30/11/2024

HOMEWORK 1

Messar Aya
Benamara Icherak
Project Title
Design and Implementation of a Chatbot Based on Internal Company Data

Project Summery
This project aims to design and implement a chatbot tailored to an organization's specific needs
using internal data. CM Consulting (CMC), a leader in digital transformation, incorporates ERP
solutions and SAP HANA to optimize processes. The chatbot will utilize these tools and data to
deliver real-time assistance to employees and customers, enhancing efficiency and decision-
making.

1. The most recent related article

Title:
Developing Chatbot Conversational Systems & the Future Generation Enterprise Systems

Source:
View of Developing Chatbot Conversational Systems & the Future Generation Enterprise
Systems

Objective:
The article explores developing chatbots for enterprise systems, focusing on aiding employees
in adapting to new ERP systems and improving efficiency.
Using Research Rabbit

Title:
CHATBOT: Architecture, Design, & Development

Source:
(4) CHATBOT: Architecture, Design, & Development | Andrew Hutapea -
Academia.edu

Objective:
The article aims to provide a comprehensive overview of the architecture, design, and
development of chatbots. It focuses on the various components and methodologies involved in
creating effective chatbot systems.

Summery:
1. Architecture:

- Components: The architecture of a chatbot typically includes several key components:

- Natural Language Processing (NLP): Converts user input into structured data that the
chatbot can understand. This involves tokenization, normalization, and entity recognition.

- Dialogue Management: Manages the flow of conversation, ensuring that the chatbot
responds appropriately to user inputs.

- Backend Integration: Connects the chatbot to external systems and databases to retrieve
and process information.

- User Interface: The front-end interface through which users interact with the chatbot, such
as a web chat window or messaging app.

2. Design:

- Intent Recognition: Identifying the user's intent based on their input. This is crucial for
understanding what the user wants to achieve.

- Entity Extraction: Extracting relevant entities from the user's input, such as dates, names, or
product IDs.
 - Response Generation: Generating appropriate responses based on the user's intent and
extracted entities. This can be rule-based or use machine learning models.

3. Development:

- Rule-Based Systems: These chatbots follow predefined rules to generate responses. They
are simple to implement but lack flexibility.

- Machine Learning Models: These chatbots use machine learning algorithms to understand
and respond to user inputs. They can learn from interactions and improve over time.

- Hybrid Models: Combining rule-based and machine learning approaches to leverage the
strengths of both.

Title: Chatbot using NLP

Source: AnaPub - Books
Objective:
The article aims to explore the development and implementation of chatbots using Natural
Language Processing (NLP). It focuses on how NLP can be leveraged to create intelligent
conversational agents that can understand and respond to human language effectively.

Summery:
Methodology: The article employs a practical approach to developing chatbots using NLP
techniques. It discusses various components and methodologies involved in creating effective
chatbot systems, including:

• Natural Language Processing (NLP): Techniques for understanding and processing

human language.
• Machine Learning: Algorithms and models used to train the chatbot to understand and
respond to user inputs.
• Dialogue Management: Strategies for managing the flow of conversation and ensuring
coherent interactions.

Technical Details:
 1. Natural Language Processing (NLP):
a. Tokenization: Splitting user input into individual words or tokens.
b. Normalization: Converting words to their base forms and correcting typos.
c. Entity Recognition: Identifying and classifying entities within the user's input.
2. Machine Learning:
a. Training Models: Using machine learning algorithms to train the chatbot on a
dataset of user interactions.
b. Intent Recognition: Identifying the user's intent based on their input.
c. Response Generation: Generating appropriate responses based on the user's
intent and extracted entities.
3. Dialogue Management:
a. State Management: Keeping track of the conversation state to provide
contextually relevant responses.
b. Flow Control: Managing the flow of conversation to ensure a natural and
coherent interaction.

Using Connected Papers

Title:
Developing an Enterprise Chatbot using Machine Learning Models: A RAG and NLP based
approach

Source:
Developing an Enterprise Chatbot using Machine Learning Models: A
RAG and NLP based approach - Webthesis

Objective:
The thesis aims to create an innovative chatbot to automate specific business processes. It
focuses on developing a conversation system capable of processing complex information and
providing comprehensive answers in textual, tabular, or graphic formats, extracted directly
from company documents or databases.
Summery:
Methodology: The thesis employs advanced open-source machine learning models and search
optimization methodologies, such as Retrieval Augmented Generation (RAG) and its variants, to
improve the efficiency and quality of the chatbot's responses. Data preprocessing techniques
were implemented to maximize performance on both the data and the entire pipeline.
Experiments were conducted on business data from documents and relational databases to
identify optimal methods and parameters.

Technical Details:

1. Machine Learning Models:

a. SQL Querying: A specific model was selected for SQL querying to retrieve data
from databases.
b. Information Retrieval: Another model was used for retrieving relevant
information from documents.
c. Natural Language Generation: A model was chosen for generating natural
language responses.
2. RAG Optimization:
a. Retrieval Augmented Generation (RAG): Combines retrieval-based methods
with generative models to enhance the chatbot's ability to provide accurate and
contextually relevant answers.
b. Data Preprocessing: Techniques were implemented to clean and prepare data
for optimal performance.
3. Experiments and Results:
a. Task-Specific Models: Different models were tested for specific tasks, leading to
the selection of the most effective ones.
b. Performance Trade-offs: Two main pipelines were identified: one with optimal
performance but longer execution time, and another less accurate but
significantly faster. The faster method was chosen for its acceptable accuracy
and improved response speed.

Title:
Build Your First Chatbot with SAP Conversational AI

Source:
Build Your First Chatbot with SAP Conversational AI | SAP Tutorials

Objective:
The article aims to guide users through the process of creating a simple chatbot using SAP
Conversational AI. The goal is to help users understand how to set up, train, and deploy a
chatbot that can handle basic interactions such as recognizing greetings, telling jokes, and
responding to user reactions.

Summery:
Methodology: The article uses the SAP Conversational AI framework. It provides a step-by-step
tutorial on:

1. Creating an SAP Conversational AI account.

2. Setting up a new bot project.
3. Training the chatbot to understand user intents and expressions.
4. Managing the conversation flow with skills.
5. Testing the chatbot to ensure it functions correctly.

Technical Details:

1. Account Creation:
a. Users are guided to create an SAP Conversational AI account and validate it via
email.
2. Bot Project Setup:
a. Instructions on setting up a new bot project, including defining intents and
adding expressions.
3. Training the Chatbot:
a. Training the chatbot involves creating intents and adding expressions to help the
bot understand user inputs.
 4. Managing Conversation Flow:
a. Skills are created to manage the conversation flow, ensuring the chatbot
responds appropriately to different user inputs.
5. Testing:
a. The chatbot is tested by chatting with it to ensure it recognizes and responds to
various inputs correctly.

Table
Article Title Problem Addressed Methods Used Results Obtained
Chatbot using NLP Development of NLP techniques, Improved user
chatbots using NLP machine learning, interaction, increased
dialogue efficiency, scalability,
management customization
Developing an Automation of Machine learning Increased efficiency,
Enterprise Chatbot specific business models, RAG accurate responses,
using Machine processes scalability,
Learning Models: A customization
RAG and NLP based
approach
Build Your First Creation of a simple SAP Conversational Ease of use,
Chatbot with SAP chatbot AI customization,
Conversational AI integration
CHATBOT: Overview of chatbot Key components, Improved user
Architecture, Design, architecture, design, design and interaction, increased
& Development and development development efficiency, scalability,
methodologies customization