Creative Commons Attribution Non-Commercial License

Copyright: © the author(s), publisher and licensee Technoscience Academy. This is an open-access article distributed under the
terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use,
distribution, and reproduction in any medium, provided the original work is properly cited
International Journal of Scientific Research in Computer Science, Engineering and Information Technology
ISSN : 2456-3307 (
doi :
Designing and Implementing Conversational Intelligent Chat-bot Using
Natural Language Processing
Asoke Nath*1
, Rupamita Sarkar1
, Swastik Mitra1
, Rohitaswa Pradhan1

1Department of Computer Science, St. Xavier’s College (Autonomous), Kolkata, India

Article Info
Volume 7, Issue 3
Page Number: 262-266

Publication Issue :

Article History
Accepted : 15 May 2021
Published : 22 May 2021
In the early days of Artificial Intelligence, it was observed that tasks which
humans consider ‘natural’ and ‘commonplace’, such as Natural Language
Understanding, Natural Language Generation and Vision were the most difficult
task to carry over to computers. Nevertheless, attempts to crack the proverbial
NLP nut were made, initially with methods that fall under ‘Symbolic NLP’. One
of the products of this era was ELIZA. At present the most promising forays into
the world of NLP are provided by ‘Neural NLP’, which uses Representation
Learning and Deep Neural networks to model, understand and generate natural
language. In the present paper the authors tried to develop a Conversational
Intelligent Chatbot, a program that can chat with a user about any conceivable
topic, without having domain-specific knowledge programmed into it. This is a
challenging task, as it involves both ‘Natural Language Understanding’ (the task
of converting natural language user input into representations that a machine
can understand) and subsequently ‘Natural Language Generation’ (the task of
generating an appropriate response to the user input in natural language).
Several approaches exist for building conversational chatbots. In the present
paper, two models have been used and their performance has been compared
and contrasted. The first model is purely generative and uses a Transformerbased architecture. The second model is retrieval-based, and uses Deep Neural
Keywords: Natural Language Processing, Natural Language Understanding,
Natural Language Generation, Deep Neural Networks, Artificial Intelligence,
Transformer Model, Intelligent Agent, Chatbot.
Natural Languages are those that came into
existence for the purpose of being used by
humans in order to interact with one another.
Such languages have evolved alongside humans,
without any conscious planning. Natural
Language Processing resides at the overlap
between Linguistics, Computer Science and
Artificial Intelligence and involves
understanding, analysing, processing and
manipulating Natural Language data.
Building a Conversational Bot using Artificial
Intelligence is a problem in the field of Natural
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
Language Processing, whose aim is to construct an
intelligent agent which is able to hold a
conversation in human understandable language,
with another human or in some cases, another
intelligent agent.
The field of conversational chatbots can prove to
be a gateway for machines to understand humans
and human connections better, since language is
one of our largest tools of communication and for
improvement in quality of conversation it is a
must for a chatbot to master a language and
explore its depths. Chatting experience can
further be enhanced by enabling text-based
sentiment analysis in a chatbot model which shall
help the bot to better understand the sentiment
and feelings behind a user message.
Chatbots are currently seeing extensive
implementation in online customer services,
appointment booking (doctor’s appointment,
business appointments etc.), providing
psychological counselling, and question
answering. Bots like Facebook messenger chatbot
allows businesses to communicate effectively
with their customers.
Such chatbots can range from simple domainspecific question answering bots to more random
chatter bot which can even simulate personality
traits such as empathy, sense of humour,
compassion etc.
The generative chatbot built for this paper uses a
transformer architecture and the Retrieval based
bot uses a Deep Neural Net architecture. A
generative Transformer model takes each word of
an input sequence, processes them in parallel and
outputs the probability of the next word in the
output sequence. The Transformer model
implemented in this paper is based directly on
the paper by Vaswani et. al., 2017 [1], which first
brought transformers into existence. The
Transformer model discussed in this paper had
been trained on the WMT 2014 English-German
dataset, which is an English to German
translation dataset consisting of about 4.5 million
sentence pairs and on the significantly larger
WMT2014 English-French translation dataset
consisting of 36M sentences and split tokens into
a 32000 word-piece vocabulary. This is also the
base model for many architectures performing
different NLP tasks, including building chatbots.
This model does quite well on Neural Machine
Translation tasks, with a BLEU score of 41.0 on
the big model trained on English-French dataset.
BLEU score is a popular metric in NLP for
evaluating the quality of a generated sequence by
computing the number of n-grams in the machine
generated sequence matching against the n-grams
in the target sequence provided by a human. An
n-gram here is a n-word ordered set which is a
subsequence of any given sequence. In the Neural
Machine Translation task, a model learns to relate
words/phrases of the input language to
words/phrases of the target language. When it
comes to building a chatbot, this particular base
model needs to be altered to do the job. The
Blenderbot (Roller et. al., 2020) [2] uses the
transformer of [1] as its base model and builds
three different architectures on top of this base
model: a retrieval-based architecture, a generative
architecture and a retrieve-and-refine model. The
retrieval-based architecture of Blenderbot uses
two poly-encoder models (Humeau et. al., 2020)
[3] which is a pre-trained architecture with an
additional pre-trained attention mechanism for
increasing the number of global feature
representations. Also, it uses a ranking system
based on cross-entropy loss of one correct sample
and multiple negative samples. The generative
model is inspired by ParlAI (Miller et. al., 2017)
[4], which comprises multiple tasks and agents
where the tasks are APIs representing various
problems and the agents are applied to solve these
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
tasks. The tasks can be of type conversation logs,
online or reinforcement learning tasks, and also
real language and simulated tasks. ParlAI is based
on the three main concepts of world- this is an
environment which can have two conversing
agents and it can also be an environment with
multiple conversing agents; agent- can be a
human, a hard coded bot or a system trained with
machine learning and an agent can speak i.e.,
converse; teacher- an entity that converses with
the learner to help it learn to speak. For decoding
from the set of output word probabilities and
deciding on the output word in Blenderbot, the
Beam Search decoding has been experimented. In
Beam Search, instead of always choosing the word
with highest probability (greedy search), a beam
length b is decided. For the first output word, the
b highest probability words are chosen. For each
of these b first words, b words are chosen such
that the combined probability of the previous and
the new word is amongst the b highest
probability combinations. This goes on until all
the possible word sequences reach the maximum
number of words allowed in the output, at which
point, the word sequence with the highest
probability (product of the probabilities of each
constituent word) is selected for output. For
sampling from decoder outputs, top-k and search
and rank approaches have been tried out. The
top-k approach has been used in the Transformer
model chatbot discussed in this paper and the topk method was used in Holtzman et. al., (2020)
[14]. This paper comes up with the top-k
sampling approach while trying to discuss a very
common problem faced by purely generative text
generation models and this problem is called
Neural Text Degeneration, and it entails the
repetition of the same word or phrase in the
output sequence, such that the frequency of this
repetition makes the sequence look like an
incoherent loop. The top-K approach randomly
samples an output word from a set of k highest
probability words. This reduces the chance of text
degeneration due to repetition quite a bit.
This paper discusses two algorithms: the
transformer algorithm for purely generative
chatbot and the Deep Neural Net algorithm for
retrieval-based bot.

A. Transformer:
The transformer model uses two different data
sets, the first is the Cornell Movie Dialog Corpus.
The next dataset is a custom dataset created from
multiple famous books and novels (Including
Sherlock Holmes, The Shining, The Picture of
Dorian Grey etc). This custom dataset has been
used to increase the size of the training dataset as
well as for providing well-structured text
generation examples for the Transformer.
The transformer model is based on the model
discussed in the paper: Attention is All You Need
(Vaswani et. al., 2017). The basic modules of this
architecture are briefed as follows:
1) Preprocessing: The sequences with less than
5 words are removed and all words that
occur less than 50 times in the individual
datasets are also removed. The sequences are
then tokenized and padded.
2) Word Embedding: Word embeddings have
been created with a dimension d_model =
128. This dimension has been maintained for
the parameter matrices throughout the
architecture. The code snippet in Fig. shows
the creation of word embedding in
3) Positional Encodings: Positional embeddings
are used for capturing information regarding
the absolute position of a word in a sequence
as well as the relative position of each word
with respect to the other words in the
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
sequence. Sine and Cosine waves have been
used for capturing the positional information
of even and odd positions respectively.
4) Multi Headed Self Attention: The task of the
self-attention layer is to find out the
relationship of each word in a sequence with
all the other words in that sequence. The
input to the self-attention is the positional
embedding matrix (in case of encoder) and
the masked positional embedding matrix (in
case of decoder). Three copies of the input
are made. What it does is take a single word
of a sequence and computes the degree of
relevance (attention) of every other word in
the sequence with that word. A single selfattention mechanism is called a head. The
transformer model uses multiple such selfattention heads in both encoder and decoder.
5) Look Ahead Mask: The solution to this is the
look-ahead mask that creates a matrix which
is to be added to the output of every selfattention head. This mask is designed in such
a way that for every word (with respect to
which self-attention is computed at any
instant), all the words coming after it in the
sequence will be masked out.
Total trainable parameters: 1,539,157

B. Retrieval Based Bot:
The authors created the intents dataset for this
bot and preprocessed it using the Natural
Language ToolKit (NLTK). The retrieval-based
bot has its data in JavaScript Object Notation
(JSON) format stored in a file called

Figure 1. Transformer Model Summary
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
The various intent classes in the dataset include:
purpose, feelings, family, politics, suicidal,
sadness, covid etc. There are a total 38 intents in
the dataset. The model used for this bot is a Deep
Neural Network model with alternating Dense
and Dropout layers. All the non-final Dense
layers use ReLU activation and the final Dense
layer uses Softmax activation function.
It is a classification model that classifies the user’s
input into one of the 38 intent classes, following
which a random response is selected from the set
of predefined responses. This randomly chosen
response is then displayed as the chatbot’s reply.
Total Trainable parameters: 51,174

● epochs – 10,
● loss (sparse categorical cross-entropy) –
● accuracy (sparse categorical accuracy)-
Retrieval Based bot:
● epochs – 200,
● loss – 0.2081,
● accuracy – 0.9142
i used to rule the world
sure as door this himself before take her every find we ever other house many
seas would rise when i gave the word
hands find had room always might all and under most but think life such no
one-minute i held the key then the walls were closed on me and i discovered that my castles stand upon
pillars of salt and pillars of sand
yet through any day at work always made their house right little found now against
her eyes and words are so icy oh but she burns like rum on fire . hard and fast and angry as she can be i walk
my days on a wire
what turned how you home might something time never ever it’s other her place last
Conversation of Purely Generative Transformer Bot where the normal fonts and messages from user and the
bold fonts are replies from bot
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
Figure 2. questions that test the self-awareness of the bot

Figure 3. Some imperfect replies from the bot

Figure 4. Bot’s awareness about COVID-19
C. Reason for the Performance Difference There is an obvious difference in the quality of
conversation of the Generative Chatbot and the
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
Retrieval Based Chatbot. The Retrieval based bot
is superior when it comes to providing
grammatically correct and meaningful replies.
The replies are meaningful with respect to the
user’s message so long as the bot can successfully
classify the intent of the user message. On the
other hand, the generative chatbot is uttering
unintelligible sentences. The reason for this is
given in the comparison table below:
1) The purely generative Transformer Bot
literally generates the output sequence one
word at a time, using no outside help. All it has
is the input sequence and its own internal
mechanism to rely upon. This is why a purely
generative chatbot falls under the category of
Language Models, which are models that
predict the probability of each word in the
sequence. Chatbots as well as other text
generators fall under this category.
Retrieval based bots are provided with a set of
intents which serve as topics on which the
chatbot is capable of holding conversation. For
each topic there are a set of sequences which
serve as the examples of the kind of user
messages the bot might expect and using these
the bot is trained to classify the user message
under the correct intent. So the task of this
chatbot is not to generate an output sequence
word by word, but to figure out which topic
the user message falls under. Once classified, it
randomly selects one of the reply strings from
the set of reply strings provided for that intent
and outputs it.

Figure 5. Model Summary of Retrieval Based Deep Neural Net Bot
2) The final layer of the generative chatbot has to
compute the probability of each word in the
vocabulary of the corpus. If all words are kept,
this number can reach well over 20,000. For this
model, the authors eliminated all words below a
predetermined threshold frequency and there
were still 3412 words. So the final Dense layer of
the model had 3412 units and had to compute
the probability for each unit i.e., probability for
the word that unit represents. This adds to the
difficulty of making such predictions.
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
The final layer of the retrieval based bot has to
compute probability for 38 intents classes. This
significantly reduces the number of parameters
needed and also makes it easier for the model to
learn to classify correctly.
3) A transformer model has different kinds of
modules with different functionalities. The
SOTA (State Of The Art) Transformer models
that are used for text generation and as chatbots
all have a huge number of parameters.
Facebook’s Blenderbot for example have
variations that have been pre trained using 256M
parameters and 622M parameters [1]. To
parallelize computation on such a huge volume
of parameters (tremendously large matrices), it
requires memory and computational resources
like multiple distributed GPU or TPU that at this
moment only the largest corporates can support
and even then such bots are often a hybrid of
retrieval based and generative. The transformer
bot built in this paper has less than 2M
parameters (1,539,157 to be precise, with 6
layers for each Encoder and each Decoder Stack;
8 heads for each of the two multiheaded
attention and multiple, dense, dropout and
normalization in both Encoder and Decoder),
and trained on a single remote TPU provided by
google colab. So it is fair to say that the resources
available for this model do not allow it to reach
its full potential but at the same time the
Transformer Chatbot built for this project
demonstrates the functioning of such models.
The Retrieval based bot is built using a
sequential model with 51,174 parameters and 6
layers. The model is simpler and smaller and the
considerably less number of parameters makes it
trainable on a CPU. As a result, the model learns
fast with very little demand for resources.
4) To save time, it was necessary to cut short the
size of the training dataset of the Generative
Transformer Bot. The number of training
examples in our final corpus was 63846. Even
then the time it took for a single epoch to
complete was 380s i.e., 6 minutes. We have
limited the total number of epochs of the final
model to 10 and the total time taken to train the
model for 10 epochs was a little over 1 hr. When
trained on the entire dataset (size of which is
over 2,000,00), a single epoch took over 1 hr.
The size of our dataset was very small compared
to even those Transformer models that are
considered as “small”. For example, the size of
our data corpus is 6 MB whereas a small
RoBERTa-like model (which is a special variant
of the Transformer) needs 3 GB sized corpus to
train on [13]. The reason for requiring this huge
dataset is that this particular chatbot is to be
trained to understand human language and
generate text on its own and from scratch. This
is the same as a human child learning to build
comprehensive and coherent sentences on its
own. While a SOTA model still takes way less
time than a human child needs to learn to
generate meaningful texts, it still requires too
many resources which is very hard to provide for
The Retrieval based bot was trained with 200
epochs and each epoch took less than a minute
to complete. It does not need to learn to generate
text from scratch. It only needs to learn to
classify the user’s message into one of the classes
available to it, based on the word patterns
present in the user message. For a Retrieval
based model it is important to have a good
variation of user messages for each class and to
be very particular when deciding which user
message goes under which intent class, so as to
minimize the chance of misclassification. A
Retrieval based model is capable of holding
meaningful conversations even when trained
with very few parameters and on minimal
computational and memory resources, provided
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
that the Training dataset is well planned and the
intents and patterns are made to be specific and
the ambiguity is kept to the minimum.
How can a conversational bot be infused with a
sense of personality? The PERSONA-CHAT
dataset [15] has been created with help of crowd
workers who were asked to chat in pairs and
attempt for a natural conversation. The
conversations held between each pair had
different tones (topic wise and mood wise),
called profiles. Different personas were also
collected and for each persona there were five
profile sentences. These personas were for
training a conversational bot to infuse with the
essences of individual personalities. Here each
persona is essentially a short character summary
and the profile sentences assigned to each
persona, define that particular character. This
dataset can be used either for creating a dialog
agent capable of showing different personalities
or for training models that, given a dialog
history, will make predictions on the persona of
that dialog.

A. Limitations
1) Retrieval-Based Model
• Does not do well on topics that are outside its
scope – In this context, “scope” means the intents
that are defined for the chatbot. For instance, a
retrieval-based chatbot might have an intent
called “name” which indicates that the user is
asking for the chatbot’s name. A retrieval-based
chatbot tries to classify the intent of a user
message into one of those that are preprogrammed into it. However, the user might
enter a message such that the chatbot fails to
find an intent that fits the user message. This can
easily happen, as it is impossible for the
programmer to foresee all the kinds of messages
that a user can provide. In such a case, a
retrieval-based chatbot fails to provide
meaningful output.
• Might repeat replies for multiple questions that
belong to the same topic – The best that can be
done to prevent a retrieval-based chatbot from
repeating the exact same response for a question
on a given topic (intent), is to provide multiple
possible replies under each topic (intent) and
have it choose one at random to give as a reply.
Even under this scheme, since the number of
possible replies under each intent is not very
high, there can be frequent repetition of
response, when asked multiple questions on the
same topic.
2) Transformer-Based Model
• Might not provide grammatically correct or even
relevant replies – As seen in the “Results and
Discussion” section, the transformer-based
chatbot gives nonsensical output that is neither
related to the user’s query, nor grammatically
correct. The reason for this is closely linked to
the following limitation.
• Requires intensive computational resources to
train to desired performance – In this paper, the
term “desired performance” simply means
“coherent replies somewhat linked to the user’s
message”. Yet, even to achieve this simple
milestone, a transformer-based bot requires
massive computational resources. A detailed
breakdown of this is given in the table in section
C in Results and Discussion. Hence, it is usually
only feasible for big corporations having massive
computational assets, to implement productionlevel transformer-based chatbots, and even then,
they use hybrid models combining the two
classes viz. Retrieval-based and Transformerbased.
3) B. Scope for Future Improvement
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
• Increasing scale – For the transformer-based
model, increasing the corpus size, parameters
and computational resources (resources used to
train) could make a huge difference to its
performance. As for the retrieval-based chatbot,
more intents could be added to make the chatbot
able to answer a wider range of questions.
• Using a hybrid model – Even production-level
generative chatbots are not purely generative.
They use retrieval-based models in addition, to
get the best of both worlds. For instance, one
strategy could be to let the retrieval-based model
reply if it is able to classify the intent of a user
message, but if it is unable to classify the intent,
then to let the generative model generate a
• Better User Interface – An application with a
user-interface like those of mobile chat
applications could be designed. Further, the
processing of the user message and generation of
replies could be moved to a server, requiring a
user to only install a front-end application,
reducing the processing power needed by an
end-user to run the application. Of course, in
that case, the server would receive requests in
the form of user messages and send back replies.
• Text-To-Speech and Speech-To-Text – For users
who require assistive technologies (visually
impaired users) or simply for lazy convenience, a
user could speak their message instead of typing
it and the application would automatically
convert it into text. Similarly, when the chatbot
replies to a user’s message, the reply could be
read out aloud by the application for the user to
• Performance Measurement using NLP Metrics –
In the future, the performance of this Chatbot
could be more accurately measured using metrics
such as BLEU score(Bilingual Evaluation
Understudy Score), perplexity, etc.
The authors are indebted to the Post Graduate
Department . of Computer Science, St. Xavier’s
College(Autonomous), Kolkata, for providing them
an opportunity to work on “Intelligent Chatbot”. The
authors Dr. Asoke Nath, Rupamita Sarkar, Swastik
Mitra, and Rohitaswa Pradhan are very much
grateful to Dr. Rev. Dominic Savio, S.J, Principal of
St. Xavier’s College(Autonomous), Kolkata for giving
all time support to do research work in Science and

[1]. Ashish Vaswani, Noam Shazeer, Niki Parmar,
Jakob Uszkoreit, Llion Jones, Aidan N. Gomez,
Łukasz Kaiser, llia Polosukhin, “Attention is All
You Need”, Page-2-6, 2017, Conference on
Neural Information Processing Systems, Long
Beach, CA, USA.
[2]. Stephen Roller, Emily Dinan, Naman Goyal, Da
Ju, Mary Williamson, Yinhan Liu, Jing Xu,
Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan
Boureau, Jason Weston, “Recipes for building
an open-domain chatbot”, Facebook AI
Research, 30th April, 2020.
[3]. Humeau et al, Facebook AI research, 25th
March, 2020.
[4]. Miller et al, Facebook AI research, 2017.
[5]. Denny Britz, “Recurrent Neural Network
Tutorial, Part 4 – Implementing a GRU/LSTM
RNN with Python and Theano.,” WILDML, 27-
[6]. Dzmitry Bahdanau, KyungHyun Cho, Yoshua
Bengio, “Neural Machine Translation By Jointly
Learning to Align and Translation”, Page-3-4,
2015, Conference paper at ICLR.
[7]. “Neural Machine Translation with Attention.”,
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
[8]. Amir Ali, Muhammad Zain Amin,
“Conversational AI Chatbot Based on EncoderDecoder Architectures with Attention
Mechanism”, Page-8-10, 2019, Artificial
Intelligence Festival 2.0, NED University of
Engineering and Technology, Karachi,
[9]. Abonia Sojasingarayar, “Seq2Seq AI Chatbot
with Attention Mechanism”, Page-3-10, 2020,
IA School/University, Boulogne-Billancourt,
[10]. J.Prassanna, Khadar Nawas K, Christy Jackson J,
Prabakaran R, Sakkaravarthi Ramanath,
“Towards Building a Neural Conversation
Chatbot through Seq2Seq Model”, International
Journal of Scientific & Technology Research,
ISSN 2277-8616, Vol-9, Issue-3, Page-2, 2020.
[11]. A. Gillioz, J. Casas, E. Mugellini and O. A.
Khaled, “Overview of the Transformer-based
Models for NLP Tasks,” 15th Conference on
Computer Science and Information Systems
(FedCSIS), Sofia, Bulgaria, 2020, pp. 179-183,
doi: 10.15439/2020F20.
[12]. Julien Chaumond, Patrick von Platen, Orestis
Floros, Anthony MOI, Aditya Malte, “How to
train a new language model from scratch using
Transformers and Tokenizers”,
blog/01_how_to_train.ipynb at master •
huggingface/blog (, 11th March,
[13]. Ari Holtzman, Jan Buys, Li Du, Maxwell
Forbes, Yejin Choi, “The Curious Case Of
Neural Text DeGeneration”, Conference Paper
at ICLR, 14th February 2020.
[14]. Saizheng Zhang, Emily Dinan, Jack Urbanek,
Arthur Szlam, Douwe Kiela, Jason Weston,
“Personalizing Dialogue Agents: I have a dog,
do you have pets too?”, Montreal Institute for
Learning Algorithms, MILA and Facebook AI
Research, 25th September.
Dr. Asoke Nath is working as Associate Professor in
the department of Computer Science, St. Xavier’s
College (Autonomous), Kolkata. He is engaged in
research work in the field of
Cryptography and Network
Security, Steganography, Green
Computing, Big Data Analytics,
Li-Fi technology, Mathematical
modelling of Social Area
Networks, MOOCs, etc.
He has published 253 research articles in different
journals and conference proceedings.
Rupamita Sarkar is a
Postgraduate student of
Computer Science at St.
Xavier’s College
(Autonomous), Kolkata. Her
interests include Data Science
and Machine Learning, with a
particular leaning towards Natural Language
Swastik Mitra is a
Postgraduate student of
Computer Science at St.
Xavier’s College
(Autonomous), Kolkata. He is
interested in research work in
the field of Machine Learning
and Artificial Intelligence. Specifically, his topics of
interest are Deep Learning and Natural Language
Processing. He also shares an interest in Data Science,
Big Data Analytics, Algorithms, and Distributed
Volume 7, Issue 3, May-June-2021 |
Asoke Nath et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol, May-June – 2021, 7 (3) : 262-266
Rohitaswa Pradhan is a
Postgraduate student of
Computer Science at St.
Xavier’s College
(Autonomous), Kolkata. He is
interested in the fields of
Analysis of Algorithms, Data
Science and Machine Learning. He is also
enthusiastic about various Software Development
technologies and Development Methodologies.
Cite this Article
Asoke Nath, Rupamita Sarkar, Swastik Mitra,
Rohitaswa Pradhan , “Designing and Implementing
Conversational Intelligent Chat-bot Using Natural
Language Processing “, International Journal of
Scientific Research in Computer Science, Engineering
and Information Technology (IJSRCSEIT), ISSN :
2456-3307, Volume 7 Issue 3, pp. 262-266, May-June
2021. Available at doi
Journal URL :

Get Professional Assignment Help Cheaply

Buy Custom Essay

Are you busy and do not have time to handle your assignment? Are you scared that your paper will not make the grade? Do you have responsibilities that may hinder you from turning in your assignment on time? Are you tired and can barely handle your assignment? Are your grades inconsistent?

Whichever your reason is, it is valid! You can get professional academic help from our service at affordable rates. We have a team of professional academic writers who can handle all your assignments.

Why Choose Our Academic Writing Service?

  • Plagiarism free papers
  • Timely delivery
  • Any deadline
  • Skilled, Experienced Native English Writers
  • Subject-relevant academic writer
  • Adherence to paper instructions
  • Ability to tackle bulk assignments
  • Reasonable prices
  • 24/7 Customer Support
  • Get superb grades consistently

Online Academic Help With Different Subjects


Students barely have time to read. We got you! Have your literature essay or book review written without having the hassle of reading the book. You can get your literature paper custom-written for you by our literature specialists.


Do you struggle with finance? No need to torture yourself if finance is not your cup of tea. You can order your finance paper from our academic writing service and get 100% original work from competent finance experts.

Computer science

Computer science is a tough subject. Fortunately, our computer science experts are up to the match. No need to stress and have sleepless nights. Our academic writers will tackle all your computer science assignments and deliver them on time. Let us handle all your python, java, ruby, JavaScript, php , C+ assignments!


While psychology may be an interesting subject, you may lack sufficient time to handle your assignments. Don’t despair; by using our academic writing service, you can be assured of perfect grades. Moreover, your grades will be consistent.


Engineering is quite a demanding subject. Students face a lot of pressure and barely have enough time to do what they love to do. Our academic writing service got you covered! Our engineering specialists follow the paper instructions and ensure timely delivery of the paper.


In the nursing course, you may have difficulties with literature reviews, annotated bibliographies, critical essays, and other assignments. Our nursing assignment writers will offer you professional nursing paper help at low prices.


Truth be told, sociology papers can be quite exhausting. Our academic writing service relieves you of fatigue, pressure, and stress. You can relax and have peace of mind as our academic writers handle your sociology assignment.


We take pride in having some of the best business writers in the industry. Our business writers have a lot of experience in the field. They are reliable, and you can be assured of a high-grade paper. They are able to handle business papers of any subject, length, deadline, and difficulty!


We boast of having some of the most experienced statistics experts in the industry. Our statistics experts have diverse skills, expertise, and knowledge to handle any kind of assignment. They have access to all kinds of software to get your assignment done.


Writing a law essay may prove to be an insurmountable obstacle, especially when you need to know the peculiarities of the legislative framework. Take advantage of our top-notch law specialists and get superb grades and 100% satisfaction.

What discipline/subjects do you deal in?

We have highlighted some of the most popular subjects we handle above. Those are just a tip of the iceberg. We deal in all academic disciplines since our writers are as diverse. They have been drawn from across all disciplines, and orders are assigned to those writers believed to be the best in the field. In a nutshell, there is no task we cannot handle; all you need to do is place your order with us. As long as your instructions are clear, just trust we shall deliver irrespective of the discipline.

Are your writers competent enough to handle my paper?

Our essay writers are graduates with bachelor's, masters, Ph.D., and doctorate degrees in various subjects. The minimum requirement to be an essay writer with our essay writing service is to have a college degree. All our academic writers have a minimum of two years of academic writing. We have a stringent recruitment process to ensure that we get only the most competent essay writers in the industry. We also ensure that the writers are handsomely compensated for their value. The majority of our writers are native English speakers. As such, the fluency of language and grammar is impeccable.

What if I don’t like the paper?

There is a very low likelihood that you won’t like the paper.

Reasons being:

  • When assigning your order, we match the paper’s discipline with the writer’s field/specialization. Since all our writers are graduates, we match the paper’s subject with the field the writer studied. For instance, if it’s a nursing paper, only a nursing graduate and writer will handle it. Furthermore, all our writers have academic writing experience and top-notch research skills.
  • We have a quality assurance that reviews the paper before it gets to you. As such, we ensure that you get a paper that meets the required standard and will most definitely make the grade.

In the event that you don’t like your paper:

  • The writer will revise the paper up to your pleasing. You have unlimited revisions. You simply need to highlight what specifically you don’t like about the paper, and the writer will make the amendments. The paper will be revised until you are satisfied. Revisions are free of charge
  • We will have a different writer write the paper from scratch.
  • Last resort, if the above does not work, we will refund your money.

Will the professor find out I didn’t write the paper myself?

Not at all. All papers are written from scratch. There is no way your tutor or instructor will realize that you did not write the paper yourself. In fact, we recommend using our assignment help services for consistent results.

What if the paper is plagiarized?

We check all papers for plagiarism before we submit them. We use powerful plagiarism checking software such as SafeAssign, LopesWrite, and Turnitin. We also upload the plagiarism report so that you can review it. We understand that plagiarism is academic suicide. We would not take the risk of submitting plagiarized work and jeopardize your academic journey. Furthermore, we do not sell or use prewritten papers, and each paper is written from scratch.

When will I get my paper?

You determine when you get the paper by setting the deadline when placing the order. All papers are delivered within the deadline. We are well aware that we operate in a time-sensitive industry. As such, we have laid out strategies to ensure that the client receives the paper on time and they never miss the deadline. We understand that papers that are submitted late have some points deducted. We do not want you to miss any points due to late submission. We work on beating deadlines by huge margins in order to ensure that you have ample time to review the paper before you submit it.

Will anyone find out that I used your services?

We have a privacy and confidentiality policy that guides our work. We NEVER share any customer information with third parties. Noone will ever know that you used our assignment help services. It’s only between you and us. We are bound by our policies to protect the customer’s identity and information. All your information, such as your names, phone number, email, order information, and so on, are protected. We have robust security systems that ensure that your data is protected. Hacking our systems is close to impossible, and it has never happened.

How our Assignment  Help Service Works

1.      Place an order

You fill all the paper instructions in the order form. Make sure you include all the helpful materials so that our academic writers can deliver the perfect paper. It will also help to eliminate unnecessary revisions.

2.      Pay for the order

Proceed to pay for the paper so that it can be assigned to one of our expert academic writers. The paper subject is matched with the writer’s area of specialization.

3.      Track the progress

You communicate with the writer and know about the progress of the paper. The client can ask the writer for drafts of the paper. The client can upload extra material and include additional instructions from the lecturer. Receive a paper.

4.      Download the paper

The paper is sent to your email and uploaded to your personal account. You also get a plagiarism report attached to your paper.

smile and order essaysmile and order essay PLACE THIS ORDER OR A SIMILAR ORDER WITH US TODAY AND GET A PERFECT SCORE!!!

order custom essay paper