Chatbot for Gurdwara Sri Panja Sahab

?Chatbot for Gurdwara Sri Panja Sahib? is a project that investigates what can be achieved with AI agents to help people from all over the world to get their gurdwara-related questions answered in a human-like and timely manner, and how the management-related paperwork for Gurdwara visitors can be s

Project Title

Chatbot for Gurdwara Sri Panja Sahab

Project Area of Specialization

Artificial Intelligence

Project Summary

“Chatbot for Gurdwara Sri Panja Sahib” is a project that investigates what can be achieved with AI agents to help people from all over the world to get their gurdwara-related questions answered in a human-like and timely manner, and how the management-related paperwork for Gurdwara visitors can be semi-automated using a modern Web App. Businesses, by now, are taking benefit of developing bots of their own to accompaniment their social media existence. Why not for a Gurdwara?

Bots are becoming faster and smarter and will shortly turn out to be straightforward to use than a Website or a traditional app. Microsoft has hastened the novelty and superiority of bots and web apps, and we are realizing this project both interesting and provoking as we are diving into Microsoft Bot Framework and ASP. NET Core.

Through many companies has even now created bots of their own using the Microsoft Bot Framework, such as Skyscanner, Ebay, Foursquare, and StubHub. So, it is an opportunity for us to create an intelligent bot of Gurdwara Sri Panja Sahib using this cutting-edge Microsoft Bot Framework and a novel web app using ASP .NET Core.

Project Objectives

The key objective of this project is to present the strength of a bot and in what way it can be useful in answering the questions of people who visit the gurdwara, and the power of a modern web app that can reduce the agony of paperwork of gurdwara’s management.

The bot should be easier to use, reply in a well-timed fashion, able to respond to customer’s questions to make sure they never go unanswered, and be all around user friendly. It ought to make the user interaction as fast and easy as achievable to confirm that the users get what they want without any misunderstanding or difficulty, and that their time is not wasted. The conversation flow must continually retain the user in control of the conversation. This saves time for potential employees to do further work and lets the bot to handle the straightforward jobs.

The web app should be usable, and it should substitute the paperwork at this time done by the gurdwara’s management. The web app will be a unified spot for both the pilgrims or visitors and the administrators. Gurdwara’s potential data can be securely manipulated in the database via this web app, by constrained access rights on the resources to the specified roles that are assigned to the users of the app.

Many pilgrims come from different regions of the world. They may have many questions about gurdwara’s activities, so there must be some human resource to answer these questions 24/7 360. The problem is that a single person cannot be hired for such a tough duty. Other solution is to employ many different people to do such a job at different times of a day. Thus, our objective is to provide them a user-friendly bot that might be able to answer almost every question that the visitors may raise, and is available 24/7 360, hence reducing the human resource.

Project Implementation Method

The bot logic is written using Microsoft Bot Framework SDK v4, C#, and JSON. It exchanges messages with the users in the form of activities that are encapsulated in a Turn Context object. Such activities are represented in JSON payloads that are deserialized when working with the bot logic. Moreover, we have incorporated the following Azure Cognitive Services in our bot.

• LUIS lets us to do so that our bot can recognize the intent of user messages. We have written LUIS model in JSON in which we have created intents, one for each topic that we wanted LUIS to deal with, entities for some of the intents, and defined 20 representative user utterances per each intent. Subsequently, this model is trained and published over LUIS Endpoint service which is then registered via Dependency Injection in our project’s entry point. Furthermore, we recognize intents form user utterance by applying K-Nearest Neighbors with Cosine Similarity as the proximity measure, and recognize entities form them via Recurrent Neural Nets. Most of the replies produced in result of LUIS Model utilize Adaptive Cards schema that can be used to create elegant and interactive rich cards other than plain text messages.
• We have created Gurdwara’s Knowledge Base (KB) in JSON. This JSON KB is trained and published on QnA Maker Endpoint Service. This endpoint service is also registered via Dependency Injection in our project’s entry point. Where the utterances suitable for this service are directed towards QnA maker Logic Handler that is a plain C# class with asynchronous Turn Handlers. QnA Maker endpoint service retrieves top scored QnA pair form the KB via Azure Cognitive Search. Finally, the KB is populated with around 500 custom QnA Pairs written in JSON.
• At the top level of the hierarchy of Azure Cognitive Services resides our Dispatch Model. Dispatch is trained on representative utterances from our bot’s LUIS model, and QnA Maker KBs to route user utterances to the correct bot component.

Benefits of the Project

There are more reasons why our bot is a good choice of platform for building applications:

1. The paper-work can be reduced.
2. The need for human resources for answering to the queries of pilgrims and visitors, 24/7 360, is not required anymore.
3. Because conversation is natural, a bot is an ideal platform for creating an interface for anyone to use.
4. One problem with desktop applications or apps is all of the additional work associated with deployment. With desktop applications, you have to download and install the program. For apps, you visit a store and install on your device. Our bot resides in a Web app that people can easily use. The deployment process is as simple as inviting the bot into their workspace. There isn’t any heavy deployment or installation. People just say “Hi” to the chatbot and it begins communicating.
5. Besides messaging, bots can reside in apps, websites, and desktop applications. Thus, our bot is more versatile because it isn’t constrained to traditional computing devices.
6. One bot can serve several platforms, such as Cortana Skills, Slack, Skype, Facebook Messenger, and many more.
7. Our bot has one convention: plain language with the user. The conversations with the user are driven by the purpose of the chatbot and the desires of the user. Instead of following a call tree or hierarchy for a user to arrive at the desired functionality, users just ask the chatbot directly what they want.
8. A bot’s UI is conversational which is easier to use than GUIs.

Technical Details of Final Deliverable

The bot logic is written using Microsoft Bot Framework SDK v4, C#, and JSON. It exchanges messages with the users in the form of activities that are encapsulated in a Turn Context object. Such activities are represented in JSON payloads that are deserialized when working with the bot logic. Moreover, we have incorporated Azure Cognitive Services in our bot such as, LUIS, QnA Maker, and Dispatch.

Along with Azure Cognitive Services, we have written two distinct ComponentDialog classes to for taking user’s feedback, and requesting rooms for booking on the behalf of the user. These classes have their own dialog sets, and avoid any name conflicts with the dialog sets outside of them.

For collecting user’s details, we have created an Azure Cosmos DB account in our AAD along with Table API. The user’s data collected via RoomBookingDialog is stored on this database on Azure cloud. Our local web app reads data from Table Entities and merges it with SQL Server LocalDB.

We have published bot code on Azure, so it can be hosted on our local web app by creating an Azure Web App Bot resource and uploading the bot code on Azure. we have created GitHub private repository on which our bot logic is synchronized. After linking Azure App Services with our private repository, our bot is set up for continuous deployment. Finally, we are using Direct Line API 3.0 for our bot, since it provides full customization of the chat window.

Since the bot is being hosted on our local app, it can be used on every device that supports Internet connection, and a Web browser with JavaScript enabled.

Final Deliverable of the Project

Software System

Type of Industry

Media , Telecommunication

Technologies

Artificial Intelligence(AI), Cloud Infrastructure

Sustainable Development Goals

Industry, Innovation and Infrastructure, Responsible Consumption and Production

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com