Professional Documents
Culture Documents
Group Members:
Muhammad Ibrahim BB-6451
Huzair Nadeem BB-6426
Muhammad Hassan Ali Khan BB-6433
Abdul Basit BB-6252
Course & Code: Computer Vision (CS-448)
Instructor: Sir Adnan
ACKNOWLEDGMENT:
This vote of thanks and acknowledgment is given to all those
who helped us during the period of our final year project. We
wish this special occasion to thank and dedicate our work to
our Family. From the start till the end, their prayers, blessings,
and suggestions for the odds have been the unconditional
capability to us. We are profoundly grateful to our Supervisor
Sir Muhammad Adnan Kaim Khani for all the support and
boosting and the way that he has shown us in the complicated
days of our project. Without him, we could never have
visualized making it to the end. Our final year project would not
have arisen without the energetic achievement and restless
hardworking of our Project. It has been our good luck and
massive prosperity to be associated with such a visionary and
liberal personality. He has been the illumination of our path
and the innovation of our ideas. His suggestions, guidance,
dedication, advisers, support, affection, and calmness towards
us, are the backbone of this project. We can never thank him
enough for being the right chief and man behind the
innovativeness of our Project of the final year.
ABSTRACT:
Customer Service chatbot are based on human
dialogue, In this project we explore Multi-Turn Queries
(MTQ) in e-commerce sites. Simple that Computer talks
like human in conversation. So we create MTQ chatbot, a
smart system that keeps track what users say. It means the
user formal reply then MTQ chatbot correct wording and
add more replies back. So it will adapt and respond better.
We will test it and make sure it works well in real
conversation and ask user feedback to keep making it
even better. The main goal is talking our MTQ chatbot just
like a friend.
Introduction:
In Customer Service Chatbot, making computer
conversation sound like human is a main focus of this project. This
project explores Multi-Turn-Queries (MTQ) in the context of
Online Shopping. We aim to make the computer talk in a way that
feels like chatting with human. The MTQ chatbot main strength is
understanding and remembering what user say in a conversation
and this help it give better and more fitting response.
The MTQ chatbot work just like smart
system. The user ask for any object of online shopping and then
chatbot give the answer of this object and further detailed of this
object. When user use formal language, the MTQ chatbot does
more than say "Okay". It fixes words and adds more information,
making the conversation evolve and feel personalized. This
chatbot aims to enhance user satisfaction by delivering responses
that align more closely with individual communication styles.
We'll thoroughly test the MTQ chatbot by
creating situation that mimic real conversation, ensuring it works
well in practical scenarios. We will ask user what they think and
use their feedback to make their chatbot better. This will happen
regularly, so the chatbot keep getting improved all the time. The
ultimate goal of this project is to create an MTQ chatbot that not
only fulfills its functional purpose but also establishes a friendly
connection, and we want users to feel like they are talking to a
friend when using it. Through the exploration, Multi-Turn-Queries
applied for this chatbot and detailed relevant response and
making interaction not only efficient but also enjoyable and
human-like.
LITERATURE REVIEW:
Advances in artificial intelligence (AI) and
natural language processing (NLP) technologies have significantly
changed the customer service landscape, especially in the e-
commerce sector. A major development in this area has been the
introduction of multi-turn queries (MTQ) chatbots, designed to
mimic human-like conversations more effectively by remembering
and adapting the context of the conversation over multiple turns.
This literature review explores the foundation, progress, and
current status of MTQ chatbots with a focus on their application in
e-commerce customer service.
Evolution of Chatbots: Initially, chatbots were based on simple
rule-based systems that could only handle single-turn interactions
based on pre-defined scripts. The advent of machine learning and
NLP has enabled the development of more sophisticated models
that understand and generate human-like text, leading to the
creation of MTQ chatbots.
Importance of Context in Conversations: One of the key
challenges in developing effective chatbots is the ability to
maintain context over multiple turns of conversation. This involves
not only understanding the intent behind a user's query but also
remembering previous interactions within the same session to
provide coherent and contextually relevant responses.
Core Technologies
Natural Language Understanding (NLU) and Generation (NLG)
NLU is crucial for interpreting user intent, while NLG allows the
chatbot to generate human-like responses. Advances in deep
learning have significantly improved these capabilities, enabling
more nuanced conversation.
Dialogue Management: This involves managing the state and
flow of the conversation, ensuring that the chatbot can handle
transitions between topics smoothly and maintain context over
multiple exchanges.
Memory Networks: These are specialized neural network
architectures that enable chatbots to remember information over
long sequences of dialogue, essential for effective MTQ handling.
Applications in E-Commerce
Personalized Customer Service: MTQ chatbots can provide
personalized assistance by remembering user preferences and
past interactions, thereby enhancing the shopping experience.
Efficiency and Scalability: By automating multi-turn
conversations, e-commerce businesses can handle a larger
volume of customer queries without compromising on the quality
of service.
Feedback Loop: Incorporating user feedback into the chatbot's
learning process allows for continuous improvement, ensuring
that the chatbot becomes more effective over time at
understanding and responding to user needs.
Turn-3 A: Oh.
Turn-4 B: Hum!