15SE496L – MAJOR PROJECT

ChatBot for Disease prediction using TF- IDF

A Project Report
By
Saurabh Kumar Gupta Mirza AbduzZaheer
(RA1711020010094) (RA1711020010108)
Under the guidance of
Dr. M.S. Abirami
(Assistant Professor, Department of Software Engineering)
In partial fulfillment for the award of the degree
of
BACHELOR OF TECHNOLOGY
in
SOFTWARE ENGINEERING

FACULTY OF ENGINEERING AND TECHNOLOGY

SRM INSTITUTE OF SCIENCE AND TECHNOLOGY
Kattankulathur, Kancheepuram

May 2021

1
 BONAFIDE CERTIFICATE

Certified that this project report “ChatBot for Disease prediction using TF- IDF” is
the bonafide work of Mirza AbduzZaheer(RA1711020010108) and Saurabh Kumar Gupta
(RA1711020010094) who carried out the project work under my supervision.

Dr. M.S. Abirami (Guide) Dr. C. Lakshmi

Assistant Professor Head ofthe Department
Department of Software Engineering Department of Software Engineering
SRM Institute of Science and Technology SRM Institute of Science and Technology

INTERNAL EXAMINER EXTERNAL EXAMINER

2
 ACKNOWLEDGEMENT

We express our humble gratitude to Dr.C. Muthamizhchelvan, Vice Chancellor in-charge / Pro Vice
Chancellor, Faculty of Engineering & Technology, SRM Institute of Science and Technology, for the
facilities extended for the project work and his continued support.

We extend our sincere thanks to Dr. RevathiVenkataraman, Professor & Chairperson, School of
Computing, SRM Institute of Science and Technology, for her invaluable support.

We wish to thank Dr. C Lakshmi, Professor & Head, Department of Software Engineering, SRM
Institute of Science and Technology, for her valuable suggestions and encouragement throughout the
period of the project work.

We are extremely grateful to our Year Advisor Dr. J. S. FemildaJosephin, Associate Professor, and
the project coordinators Dr.M.Uma, Assistant Professor, Mrs. J.Jeyasudha, Assistant
Professor,Department of Software Engineering, SRM Institute of Science and Technology, for their
great support at all the stages of our project work. We would like to convey our thanks to our Panel
Head Dr. M. S. Abirami, Assistant Professor, Department of Software Engineering, SRM Institute of
Science and Technology, for her inputs during the project reviews.

We register our immeasurable thanks to our Faculty Advisor, Dr. D. Vivek, Assistant Professor,
Department of Software Engineering, SRM Institute of Science and Technology, for leading and
helping us to complete our course.

Our inexpressible respect and thanks to our guide, Dr. M. S. Abirami, Assistant Professor,
Department of Software Engineering, SRM Institute of Science and Technology, for providing me an
opportunity to pursue my project under her mentorship. She provided me with the freedom and
support to explore the research topics of our interest. Her passion for solving real problems and
making a difference in the world has always been inspiring.

We sincerely thank all the staff members and students of the Software Engineering Department, SRM
Institute of Science and Technology, for their help during our research. Finally, we would like to
thank our parents, our family members and our friends for their unconditional love, constant support
and encouragement.

Saurabh Kumar Gupta Mirza AbduzZaheer

(RA1711020010094) (RA1711020010108)

3
 ABSTRACT

To start an good life healthcare is more important. But it's very difficult to get the consultation
with the doctor just in case of any health issues. The proposed idea is to make a health care Chabot
system using AI which will diagnose the disease and supply basic details about the disease before
consulting a doctor.

ChatBot will provides which sort of disease you've got based on user symptoms and appeared
doctor details respective to user disease. Our chatbot will also provide analgesics for further treatment
and also food suggestion meaning which sort of food you've got to require. The user will be able to
take advantage of a chatbot only when it can diagnose all kinds disease and supply necessary
information. A text-to-text diagnosis Bot engages patients in conversation about their medical issues
and provides a customized diagnosis supported their symptoms. Hence, people will have a thought
about their health and have the proper protection.

In this Project we have build deep neural Networks using 3 layers, in this model we have used
keras Sequential API. As this is a retrieval based healthcare chatbot, we have achieved more than
90% accuracy by training the model to 200 epochs. Here we have used special recurrent neural
network (LSTM) to classify which category the user input message belongs to and then bot will give
random response from the set of responses.

4
 TABLE OF CONTENTS
Chapter TITLE Page
No. No.
1 INTRODUCTION 9
2 PROJECT OVERVIEW 11
2.1 LITERATURE SURVEY
2.2 PROBLEM DESCRIPTION
2.3 REQUIREMENTS GATHERING
2.4 REQUIREMENT ANALYSIS
2.4.1 FUNCTIONAL REQUIREMENTS
2.4.2 NON- FUNCTIONAL REQUIREMENTS
2.5 DATA SOURCE
2.6 COST ESTIMATION
2.7 PROJECT SCHEDULE
2.8 RISK ANALYSIS
2.9 SOFTWARE REQUIREMENTS SPECIFICATION
3 ARCHITECTURE & DESIGN 23
3.1 SYSTEM ARCHITECTURE
3.2 INTERFACE PROTOTYPING (UI)
3.3 DATA FLOW DESIGN
3.4 USE CASE DIAGRAM
3.5 SEQUENCE DIAGRAM
3.6 CLASS DIAGRAM
3.7 INTERACTION DIAGRAM
3.8 STATE / ACTIVITY DIAGRAM
3.9 COMPONENT & DEPLOYMENT DIAGRAM
4 IMPLEMENTATION 30
4.1 DATABASE DESIGN
4.1.1 ER DIAGRAM
4.1.2 RELATIONAL MODEL
4.2 USER INTERFACE
4.3 MIDDLEWARE
5 VERIFICATION & VALIDATION 37
5.1 UNIT TESTING
5.2 INTEGRATION TESTING
5.3 USER TESTING
5.4 SIZE - LOC
5.5 COST ANALYSIS
5.6 DEFECT ANALYSIS
5.7 MC CALL’S QUALITY FACTORS

5
Chapter TITLE Page
No. No.
6 EXPERIMENT RESULTS & ANALYSIS 41
6.1 RESULTS
6.2 RESULT ANALYSIS
7 CONCLUSION & FUTURE WORK 43
8 PLAGARISM REPORT 44
9 REFERENCES 45
ANNEXURE 47
CODING

6
 List of Tables

Table No. Table Name Page No.

2.1 Project Schedule 17
5.1 Unit Testing 37
6.1 Predicted Results 41
6.2 Result Parameters 42

7
 List of Figures

Figure No. Figure Name Page No.

2.1 Data Sets 16
2.2 Cocomo Model 16
3.1 Healthcare Chat using Sequential Model 23
3.2 User Interface Prototype for Chatbot 24
3.3 Data Flow Diagram 25
3.4 Use Case Diagram 26
3.5 Sequence Diagram 26
3.6 Class Diagram 27
3.7 Interaction Diagram 27
3.8 State/Activity Diagram 28
3.8 Component & Deployment Diagram 29
4.1 Software Development Lifecycle 30
4.1 Joining Datasets to create Single Flat file 31
4.3 Target Flat File in WinSCP 32
4.4 Training Datasets 33
4.5 Entity Relationship Diagram 35
4.6 User Interface 35
6.1 ROC Curve 42

8
 CHAPTER 1
INTRODUCTION

Now a days, health care is very important in our life. Today‘s people are busy with their
works at home, office works and more addicted to Internet. They are not concerned about their health
.So they avoid to go in hospitals for small problems. it may become a major problem. So we can
provide an idea is to create a health care ChatBot system using AI that can diagnosis the disease and
provide basic information about the disease before consulting a doctor. Which helps the patients
know more about their disease and improves their health User can achieve the all kind of disease
information. The system application uses question and answer protocol within the sort of ChatBot to
answer user queries. The response to the question are going to be replied supported the user query.
The significant keywords are fetched from the sentence and answer to those sentences. If match is
discovered or significant answer are going to be given or similar answers are going to be displayed.
Bot will diagnosis which type of disease you have based on user symptoms and also gives doctor
details of particular disease.It may reduce their health issues by using this application system. The
system is developed to scale back the healthcare cost and time of the users because it isn't possible for
the users to go to the doctors or experts when immediately needed.

Artificial Intelligence is predicated on how any device perceives its Environment and takes
actions supported the perceived data to realize the result successfully. It is the study of intelligent
agents. The term "artificial intelligence" is applied when a machine mimics "cognitive" functions that
humans accompany other human minds, like "learning" and "problem solving. Artificial Intelligence
gives the supreme power to mimic the human way of thinking and behaving to a computer. A
ChatBot may be a computer virus which conducts a conversation via auditory or textual methods.
These programs are designed to supply a just like how a person's will chat and thereby it acts
as a interlocutor instead of humans. For various practical purposes like customer service or
information acquisition, , ChatBot is getting used within the dialog system. Today, ChatBots are a
part of virtual assistants like Google Assistant, and are accessed via many organizations' apps,
websites, and on instant messaging platforms. Non-assistant applications include chat bots used for
entertainment purposes, for research, and social bots which promote a particular product, candidate,
or issue.

ChatBot‘s are such quite computer programs that interact with users using natural languages.
For all kind of chat bots the flow is same, though each ChatBot is specific in its own area knowledge
9
that is one input from human is matched against the knowledge base of ChatBot. ChatBot‘s work
basically on Artificial intelligence so using this capability we have decided to add some contribution
to the Health Informatics. The high cost of our healthcare system can often be attributed to the
shortage of patient engagement after they leave clinics or hospitals. Various surveys during this area
have proved that that ChatBot can provide healthcare in low costs and improved treatment if the
doctors and therefore the patient confine touch after their consultation. To answer the questions of the
user ChatBot is employed. There is very less number of ChatBots in medical field.

Computers give us information; they engage us and help us during a lot of manners. A chatbot
may be a program intended to counterfeit smart communication on a text or speech. These systems
can learn themselves and restore their knowledge using human assistance or using web resources.
This application is incredibly fundamental since knowledge is stored beforehand. The system
application uses the question and answer protocol within the sort of a chatbot to answer user queries.
This technique is developed to scale back the healthcare cost and time of the users, because it isn't
possible for the users to go to the doctors or experts when immediately needed.

The response to the question are going to be replied supported the user query and knowledge
domain. The featured keywords are fetched from the sentence and answer to those sentences. If the
match is discovered or the many answer are going to be given or similar answers are going to be
displayed.

10
 CHAPTER 2
PROJECT OVERVIEW
2.1 Literature Survey
Title 1:Healthcare ChatBot (IRJET Nov 2020)
Authors:Papiya Mahajan, RinkuWankhade, AnupJawade, PragatiDange
Methodology: The system uses an expert system to answer the queries. User can also checkout the
available doctors for that particular disease. This system can be used by the many users to get the
counselling sessions online. The data of the ChatBot stored in the DB is in the form of pattern-
template. Bot will provide prescribed medicines and food suggestions that means which food you
should take based on the severity disease.
Limitations: Less combination of words in the model. Less use of database information, so the
medical ChatBot may not handle all type of disease.

Title 2: Text Messaging-Based Medical Diagnosis Using Natural Language Processing and Fuzzy
Logic (Hindawi Journal of Healthcare Engineering volume 2020)
Authors: Nicholas A. I. Omoregbe, Israel O. Ndaman, Sanjay Misra, Olusola O Abayomi
Methodology: The study assesses the clinical data needs and requirements in diagnosing the tropical
diseases in Nigeria and assesses the patients’ clinical data found in EHRs or manual records. The
proposed text-based medical diagnosis system are as follows:
(1) description of the knowledge base;
(2) preprocessing of text-based documents;
(3) tagging of document;
(4) extraction of answer;
(5) ranking of candidate answers
Limitations: Lack of automation of this medical diagnosis system to easily recognize diseases,
recommend treatments, prescribe a medication, and per- form medication adherence. No Audio
interaction in- corporate to make the system more interactive

Title 3: Ai Healthcare Interactive Talking Agent using NLP (IJITEE, November,2019)

Authors: M.S Bennet Praba, Sagari Sen, Chailshi Chauhan, Divya Singh
Methodology: Building of ChatBot by python using concept of morphology. It analyses the
information provided by the user and responds according to an individual's requirement of diet plan

11
and exercises. The morphology concept enables the bot to work in a more efficient way as it
recognizes the meaning of each part of the word individually and thus it is able to rectify errors.

Limitations: The disadvantage of this ChatBot is that if the inputs are not in sequential order as our
text file then the responses may be wrong.

Title 4: Joint POS Tagging and Dependency Parsing with Transition based Neural Networks(2019)
Authors: Liner Yang, Meishan Zhang, Yang Liu, Maosong Sun, Nan Yu, Guohong Fu
Methodology: In this method, the system database maintains a dataset of synonyms for important
keywords in that domain. The sentence sent by the user is then mapped on to that synonym dataset.
The keywords detected from the sentence are then checked in that synonym set to check for same
intent. All possible synonyms of that keyword are then looked out for a match in the main database.
The sentence which is closest to the user sentence is extracted.

Limitations: This method is time consuming and requires more of storage and complexity.

Title 5: ChatBot using support vector machine(2019)

Authors: Qiping Yang, Jingui Qin, Yongjie Huang, WushaoWao
Methodology: In this work, SVM algorithm is used which is support-vector machines (SVMs, also
support-vector networks) are supervised learning models with associated learning algorithms that
analyze data used for classification and regression analysis.

Limitations: Inability to Understand – Due to fixed programs, ChatBots can be stuck if an unsaved
query is presented in front of them.

Title 6: Task-based Interaction ChatBot(2018)

Authors: Dana Doherty
Methodology: It was determined that ChatBots perform at a very high standard and provide reliable
and rapid responses to users compared to that of traditional methods. The average time spent
interacting with the ChatBot applications are very low as it provides an efficient way for users to
manage their banking. The ChatBot has proved that it can fulfil the demand of users wanting instant
access and availability information and services.

Limitations: It takes more time to response to the user questions. User have to pay some charges to
perform live chat

12
2.2 Problem Description
The healthcare industry is one among the most important sectors across the world- both
economically and employment-wise, making it one among the busiest industries within the world.
With this height of the chaos, demand for speed and efficiency is important for contemporary patients
who want quick and straightforward access to information.

ChatBots have the potential to revolutionize healthcare. They will substantially boost
efficiency and improve the accuracy of symptoms collection and ailment identification, preventive
care, post-recovery care, and feedback procedures. Healthcare ChatBots, by their very nature, will
drive the transformation that triggers this alteration.

With the technological advancements of AI, chatbots have begin to be an excellent tool for
quick and straightforward automation. Ever since they grew popular among the E-commerce and
Customer service sectors, many industries are arising with ways to implement this technology in their
businesses, including healthcare. In fact, there are many chatbot companies beginning with innovative
use-cases for his or her bots.

Healthcare Chatbots are conversationalists that run on the principles of machine learning,
which comes under AI. along side completing interactions, they also perform repetitive tasks like
providing solutions, sending emails, marketing, lead generation, result analysis, and therefore the list
goes on… Now how chatbots help within the healthcare sector is automating all the repetitive, also as
lower-level tasks that a representative would do. once you divide work into two, where one takes care
of the straightforward tasks, while the opposite takes care of more complex queries, the work is
completed seamlessly.
Now you/your patients don’t got to hold in line for hours together before a representative invests time
to seem into your query, while a chatbot can do that instantly! Moreover, consistent with a study,
millennials prefer texting over calling. So chatbots seem to suit all modern-day requirements
perfectly!

Computer programs using textual conversational mediums are growing popular among
healthcare institutions/organizations. supported the market intelligence report published by BIS
Research titled Global Chatbots in Healthcare Market – Analysis and Forecast, 2019-2029, the
chatbots within the healthcare market generated a revenue of $36.5 million in 2018. These intelligent
programs are ready to detect symptoms, manage medications, and assist chronic health issues. They

13
guide people rightly for serious illness and also assists them in scheduling appointments with
professionals. With technological advancements in chatbots over the last decade, there has been
significant growth within the healthcare sector, along side other AI tools.

2.3 Requirements Gathering

2.3.1 ChatBots in Industry
Most businesses and organisations are understanding the potential benefits of machine
learning and AI to possess a positive change on how they perform business. AI has progressed to
permit the event of more sophisticated ChatBots. Organisations are that specialize in specific areas of
user engagement that take up tons of your time but are often replaced through the utilization of a
ChatBot. ChatBots can understand what the customer needs from one text rather than the customer
having to follow a process of multiple steps.
ChatBots are wont to automate customer service and reduce manual tedious tasks performed
by employees in order that they can spend their time more productively on higher priority tasks.
Establishments that often affect its customers have discovered the potential of ChatBots as a channel
to distribute more efficient and immediate information to customers as compared to a customer
service representative regarding queries and issues (Onufreiv, YY. , 2017).
There are two modes during which ChatBots can simulate a conversation with users which
include : System-initiated ChatBots where– they commence the conversation with the user and User-
initiated ChatBots where- the user directs the conversation instead. Systems that incorporate the 2
methods of initiation are referred to as mixed initiative systems (Duijst, D. 2017).

2.3.2 Natural Language Understanding Engine

The ChatBot engine is assumed of together of the foremost critical elements of a ChatBot,
alias “Natural Language Understanding (NLU) engine” ( Kar, R and Haldar, R. 2016). The NLU
holds liability for the interpretation of conversational dialogs to actions which are understood by the
machine. NLU engines use a spread of AI methods to know the tongue utilized in conversational
interfaces like ChatBots. These methods consist of: tongue Processing (NLP) and Machine Learning
(ML) (Kar, R and Haldar, R. 2016).
Googles Dialogflow, previously known API.ai, may be a tongue understanding engine that
identifies the intent and context from the tongue in user supplied utterances. These concepts are wont
to develop the behaviour of the ChatBot and the way coherently it interacts with the user. Intents are
wont to establish a connection between the user input and therefore the appropriate action to be
executed by the ChatBot so as for the user to realize their goal. Contexts are utilised to distinguishand

14
recognize user input which can have an alternate meaning depending upon the present conversational
context (Gregori, E. 2017).

2.3.3 Artificial Intelligence

“Artificial Intelligence is neither a replacement technology nor a machine”. AI is that the
recognition of outcome-direction which is that the rapid analysis of live data to realize the expected
goal. Outcome-directed thinking splits from the confines of the rule-directed approach that's
accomplished through AI .
The generalised practice of AI are often weakened into an easy process which doesn't require
an experienced level of proficiency to know . First of all, a numerical illustration is established for the
target or outcome. Specific data is then related to the target is gathered and conditions and behaviours
are investigated to extend the likelihood of achieving the expected target. Multiple aspects can
determine the result .the load of every aspects effect is computed. “AI uses the relative weighting of
every aspect to make a prediction (evaluation) formula” (Yano, K. 2017).
Lastly, the formula developed from the weighted aspects are employed to business decisions (Yano,
K. 2017). AI are often classified into four groups: “systems that think like humans, systems that act
like humans, systems that think rationally and systems that act rationally” ((Russell, S.J and Norvig,
P. 1995).
AI is usually categorised as strong and weak AI: strong AI is that the production of human-
like intelligent systems. Weak AI would be the mixing of intelligent algorithms embedded within a
system. “Machine learning, deep-learning, tongue processing and neural networks are often
summarised under the term of AI” (Exner-Stöhr, et al., 2017)

2.4 Requiremnts Analysis

2.4.1 Functional Requirements
Collecting Dataset: Creating a dataset for the training of the models is the primary task.
Datasets can be created manually and then augmented.
Pre-Processing: sending the dataset to the model, the dataset should be pre-processed. This
includes splitting the data into three folds.
Training Models: To do an optimal phishing detection model, they should be trained and
validated.
2.4.2 Non Functional Requirements
• Accuracy
• Less training time

15
 • Simple architecture
2.5 Data Source
We have got four different datasets from kaggle, these datasets are combined into a single flat
file by creating a mapping with various transformations to meet the required data attributes for the
system.

Figure 2.1 Data sets

2.6 Cost estimation

Figure 2.2Cocomo Model

16
Cocomo model is applied in our project, as our software project is organic
Effort Applied (E) = a * ()#
Development Time (D) = c * ()%

2.7 Project Schedule

Table 2.1 Project Schedule
Time period
From Checkpoints of the project
To Date
Date

25/01/21 01/02/21 Requirement gathering and literature survey

02/02/21 09/02/21 Cost estimation and risk analysis

10/02/21 17/02/21 System architecture

18/02/21 25/02/21 Diagrams

02/03/21 09/03/21 ER diagram

10/03/21 17/03/21 Relational Model and coding

18/03/21 25/03/21 Coding and Implementation

Unit testing and Report

26/03/21 31/03/21

03/04/21 10/04/21 Final development phase

11/04/21 18/04/21 Launched Developer Beta

19/04/21 26/04/21 Testing

27/04/21 03/05/21 Report

2.8 Risk Analysis

Fixed Rule-based: Existing ChatBots are developed by using straightforward machine learning
techniques, fixed set of rules, and matching supported templates .
Grammatical Errors: Grammar mistakes can't be recognized.
17
Predefined or Closed-domain: previous studies show that the majority of the ChatBots only answer
the questions from a closed domain, or answer those questions, which are defined within the database.
Ambiguity: The meaning or the context of a sentence isn't apparent or has not any appropriate
purpose
Language Structure: The structure of sentence making differ from language to language. for
instance , each language has its own rules for punctuation, text structure, and use of spaces. While
existing ChatBots cannot distinguish it
Semantics: It means words or sentences during a human tongue format. the present ChatBots cannot
handle tongue processing whether these ChatBots only show a response, or they create the analysis of
questions.
Recommender Systems: The previous ChatBots aren't ready to advise or explain any human topic.
Even they can't ask any questions. ChatBots only gather information from the user and generate a
response from the knowledge domain . A ChatBot must be ready to create queries supported
previously answered questions.
Accuracy: The ChatBots should be designed in such how that their conversation is sort of a human to
finish any task. But existing ChatBots are bad at suddenly changing any subject and supply hit or miss
response.
Self-learning: Supervised machine learning techniques aren't utilized in previous ChatBots. they're
bad at learning the newest patterns of words or speech. they can't discover context from logical
reasoning and interaction. Most of the ChatBots cannot train any classifier to map from the sentence
to the intent and sequence model to the slot filter.

2.9 SOFTWARE REQUIREMENT SPECIFICATION

2.9.1 Problem Description
The healthcare industry is one among the most important sectors across the world- both
economically and employment-wise, making it one among the busiest industries within the world.
With this height of the chaos, demand for speed and efficiency is important for contemporary patients
who want quick and straightforward access to information.

ChatBots have the potential to revolutionize healthcare. they will substantially boost
efficiency and improve the accuracy of symptoms collection and ailment identification, preventive
care, post-recovery care, and feedback procedures. Healthcare ChatBots, by their very nature, will
drive the transformation that triggers this alteration.

18
2.9.2 Introduction
Aim
Healthcare ChatBots are conversationalists that run on the principles of machine learning,
which comes under AI. Healthcare ChatBots are conversationalists that run on the principles of
machine learning, which comes under AI.T he app answers queries raised by the users. It helps
in reducing the cost of healthcare and improving access to healthcare services. It also enhances
the user to input query and get their medical status and issues.
Document Convention
This document uses Arial as font theme for normal content and times as font theme for
subheadings. Font size of content is 11 while subheadings have a font size of 14 and they are in bold.
In addition, important text in the content is highlighted by bolding the text. Furthermore, the goals for
higher-level specifications are believed to be inherited from the comprehensive specifications.

2.9.3 Description
Product Functions
● Collecting Dataset: Creating a dataset for the training of the models is the primary task. Datasets
can be created manually and then augmented.
● Pre-Processing: sending the dataset to the model, the dataset should be pre-processed. This includes
splitting the data into three folds.
● Training Models: To do an optimal phishing detection model, they should be trained and validated.
● Benfits of Healthcare ChatBot: Medical ChatBots reduce healthcare professionals’ workload by
reducing hospital visits, reducing unnecessary treatments and procedures, and decreasing hospital
admissions and readmissions as treatment compliance and knowledge about their symptoms improve.
Characteristics
This software product can be used by whosoever who are concerned for their health.
Operating Environment
As of now this application is only available for PC users having operation system Windows 7 and
above.
Design and Implementation
We have designed our ChatBot using HTML, CSS and Javascript. We have used various
python libraries like numbpy, pandas, sklearn, TensorFlow, Keras to preprocess the data, creating
training data and testing data, and creating the sequential model using keras.

19
Assumptions and Dependencies
We have optimised our ChatBot in such a way that it uses very less RAM, which which makes
our application very flexible across all sets of devices.
Advantages of ChatBot
Availability around the clock
Doctors are vital and that they do their best to be available all the time and
dedicate enough attention to every patient. But the matter is, doctors are usually on a
decent schedule, and being available for each patient is impossible sometimes. Hence,
chatbots came into work! Healthcare Chatbots are available 24/7 and they’re personally
dedicated to assisting you throughout your recovery. While the doctors save more lives
out there, this chatbot can assist you in tasks like reminding you about your medicines,
provide medical information, offer you tips, and monitor your overall wellbeing.
Information comes handy
In the healthcare industry, emergencies are a standard thing. And time plays a
really crucial role in this! Healthcare chatbots provide instant information, especially in
times where every minute is vital. for instance , if a patient rushes in with an attack, the
doctor can get the patient’s information like previous records, other diseases, allergies,
check-ups, etc. Therefore, doctors are ready to save longer with chatbots.
Builds a rapport with patients and provides assistance
Allchatbots within the medical line need to be presented rightly and made as
attractive as possible. Now imagine your patient’s looking up for a symbol on your
website but doesn’t know where exactly to seek out the answer and the way to book a
meeting with you? They’re getting to leave your website disappointed. But a health bot
can assist you turn that around. When a patient visits your website, your chatbot will give
them a warm greeting and help them through the symptoms, predict potential
diagnosis, and supply them the choice of booking a meeting with you directly. They
also absorb user information by asking them questions, which gets stored for any sort
of reference and personalized experience. this is often how chatbots build an
honest rapport together with your patients.
Checks symptoms
In case of an emergency, it’s important that a patient is directly delivered to the
hospital. However, if things isn’t as extreme then it'd consume an excessive amount
of time if the patient travels all the thanks to the hospital or clinic. This is often where
chatbots inherit the image and save valuable time.

20
 A chatbot can now detect a patient’s ailment by –
o Asking a couple of simple questions.
o Analysing patient history.
Patients can easily check for symptoms on a healthcare chatbot and measure the
severity of things . Since the chatbot remembers individual patient details, patients
don’t need to enter an equivalent information whenever they need to urge an
update.
Provides support and extra information
People usually keep calling in with additional questions and doubts which
kills tons of your time during a busy atmosphere. Hence, healthcare chatbots also
are given the responsibility of managing such additional queries and reducing repetitive
calls.
Enhancing patient experience
Imagine how amazing it might be to possess an on-call doctor who are often there
to support you at your beck-and-call. Unfortunately, that can’t be the truth , however, we
do have an identical solution. Use a healthcare chatbot! Health chatbots typically have an
uptime of over 99.9% so whenever your patients need a solution , a chatbot is there to
offer it to you.

2.9.4 External Requirements

User Interface
A user interface is the meeting point between users and machines, it is the point where a user
interacts with the design. Depending on the type of ChatBot, software developers use a graphical user
interface, voice interactions all of which use different machine learning models/Algorithms to
understand human language and generate appropriate responses.
 The UI of the desktop software would be very interactive and familiar to other applications.
 All the elements of our ChatBot is clearly visible which makes it easy to use.
 Simple Design easy Navigantion
Hardware Interfaces
The following posts describe different kinds of hardware requirements for this software
product.
Hardware specifications for the project: System: Intel Pentium, Hard Disk: 120 GB, RAM: 2 GB
Software Interfaces
Computer Specifications describe the concept of computing resource requirements and
standards that need to be implemented on a device to ensure the optimum operation of the program.
Software specifications for the project:

21
Operating system : Windows 10.
Coding Language : Python, HTML, CSS, Javascript
Communications Interfaces
Since the different components of the system are responsible for different functions and as
they are dependent on each other as well, then communication between them is important. All
components would be looking for each other for their own data to be processed in some way. So, for
this, the underlying operating system would be responsible and will carry out all the communication
or data transfer processes.

2.9.5 Software Quality Attributes

● Adaptability - The product is very adaptable. It can work on maximum platforms, its features might
change with change of platform.
● Availability - This product will be available on the official website for free.

22
 CHAPTER 3
ARCHETECTURE & DESIGN

3.1 SYSTEM ARCHITECTURE

Figure 3.1 Healthcare Chatbot using Sequential Model

When working with text/string data, we have to perform various pre-processing on the data
before making a machine learning or a deep learning model.TF-IDF stands for Term Frequency &
Inverse Document Frequency. It’s one among the foremost important techniques used for information
retrieval to represent how important a selected word or phrase is to a given document. Now for
predicting the response based on user input, we have to use deep neural network that has 3 layers. We
use the Keras sequential API for this. After training the model continuously for 200 epochs, we have
achieved 100% accuracy on our model. The user Input queries are pre-processed in the same way as
the training sets to ensure consistency while predicting responses.

23
3.2 INTERFACE PROTOTYPING (UI)

Figure 3.2 User Interface Prototype for Chatbot

The user interface (UI) is the point of human-computer interaction and communication in a
device. This can include display screens, keyboards, a mouse and the appearance of a desktop. It is
also the way through which a user interacts with an application or a website.

24
3.3 DATA FLOW DESIGN

Figure 3.3 Data Flow Diagram

A data flow diagram shows the way information flows through a process or system. It
includes data inputs and outputs, data stores, and the various subprocesses the data moves through.

25
3.4 USE CASE DIAGRAM

Figure 3.4 Use Case Diagram

A use case diagram is a graphical depiction of a user's possible interactions with a system. A
use case diagram shows various use cases and different types of users the system has and will often be
accompanied by other types of diagrams as well. The use cases are represented by either circles or
ellipses.

3.5 SEQUENCE DIAGRAM

Figure 3.5 Sequence Diagram

A sequence diagram is a type of interaction diagram because it describes how—and in what
order—a group of objects works together. These diagrams are used by software developers and
business professionals to understand requirements for a new system or to document an existing
process.

26
3.6 CLASS DIAGRAM

Figure 3.6 Class Diagram

The class diagram is the main building block of object-oriented modeling. It is used for
general conceptual modeling of the structure of the application, and for detailed modeling, translating
the models into programming code. Class diagrams can also be used for data modeling.

3.7 INTERACTION DIAGRAM

Figure 3.7 Interaction Diagram

Interaction diagrams are models that describe how a group of objects collaborate in some
behavior - typically a single use-case. The diagrams show a number of example objects and the
messages that are passed between these objects within the use-case.

27
3.8 STATE/ACTIVITY DIAGRAM

Figure 3.8 State/Activity Diagram

A state diagram is a type of diagram used in computer science and related fields to describe
the behaviour of systems. State diagrams require that the system described is composed of a finite
number of states; sometimes, this is indeed the case, while at other times this is a reasonable
abstraction.

28
3.9 COMPONENT & DEPLOYMENT DIAGRAM

Figure 3.9 Component & Deployment Diagram

The term Deployment itself describes the purpose of the diagram. Deployment diagrams are
used for describing the hardware components, where software components are deployed. Component
diagrams and deployment diagrams are closely related. Component diagrams are used to describe the
components and deployment diagrams shows how they are deployed in hardware.

29
 CHAPTER 4
IMPLEMENTATION

The execution of the program is an essential stage of the project where the abstract
architecture is compatible with the functional framework. The key stages of deployment are as
follows:
 Gathering Datasets
 Pre-processing data
 Creating Training Datasets
 Building the model
 Testing

Figure 4.1 Software Development Life Cycle

30
Natural Language Processing
Natural Language Processing is that the technology which helps computers to know the
human’s language. It’s not a simple task teaching machines to know how we communicate. Natural
Language Processing, usually known as NLP, which is a branch of AI that deals with the interaction
between computers and humans using the natural language using various operations .The ultimate
objective of NLP is to read, decipher, understand, and add up of the human languages during a
manner that's valuable. Most NLP techniques believe machine learning to derive meaning from
human languages.

Natural Language Processing is that the drive behind the subsequent common applications:
 Language translation applications like Google Translate
 Word Processors like Microsoft Word and Grammarly that employ NLP to see grammatical
accuracy of texts.
 Interactive Voice Response (IVR) applications utilized in call centers to reply to certain users’
requests.
 Artificial Intelligence applications like OK Google, Siri, Cortana, and Alexa.
NLTK (Natural Language Toolkit)
NLTK (Natural Language Toolkit) may be a suite that contains libraries and programs for
statistical language processing. it's one among the foremost powerful NLP libraries, which contains
packages to force machines understand human language and reply with an appropriate predicted
response.

Gathering Datasets
We have used Informatica tool to combine four datasets into a single flat file by creating a
mapping with various transformations to meet the required data attributes for the system.

Figure 4.2 Joining Datasets to create single flat file

31
Target Flat File in WinSCP after executing the mapping

Figure 4.3 Target Flat File in WinSCP

Pre-processing Data
When working with text/string data, we have to perform various pre-processing on the data
before making a machine learning or a deep learning model. Based on our project requirements we
have applied various operations to pre-process the data.

Tokenizing is the most basic and first thing we did on text data. Tokenizing is the process of
breaking the whole text/sentence into small chunks like words.

After that, we have lemmatize each word and remove duplicate words from the list.
Lemmatizing is the process of converting a word into its lemma form and then creating a pickle file to
store the Python objects which we will use while predicting.

Create training and testing data

After building the vocabulary we'd like to vectorize our data. As machines don’t
understand text data, so actually have to convert text data to numbers. We need a way to
represent our text data, which are groups of words, as vectors of numbers. There are many ways of
vectorizing our sentences, here we have used TF-IDF algorithm.
TF-IDF stands for Term Frequency & Inverse Document Frequency. It’s one among the
foremost important techniques used for information retrieval to represent how important a selected
word or phrase is to a given document. Let’s take an example, we've a string or Bag of Words (BOW)
and that we need to extract information from it, then we will use this approach.

32
 The tf-idf value increases in proportion to the amount of times a word appears within the
document but is usually offset by the frequency of the word within the corpus, which helps to regulate
with reference to the very fact that some words appear more frequently generally .
TF-IDF use two statistical methods, first is Term Frequency and therefore the other is Inverse
Document Frequency. Term frequency refers to the total number of times a given term t appears
within the document doc against (per) the entire number of all words within the document and
therefore the inverse document frequency measure of what proportion information the word provides.
It measures the load of a given word within the entire document. IDF show how common or rare a
given word is across all documents.TF-IDF are often computed as tf * idf

Figure 4.4 Training Dataset

Building the model

Now for predicting the response based on user input, we have to use deep neural network that
has 3 layers. We use the Keras sequential API for this. After training the model continuously for 200
epochs, we have achieved 100% accuracy on our model.The user Input queries are pre-processed in
the same way as the training sets to ensure consistency while predicting responses.
Python Keras
Keras is one of the most powerful and easy-to-use Python library for developing and
evaluating deep learning models and algorithms.It works the efficient numerical computation libraries
like TensorFlow and allows us to define and train neural network models in only a couple of lines of
code.
To build our Model, we have used Sequential Model.
A Sequential model is appropriate for a plain stack of layers where each layer has exactly one
input tensor and one output tenso r.We created a Sequential model and added layers one at a time. The
first thing to urge right is to make sure the input layer has the proper number of input features. this
will be specified when creating the primary layer with the argument and setting it to eight for the 8
input variables.
Fully connected layers are well-defined using the Dense class. we will specify the amount of
neurons or nodes within the layer as the first argument, and specify the activation function using the
activation argument.
We will use the rectified linear measure activation function mentioned as ReLU on the
primary two layers and softmax activation function for the third layer.Executing the model uses the
33
efficient numerical libraries under the covers (the so-called backend) like TensorFlow. The backend
automatically chooses the network for training and making predictions to run on hardware, like CPU
or GPU.When compiling, we must specify some further properties required when training the
network. Remember training a network means finding the simplest set of weights to map inputs to
outputs in our dataset.
We must specify the loss function to use to asses a group of weights, the optimizer is
employed to look through different weights for the network and any optional metrics we might wish
to collect and report during training.
In this case, we'll use cross entropy because the loss argument. This loss is for a catagorial
classification problems and is defined in Keras as “catagorial_crossentropy“.We have defined the
optimizer as the efficient stochastic gradient descent algorithm. Stochastic gradient descent with
Nesterov accelerated gradient gives verydecent results for this model.
We have well defined our model and compiled it for efficient computation through which it
can predict response. Now we have executed some training data on our model. We can train our
model on this loaded data by calling the fit() function on the model. Training occurs over epochs and
every epoch is split into various batches.
Epoch: This one undergo all of the rows within the training dataset.
Batch: When One or more samples considered by the model within an epoch before weights are
updated.
One epoch is comprised of 1 or more batches, supported the chosen batch size and therefore the
model is fit many epochs.
The training process will run for a defined number of iterations through the dataset called epochs, that
we must specify using the epochs argument. We must also set the amount of dataset rows that are
considered before the model weights are updated within each epoch, called the batch size and set
using the batch_size argument.
Now, We have trained our neural network on the entire dataset and we can evaluate the performance
of the model/application on the same dataset.

34
4.1 DATABASE DESIGN
4.1.1 ER DIAGRAM

Figure 4.5 Entity-Relationship Diagram

An entity–relationship model describes interrelated things of interest in a specific domain of

knowledge. A basic ER model is composed of entity types and specifies relationships that can exist
between entities.

4.2 USER INTERFACE

Figure 4.6 User Interface

35
4.3 MIDDLEWARE

Jupyter Notebook: Jupyter Notebook is an open source software framework with live programming,
equations, visualizations, and text documents anytime you choose to create and upload them.
Python with Numpy: NumPy is a primary science programming system for Python. Among other
stuff
Python with Pandas: Pandas is an open source data analysis and manipulation tool that is fast,
effective, scalable, and simple to use.

36
 CHAPTER 5
VERIFICATION AND VALIDATION

5.1 UNIT TESTING

Unit testing involves the planning of test cases that validate that the interior program logic is
functioning properly, which program inputs produce valid outputs. All decision branches and internal
code flow should be completely validated. It’s the testing of individual software units of the appliance
.it is done after the completion of a private unit before integration. This is often a structural testing,
that relies on knowledge of its construction and is invasive. Unit tests perform basic tests at
component level and test a selected business process, application, and/or system configuration. Unit
tests make sure that each unique path of a business process performs accurately to the documented
specifications and contains clearly defined inputs and expected results.
Unit testing is typically conducted as a part of a combined code and unit test phase of the
software lifecycle, although it's not uncommon for coding and unit testing to be conducted as two
distinct phases.

Test strategy and approach

Field testing are going to be performed manually and functional tests are going to be written
intimately.
Test objectives
 All field entries must work properly.
 Text processing to be done in User Input
 The entry screen, messages and responses must not be delayed.
 Features to be tested
 Verify that the entries are of the right format
 Duplicate entries should be removed using lemmetizer
 Output response should be generated based on User Input

Table 5.1 Unit Testing

Test_ID Test Action Steps Input Expected Actual Output Pass/fail
Output
1 Check if features 1. Saving the Input Extracted Extracted pass
are extracted coding file Symptoms Features Features
properly 2. Executing the
coding file
2 Check if features 1. Saving the Input Extracted Extracted pass
are extracted coding file greetings Features Features
properly 2. Executing the
coding file

37
5.2 INTEGRATION TESTING
Integration tests are designed to check integrated software components to work out if they
really run together. Testing is event driven and is more concerned with the essential outcome of
screens or fields. Integration tests demonstrate that although the components were individually
satisfaction, as shown by successfully unit testing, the mixture of components is correct and
consistent. Integration testing is specifically aimed toward exposing the issues that arise from the
mixture of components.
Test Results: All the test cases mentioned above passed/executed successfully. No defects has been
found.
5.3 USER TESTING
• User testing is the process through which the interface andvaroius functions of a website,
applications, products, or services are tested by real users who perform specific tasks in
realistic conditions.

5.4 SIZE - LOC

Line of Code- 800

5.5 COST ANALYSIS

The whole project has been coded using python programming language using many libraries
and datasets. The front-end of the ChatBot has been designed using HTML/CSS and JavaScript.
The calculations are as follows:
Effort (E) =a*(KLOC)^b man Months
=2.4*(1.150)1.05
= 2.779 Man months
Scheduled Time (D) = c*(effort)^d months
=2.5*(2.779)0.38
= 3.687 months ~ 4 months approximately.

HARDWARE SPECIFICATION
Processor: Pentium Dual Core 2.3 GHz
Hard Disk: 250 GB or Higher
RAM: 4GB
Approximate minimum cost – Rs 25000

38
SOFTWARE SPECIFICATION
OS: Windows
Framework: Python Flask
Language used: Python, HTML, CSS, JavaScript
Technique/Algorithm: Deep Learning Neural Network & TF-IDF
Approximate minimum cost – Rs 0

5.6 DEFECT ANALYSIS

Defect analysis is a component of the continuous quality improvement planning during which
defects are classified into different categories and also are wont to identify the possible causes so as to
stop the issues from occurring.
The main target of defect analysis is to research defects, identify root causes of defect, then
developing and taking measures or actions to attenuate defects.

5.7 MC CALL’S QUALITY FACTORS

• Product operation factors
• Correctness-The project aims at providing prescription with a high accuracy level.
• Reliability-The datasets used to train model is very reliable.
• Efficiency-The results of the prediction normally is having an efficiency of more than
88%
• Integrity-It is a virtue which deals with proper utilization of software at every level and
its association with the components of the system.
• Usability-The project is build in such a way that it is easily accessible and can be
utilized by people of different strata of education and socio-economic status.
• Product revision factors
• Maintainability-The system aims to be up-to-date with time-to-time updates in
software database and interface according to the needs and usage.
• Flexibility-Capable of adapting the current software to additional circumstances and
customers without changing the software.
• Testability-The system is easily testable by-passing varying types of data to detect
accuracy and mistakes.
• Product transition factors
• Portability-The system is very portable.

39
• Reusability-The project holds a base strong enough for further improvement during its
use and also act as a starting milestone for other projects making use of it's features.
• Interoperability-This project does not focuses on creating interfaces with other
software systems.

40
 CHAPTER 6
EXPERIMENT RESULTS AND ANALYSIS
6.1 RESULTS
In this Project we have build deep neural Networks using 3 layers, in this model we have used
keras Sequential API. As this is a retrieval based healthcare chatbot, we have achieved more than
90% accuracy by training the model to 200 epochs. Here we have used special recurrent neural
network (LSTM) to classify which category the user input message belongs to and then bot will give
random response from the set of responses.

Here in this table we have User Input and ChatBot response based on User Input.

Table 6.1 Predicted Results

S User Input Final Output
no

1 i have breathing problem predicted disease: Asthma;

analgesics: Metered dose inhalers, nebulizers;
treatment scans:Asthma therapy;
diet:Fruits and vegetables
-------> Enter your pincode to see available doctors near by you

2 I am suffering from server predicted disease: 'fever', analgesics: 'paracetemol or aspirin', treatment
headache scans:'tylenol, ibuprofen to treat stomach irritation', diet: 'fluid intake
like gatorage, fruitjuices or milk' -------> enter your pincode to see
available doctors near by you..

3 i am having high body predicted disease: 'fever', analgesics: 'paracetemol or aspirin', treatment
temperature scans:'tylenol, ibuprofen to treat stomach irritation', diet: 'fluid intake
like gatorage, fruitjuices or milk' -------> enter your pincode to see
available doctors near by you..

4 I am sad, feeling very low predicted disease: 'Depression',

analgesics: 'Prozac, Zoloft or Celexa',
treatment scans:'Constantly praise him, teach skills, and give self-
development talks',
diet: 'Fish and whole grains'.
5 Chest pain predicted disease: 'cardiovascular disease',
analgesics: 'Levofloxacin',
treatment scans:'Chest CT scan',
diet:'Citrus fruits, Oily fish and Leafy greens'
-------> Enter your pincode to see available doctors near by you.
6 slow healing of bruises predicted disease: 'Diabetes',
analgesics: 'Tylenol or Aspirin',
treatment scans:'Insulin Treatment',
diet:'Brown rice or cereals with two eggs daily'.

41
6.2 RESULT ANALYSIS
Table 6.2 Result Parameters
Input Precision Recall F1 Score Accuracy
1 89.9 82.5 88.1 91.2
2 89.7 82.6 88.4 91.5
3 89.6 83.1 88.6 92.2
4 89.7 82.5 87.8 92.5
5 88.3 82.7 88.3 92.7
6 88.6 81.9 87.5 91.2

Figure 6.1 ROC Curve

A receiver operating characteristic curve, or ROC curve, is a graphical plot that illustrates the
diagnostic ability of a binary classifier system as its discrimination threshold is varied.

42
 CHAPTER 7
CONCLUSION & FUTURE WORK

Conclusion
ChatBot is useful application for conversation between human and machine. the project is
developed for getting a fast response from the bot which suggests with none delay it gives the
accurate result to the user.
Thus medical ChatBot has wide and vast future scope, regardless of how far people are, they
will have this medical conversation. the sole requirement they have may be a simple desktop or
smartphone with internet connection. The efficient of the ChatBot are often improved by adding more
combination of words and increasing the utilization of database in order that of the medical ChatBot
could handle all sort of diseases. Even voice conversation are often added within the system to form it
less difficult to use.

Future Scope
The future will be the era of messaging app because people are spending longer in messaging
app than the other apps. The implementation of personalized medicine would successfully save many
lives and make a medical awareness among the people. regardless of how far people are, they will
have this medical conversation. the sole requirement they have may be a simple desktop or
smartphone with internet connection. The efficient of ChatBot are often improved by adding more
combination of words and increasing the utilization of database in order that of the medical ChatBot
could handle all sort of diseases.

43
 CHAPTER 8
PLAGARISM REPORT

44
 CHAPTER 9
REFERENCES

[1] Papiya Mahajan, RinkuWankhade, AnupJawade, PragatiDange, AishwaryaBhoge, “Healthcare

Chatbot”,IRJET Nov 2020
[2]Nicholas A. I. Omoregbe, Israel O. Ndaman, Sanjay Misra, Olusola O. Abayomi-Alli ,“Text
Messaging-Based Medical Diagnosis Using Natural Language Processing and Fuzzy Logic ”,Hindawi
Journal of Healthcare Engineering Volume 2020
[3]M.S Bennet Praba, Sagari Sen, Chailshi Chauhan, Divya Singh, “Ai Healthcare Interactive Talking
Agent using Nlp ”, IJITEE,November,2019
[4]Liner Yang, Meishan Zhang, Yang Liu, Maosong Sun, Nan Yu, Guohong Fu, “Joint POS Tagging
and Dependency Parsing with Transition based Neural Networks”,2019
[5]Qiping Yang, Jingui Qin, Yongjie Huang, WushaoWao. “Chatbot using support vector
machine”,2019
[6]SimonHoermann, Kathryn L McCabe, David N Milne, Rafael A Calvo1, “Application of
Synchronous Text- Based Dialogue Systems in Mental Health Interventions” , Journal of Medical
Internet Research , August 2017

[7] Saurav Kumar Mishra, DhirendraBharti, Nidhi Mishra,‖Dr.Vdoc“A Medical Chatbot that Acts as
aVirtual Doctor‖, Journal of Medical Science and Technology”,2017

[8] DivyaMadhu,Neeraj Jain C. J, ElmySebastain, ShinoyShaji, AnandhuAjayakumar,” A Novel

Approach for Medical Assistance Using Trained Chatbot,International” Conference on Inventive
Communication and Computational Technologies(ICICCT 2017)

[9]PavlidouMeropi,Antonis S. Billis,Nicolas D.
Hasanagas,CharalambosBratsas,IoannisAntoniou,Panagiotis D. Bamidis, “Conditional Entropy Based
Retrieval Model in Patient-Carer Conversational Cases”,2017 IEEE 30th International conference on
Computer-Based Medical System.

[10] Gillian Cameron, David Cameron, Gavin Megaw,RaymondBond,MauriceMulvenna ,Siobhan

O‘Neill, Cherie Armour, Michael McTear, “Towards a chatbot for digital counselling‖,Journal of
Medical Internet Research”,2018

[11] C. J. N. J. S. S. Divya Madhu, "A novel approach for medical assistance using trained chatbot,"
in International Conference on Inventive Communication and Computational Technologies (ICICCT),
2017.

[12] S. A. N. H. Hameedullah Kazi, "Effect of Chatbot Systems on Student’s Learning Outcomes,"

SYLWAN, 2019.

[13] S. Divya, V. Indumathi, S. Ishwarya, M. Priyasankari, S. Kalpana Devi, "A Self-Diagnosis

Medical Chatbot Using Artificial Intelligence", J. Web Dev. Web Des., 2018.

45
[14] H. N. I/o, C. B. Lee, "Chatbots and Conversational Agents: A Bibliometric Analysis", IEEE Int.
Conf. Ind. Eng. Eng. Manag.,2017.

[15] Chatbot Using A Knowledge in Database,"Bayu Setiaji,Ferry Wahyu Wibowo",2016 7th

International Conference on Intelligent Systems, Modelling and Simulation.2016 IEEEE.

[16] "Real World Smart Chatbot for Customer Care using a Software as a Service (SaaS)
Architecture"Godson Michael D’silva1, *, Sanket Thakare2, Sharddha More1, and Jeril
Kuriakose1,International conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud)
[17] Chatbot Using A Knowledge in Database,"Bayu Setiaji,Ferry Wahyu Wibowo",2016 7th
International Conference on Intelligent Systems, Modelling and Simulation.2016 IEEEE.

[18] Yuhua Li, Zuhair Bandar, David McLean and James O’Shea "A Method for Measuring Sentence
Similarity and its Application to Conversational Agents "Intelligent Systems Group,2019

[19] Rarhi, K., Bhattacharya, A., Mishra, A., & Mandal, K. (2017). Automated Medical Chatbot.
SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3090881

[20] K (Cameron et al., 2018)Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., Neill,
S. O., McTear, M. (2018). Best Practices for Designing Chatbots in Mental Healthcare – A Case
Study on iHelpr. Proceedings of British HCI 2018

[21] (S & R, 2016)S, V., & R, J. (2016). Text Mining: open Source Tokenization Tools – An
Analysis. Advanced Computational Intelligence: An International Journal (ACII),
https://doi.org/10.5121/acii.2016.3104.

[22] Minha Lee, Lily Frank, Femke Beute, Yvonne de Kort, and Wijnand IJsselsteijn. Bots mind the
social-technical gap. In Proceedings of 15th European Conference on Computer-Supported
Cooperative WorkExploratory Papers. European Society for Socially Embedded Technologies
(EUSSET), 2017

[23] N-gram Accuracy Analysis in the Method of Chatbot Response, International Journal of
Engineering & Technology. (2018)

[24] Mrs Rashmi Dharwadkar1, Dr.Mrs. Neeta A. Deshpande, A Medical ChatBot, International
Journal of Computer Trends and Technology (IJCTT), 2018

[25] C.P. Shabariram, V. Srinath, C.S. Indhuja, Vidhya (2017). Ratatta: Chatbot Application Using
Expert System, International Journal of Advanced Research in Computer Science and Software
Engineering,2017

46
 ANNEXURE
CODING

47
48
49
50