh/media_set?set=a.10209971335074336.1073741828.1663955416&type=3