Description
Companies have increasingly started adopting chatbots as a customer service support system to provide more personalized service experience to customers. This technology is catching attention in the customer service context because chatbots can support multiple customers at one, have the ability to provide multi-lingual support, they are available 24*7 with no extra cost, relieves humans from repetitive task and can respond to a variety of troubleshooting queries without any latency. This work aims at developing a customer support chatbot for the companies. The chatbot is developed in web frame work and spaCy NLP library for Python has been employed to understand utterances in English language.
TABLE OF CONTENTS
COVER PAGE
TITLE PAGE
APPROVAL PAGE
DEDICATION
ACKNOWELDGEMENT
ABSTRACT
CHAPTER ONE
INTRODUCTION
- BACKGROUND OF THE STUDY
- PROBLEM STATEMENT
- AIM/OBJECTIVE OF THE STUDY
- RESEARCH QUESTION
- MOTIVATION OF THE STUDY
- SCOPE OF THE STUDY
- DEFINITION OF TERMS
CHAPTER TWO
LITERATURE REVIEW
- CUSTOMER RELATIONSHIPS
- BENEFITS OF AUTOMATION CUSTOMER SERVICE, SALES AND MARKRTING MESSAGES
- OVERVIEW OF CHATBOT
- HISTORICAL BACKGROUND OF CHATBOT
- DEVELOPMENT OF CHATBOTS
- APPLICATION OF CHATBOTS
- TYPES OF CHATBOTS
CHAPTER THREE
METHODOLOGY
- THE CHATBOT CONCEPT
- SYSTEM BLOCK DIAGRAM
- System design of the chatbot
- CHATBOT SOFTWARE APPLICATION
- DESIGN AND IMPLEMENTATION
- DESIGN PROCESS
- FINAL UI AND FUNCTIONALITY
- EVALUATION OF THE DESIGN
- EVALUATION
CHAPTER FOUR
RESULT ANALYSIS
4.1. RESULT ANALYSIS
4.2 DISCUSSION
CHPATER FIVE
CONCLUSION, FUTURE AND REFERENCES
- CONCLUSION
- SUMMARY
- FUTURE WORK
- REFERENCES
CHAPTER ONE
1.0 INTRODUCTION
1.1 BACKGROUND OF THE STUDY
The chatbot is often referred to as a “conversational interface” or “conversational user interface”, indicating a technology that enables communication between people and information systems (ISs), using a human language (Bahdanau et al., 2017). While chatbots are usually associated with text‐based communication in a messenger or chat window (Bahdanau et al., 2017), conversational interfaces can also include voice assistants integrated into smart or wearable devices, smartphones, social robots, autonomous vehicles, and other devices. Over the past decade, the growth in the successful use of chatbots has often been linked to advances in the development of AI technology. Virtual assistants have been developed for a range of different functions but all use AI technologies and natural language processing (NLP) algorithms (Gregori, 2017). For the past decade, new approaches based on non‐lin‐ ear neural networks have become increasingly evident, and the use of recurrent neural networks (RNNs) has demonstrated significant success in solving typical NLP problems (Gregori, 2017). Molnár et al. (2018) have shown how competitive pressures in global markets and rapid advances in hardware performance—making new technologies increasingly accessible—have led to a significant increase in interest in the potential of AI. In 2017, Qiu et al. (2017) introduced BERT (Bidirectional Encoder Representations from Transformers). This model significantly improved conversational interface performance, and for chatbot developers, BERT provided new opportunities to solve applied NLP tasks typical of most chatbots: user intent recognition, question answering, and classification of what? (Qiu et al., 2017). The subsequent improvements in Large Language Models (LLMs) and related text generative models have had a significant impact on the development of chatbots. The latest and most significant development in the field of NLP is the creation of the GPT (Generative Pre‐trained Transformer) model by OpenAI in 2018. It shows excellent results in tasks involving the processing of human‐language text. The GPT model is trained on a wide range of open text data, including Wikipedia, and is able to generate coherent text of almost indistinguishable quality from human‐written text. GPT‐4,released in March 2023, is the latest and most advanced version of the GPT family of models to date. It is a multimodal model that works not only with textual information but also with images and has pioneering abilities to handle a wide range of real‐world tasks performed by humans (Sultan et al., 2014).
Chatbots are attractive to businesses as they facilitate customer service through a conversational style that is natural and intuitive for humans (Sultan et al., 2014). Despite the rapid development of technology in this area, the increasing capabilities of AI algorithms and the growing number of successful real‐world applications, chatbot implementation projects face a number of difficulties. The expectation of quick and immediate success from chatbot implementation that many companies had in the middle of the last decade has been replaced by a more sceptical attitude towards the potential of virtual assistants. Practical experience has shown that chatbots cannot be used effectively in all situations, whilst creating a high‐quality service requires significant investment. The technological complexity and unpredictability of project outcomes present additional difficulties for implementation teams, who often lack practical experience and theoretical knowledge (Sultan et al., 2014).
This research identifies common patterns and rules that can be useful for a wide range of organisations whose activities are related to customer service, which can be transformed into a client – chatbot interaction, benefiting both the company and the client.
1.2 STATEMENT OF THE PROBLEM
Nowadays, most companies that offer products or services have helpdesk teams available for different purposes. The existence of these teams is expensive and cannot guarantee that all users are served immediately or in a short time frame. In company, there is a lack of customer query management in customer care facilities provided. A customer may have number of queries regarding the services offered. For example, a consumer may want to know the process of obtaining new service connection, list of requisite documents to be enclosed with new connection application form and the charges payable for new service connection in a particular consumer category. Customers who contact customer care center to get answers of their queries may have to wait for the responses. Web sites and mobile apps provide links to PDF files that contain frequently asked questions to answer common queries but customers have to navigate the entire content of the PDF file so that they can extract the information they are interested in.
This work attempts to overcome the challenge of processing the customer queries by creating a chatbot. A chatbot is a software application that comprises of conversational interface and a natural language processor. Using conversation interface, the end user interacts with chatbot using natural language phrases (Sultan et al., 2014). The natural language processor understands the meaning of the input text and then generated an appropriate response. In this work, the purpose of chatbot is to provide efficient, cost effective and consumer friendly services to the consumers. Consumers can chat with the bot to inquiry about the services they seek to avail and register a request/complaint. With chatbot, consumers will get exactly what they want.
1.3 OBJECTIVES OF THE STUDY
The objectives of the chat system includes are as follows:
- To develop an internal chatting system that applies peer-to peer concept and applies multicast technique.
- To develop a system that will automatically extract messages from the company’s customer support software.
- To develop a system that will able to reply to users in real-time.
1.4 RESEARCH QUESTIONS
This study addresses the following research questions:
- What are the main challenges which companies face when implementing chat‐ bot projects?
- What are the critical success factors (CSFs) that contribute to the successful implementation of chatbots in companies?
- What are the advantages of chatbots to organisations?
1.5 MOTIVATION OF THE STUDY
One of the most common tasks of a helpdesk team consists in solving repetitive problems or replying to questions that have already been answered somewhere in the past. This takes up valuable staff time and could be solved with the existence of a chatbot. Other scenarios include cases where users need to access documentation pages, wasting time that helpdesk teams could be dedicating to other tasks. The process of searching for documentation could be done by a chatbot in an efficient and simple way. The existence of a chatbot allows users to get replies with a low response time compared to a human.
The main motivation behind the development of a chatbot to assist the helpdesk team of a consulting company lies in the possibility of having an agent capable of integrating the helpdesk team and providing support to all employees, thereby freeing the helpdesk team to work on other tasks. Besides freeing the helpdesk team, another motivation for the implementation of a chatbot involves the possibility of employees getting faster and more straightforward answers to a set of common problems and questions.
Alongside low response times, a chatbot allows users to be more productive by automating certain tasks. Another advantage of a chatbot is its relatively low maintenance cost compared to a helpdesk team with the same level of availability and the fact that it allows the helpdesk team to focus on more important
1.6 SCOPE OF THE STUDY
The Scope of this work covers developing a scalable and easily maintainable multilingual chatbot solution that can interface with the company’s existing customer support software. It should serve as the first communication layer in the company’s customer service and provide a way for users to send free-form text messages to help them with the most common issues they encounter regarding the company’s products. Chatbot for customer support services replicate the menial tasks performed by customer service agents on a daily basis. Ideally, the system should:
- Provide means to automatically extract messages from the company’s customer support software.
- Process extracted data into a suitable format for machine learning solutions.
- Provide a scalable machine learning pipeline for the system’s models, starting from the raw data.
- Be designed in such a way that it can be deployed on multiple platforms with little effort.
- Be able to reply to users in real-time.
- Help users with a subset of problems that they often experience. • Ask the user for more information if their intent is not clear enough.
- Request human intervention when appropriate.
- Allow tuning of the bot’s responses.
- Support multiple languages
1.7 DEFINITION OF TERMS
Automated Customer Service: The process of using canned replies, email automation, and other techniques in order to scale the capacity of a customer service department to serve a business’ customer base.
Application programming interface (API) – A set of definitions, protocols and tools to build application software.
Artificial intelligence (AI) – AI is the creation of computer systems that can perform tasks that require human intelligence, such as speech recognition, understanding and translating language, and decision making. AI includes computer science, neuroscience, psychology and linguistics, which are all needed by a machine to duplicate a typical human response.
Bots (see Chatbots) – A short name for computer programs that interact intelligently with humans and the internet.
Broadcast – A message that is proactively sent to users, similar to a push message in a mobile app.
Channel – The medium for chatbot conversations. Channels include Your Website, Messenger, Slack, Skype, SMS, email and web chat windows.
Chat Logs – chat data of human bot conversions.
Chatbot – A chatbot is an intelligent computer program that interfaces among humans, computer systems, and the internet. The word is a contraction of chat and robot.
Cloud or cloud computing – Internet-connected computers that share computer processing resources and data with other computers on demand.
Context – Contextual data is information from the chatbot related to specific conversations and can have a relative importance.
Conversational UI – A user interface based on human speech or language.
Conversations – This is a decision tree or logic diagram of a scripted conversation. These conversations can be linear or they have branching logic with multiple answers to questions.
Deep Learning – Algorithms used in machine learning and artificial intelligence to gain insights.
Entity – An entity modifies an intent. For example, if a user types “show me yesterday’s financial news”, the entities are “yesterday” and “financial”. Entities are given a name, such as “dateTime” and “newsType”. Entities are sometimes referred to as slots.
Framework – A structure that provides building blocks and functionality to build a chatbot that also requires programming.
Human in the Loop – Chatbots learn by collecting and monitoring data from conversations. The AI system applies what it learns from each conversation. Human in the Loop is when a human has to monitor some of the AI responses to make sure the appropriate response is given.
Intent – An intent is the user’s intention to gather a specific piece of information, such as the daily weather forecast. The intent is usually a noun-verb word combination that tells the chatbot what the human wants it to do. For example, find an ATM, book an appointment, or order food.
Interaction – A verbal or written communication between a chatbot and a human.
Interface – A shared boundary where two or more parts of a computer system exchange information.
Machine Learning (ML) – Machine learning is the process where a computer learns from experience rather than from programming. The machine learns by gathering data and it can find insights from that data without being explicitly programmed.
Multiple intents – When a user makes a complex request to the chatbot and the chatbot has to process and prioritize two or more intents simultaneously.
Natural language processing (NLP) – Natural language processing is teaching a computer to understand language and the intent behind the language. NLP is based on artificial intelligence, computer science, and computer linguistics.
Natural language processing Engineer – a person that specializes in NLP, AI, ML and chatbots
Pilot – The development stage of a chatbot where it is deployed to small group of testers.
Rails, Guard Rails – When a user asks a question that stumps the chatbot or is beyond the scope of the chatbot, the conversation must be redirected to a human. This redirection is called a Rail or a Guard Rail.
Response – Any reply from a chatbot based on user input.
Structured data – Highly organized information that is searchable when put in a database.
CHAPTER FIVE
5.1 CONCLUSION
The aim of this work was to provide an overview of the process of creating and evaluation of an automatic customer service, sales and marketing messages application using chatbot. As a result, chatbot application was created. The process was straight-forward: it started out with design process, moved quickly into implementation and lastly, evaluation was done. The main research question the thesis seeks to answer is:
The requirements found in the thesis were interactivity, enhancing customer relationships and satisfaction, good quality and fulfilling a need. Literature review shows us how these requirements bind together. Interactivity and good quality of an application increase customer satisfaction which is closely related to customer relationships. Customer relationships should be handled carefully to maintain one’s business.
5.2 SUMMARY
Chatbots are still considered an emerging technology, but they are quickly maturing and becoming a staple in many businesses’ customer service, sales, and marketing operations.
That said, a lot of current generation chatbots are still in their infancy – they have big ambitions but lack the experience and expertise needed to truly deliver. The path to greatness is not a forgiving one.
As artificial intelligence, machine learning, and deep neural network application matures, each new generation of chatbots is bound to be better and better.
During this transitional phase, it’s important for business owners to understand that poorly executed chatbot protocols can still fall short and offer a poor customer experience.
Brands should focus on implementing chatbots in a way that simplifies existing processes rather than going for the fanciest promises.
Offer guidance to customers, set expectations, and teach customers to interact with chatbots in a way that is ultimately beneficial to the growth of your brand.
5.3 FUTURE WORK
Preferably, application would have a live feed feature. This would show website visitors in real time on a map (or a list) and users could start a chat with a visitor instantly from that view. Currently, a chat has to be started by the website visitor. This would, we believe, increase sales and improve user experience on the website.
At the moment, the login screen can be used to log in to chatbot with existing account. We think it needs a feature to start a chatbot trial. This is important as the application is put to distribution to Google Play. If a user doesn’t have an account, it must be easy to create one from the app directly.
Security could be increased in the future. By switching to HTTPS from HTTP, security could be increased notably. Also, all requests could be hashed whereas now only password is hashed in the login call.
The feedback survey received can be used to think of future improvements. Few users were missing file sending, a user would have liked that chat groups were distinguished in the main list, and another user wanted to be able to send a conversation to email. These could be valuable features to be implemented in the future.
In general, we are pleased with the application. There was discussion that we would develop an iPhone version simultaneously with Android version. With the time given, this was the right choice to make the Android version first and then, if needed, concentrate on other platforms. As writing this, iPhone version is not under development.
