Hotel Chatbot

Shawna-Lee Pommells, Christian Soler, Ataberk Yanut, Nathan Chiche, Melissa Fernandez-Rubio

Product Owner: Amar Patel

Professor: Dr. Masoud Sadjadi

FLORIDA INTERNATIONAL UNIVERSITY

PROBLEM & SUMMARY

​

In the hospitality industry, a large chunk of staff allocate their time to handling customers inquiries, reservations, or general assistance during a stay. Some of these tasks can be very repetitive and can be address without necessary a staff, and some clients even just look through google for their answers if staff are a little busy. To improve efficiency, Hotels can provide guests with a personal chatbot assistance which can help them in whatever their needs and if not then it’ll send it to staff who will help them. This in the long term will make a hotel running very efficiently and can reduce operational costs.��During our Hotel Chatbot Project, we were tasks with developing a chatbot that can handle various hotel-related tasks, such as room reservations, answering FAQ’s, in-hotel services, and many more.

​

COLLEGE OF ENGINEERING AND COMPUTING

FLORIDA INTERNATIONAL UNIVERSITY

High-level scope

The high-level scope of the project is to create a chatbot for answering customer queries for hotel guests.  The information that the chatbot will need to access are details about the hotel that are typically asked for by customers; for example:

• When the check-in/check-out times are
• What is included in the breakfast items 

​

​

The customer may also report problems with their rooms or any other concerns relating to her or his stay at the hotel.  The chatbot will attempt to answer the customer's query as best as it can and will hand off to a live front desk clerk when the it cannot provide a suitable answer.  The chatbot will have a persona and can leverage generative artificial intelligence to provide answers in different words.

​

The chatbot will be trained using frequently asked questions collected from hotels' social media and hotel's online travel agencies.  The answers provided by the chatbot should be realistic and informative.  Suggestion cards will be used to filter the specific area of the customer's concern so that an accurate answer can be provided.

�The chatbot will be multi-channel based; as such, it will work via SMS, a chat widget on the hotel guests' landing page, phone call (via automated attendant), WhatsApp Business, hotel mobile application, social media platforms and Google Business Chat.

�

COLLEGE OF ENGINEERING AND COMPUTING

FLORIDA INTERNATIONAL UNIVERSITY

UML Diagram

COLLEGE OF ENGINEERING AND COMPUTING

FLORIDA INTERNATIONAL UNIVERSITY

User Stories

1. User wants to make/cancel reservations
2. User wants to ask questions about the hotel
3. User wants to report a problem
4. Admin wants to keep track of requests 

COLLEGE OF ENGINEERING AND COMPUTING

Microservices That Are Included

Repair

Customer Service

Booking

​

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Implementation

• Instant responses: guests don't want to hold out for responses.
• Multi-channel functionality: The capacity to interact with guests across all platforms such as SMS and WhatsApp
• Twilio Studio used to create ChatBot workflow, for SMS and WhatsApp

​

COLLEGE OF ENGINEERING AND COMPUTING

System Decomposition

• User Interface Module: This module handles the interaction with the users. It receives user input, such as text messages or voice commands, and presents responses from the chatbot in a user-friendly manner. The user interface can be implemented using web interfaces, mobile applications, or messaging platforms like Facebook Messenger or WhatsApp.
• Natural Language Processing (NLP) Module: The NLP module is responsible for understanding and processing user input in natural language. It includes tasks such as language detection, tokenization, entity recognition, and sentiment analysis. NLP allows the chatbot to comprehend user queries and extract relevant information for further processing.
• Intent Recognition Module: This module identifies the intent behind the user's message. It classifies user queries into specific categories, such as room booking, room service, check-in, check-out, etc. Based on the recognized intent, the chatbot can decide which module to direct the query for appropriate handling.
• Knowledge Base Module: The knowledge base module stores information about the hotel, such as room availability, pricing, amenities, policies, and frequently asked questions. It serves as the primary source of information for the chatbot to answer user queries accurately.
• Booking Module: The booking module handles room reservations and booking-related queries. It communicates with the hotel's reservation system to check room availability, book rooms, and provide booking confirmations to users.
• Customer Support Module: This module deals with customer support-related queries and issues. It can provide assistance for common guest concerns, such as lost items, maintenance requests, or complaints. For more complex issues, the module can escalate the query to a human support agent.
• Payment Module: The payment module is responsible for processing payment-related queries, such as making payments for reservations or handling payment issues. It can integrate with the hotel's payment gateway for secure transactions.
• Feedback and Review Module: This module encourages users to provide feedback and reviews about their experiences. It can prompt users to rate their stay and provide suggestions for improvement.
• Integration Module: The integration module ensures seamless communication between different modules of the chatbot and third-party services, such as the hotel's database, reservation system, payment gateway, and customer support platform.

​

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

TWILIO WORKFLOW

COLLEGE OF ENGINEERING AND COMPUTING

REQUIREMENTS

• The ChatBot will allow new guests to make reservations. 
• Existing guests will be able to cancel/change 

   an existing reservation (checkin/checkout)

• guests will have the capability to be transferred

 to the Front Desk with other questions.

• will be able to request services to their rooms such as new towels, tooth paste, etc etc.

COLLEGE OF ENGINEERING AND COMPUTING

Project development: Sprint 1

Sprint 1:

• Research on the different technologies to be used for the project: AI model, Architecture, APIs, security, SMS technologies, and user FAQ.

​

• Worked on JDL Models for the Jhipster code generator for our CRUD API(Create, Read, Update, and Delete)

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Project development: Sprint 2

Sprint 2:

• Finish creating JDL's for each respective microservice(chatbot, booking, repair, customer service, chatbot gateway) having organized fields for our entities. 

​

• Generating the code, Creating github, setting up initial technologies like docker, twilio, Jhipster, and libraries for our development environment

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Project development: Sprint 3

Sprint 3:

• Finding solutions to different errors when trying to compile. More research and figuring out what went wrong. Worked on a couple scripts

​

• After Code compiled started working on our respective microservices.

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Project development: Sprint 4

Sprint 4:

• Had a bunch of errors when posted to github and combining the microservices. Mostly due to versioning of the technologies, corrupt files and maven. Worked on figuring out how to fix it and continue through

​

• Continued work on the errors and continued work on each development branch for respective microservices

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Project development: Sprint 5

Sprint 5:

• Changed a couple technologies to have our project running smoothly(authentication mechanism, RBDMS, Versioning of technologies)

​

• Finally have the CRUD up and working with a frontend and started working on connecting twilio with the project

FLORIDA INTERNATIONAL UNIVERSITY

COLLEGE OF ENGINEERING AND COMPUTING

Summary

• The Hotel Chatbot app gave the chance to start working and developing on a product that would be highly beneficial in enhancing efficiency, especially in a busy environment like hotels, where receptionists often face overwhelming workloads.
• Our product is still in the development stages because of a lot of issues concering the backend of it but had been resolved so continuing there would be a couple more steps of integration like Twilio, using NPL, iteration and fine tuning.

​

COLLEGE OF ENGINEERING AND COMPUTING

FLORIDA INTERNATIONAL UNIVERSITY