How to Cite:
Deshpande, B., & Chandak, M. B. (2022). A survey of designing tools for chatbot
application. International Journal of Health Sciences, 6(S5), 1403–1413.
https://doi.org/10.53730/ijhs.v6nS5.8889
A survey of designing tools for chatbot application
Bhagyashree Deshpande
Dept. of Computer Science and Engineering, Shri Ramdeobaba College of
Engineering and Management, Nagpur, India
M.B. Chandak
Dept. of Computer Science and Engineering, Shri Ramdeobaba College of
Engineering and Management, Nagpur, India
Corresponding Author: [email protected]
Abstract---The invention of chatbot brought us to the new era of
technology. As Artificial Intelligence continues to gain traction in
modern customer service, chatbots are becoming an integrated part of
user experience. A chatbot has the capability to study and influence
user behavior by asking and answering user’s questions. Chatbots
help with human- computer interaction and address customers’
queries instantly, without the need for a support agent. Now a days
there are many tools available for designing and deploying the chatbot
such as Google dialogflow, Amazon Lex, Microsoft bot, Facebook
Wit.AI, RASA, IBM Watson, Botpress, Botkit, Botsify, Pandorabots,
MobileMonkey, SAP Conversational AI, CoRover, Chatfuel and many
more. This paper represents the different types of tools used for
developing chatbot.
Keywords---Artificial Intelligence, Chatbot, Tool.
Introduction
The world is moving rapidly towards digitalization which will further change how
and where people associate, gather and share information. In an era of
digitalization, everything is available on internet. The world becomes digital with
the new technologies. One of the examples is Chatbot. A chatbot is a computer
program that simulates human conversation through text chats or voice
commands or both[1]. Chatbots respond to what users speak or write. Using a
chatbot is similar to communicating with another person and therefore chatbots
are very appealing to the users. Chatbots are different from computers that use
command-line or graphical interface. When using command-line interface, syntax
is required to communicate with the computer. In graphical interface, users need
to click the appropriate icons to perform specific actions. In both interfaces, users
are confined to a set of predefined commands. The concept has existed since the
International Journal of Health Sciences ISSN 2550-6978 E-ISSN 2550-696X © 2022.
Manuscript submitted: 18 Feb 2022, Manuscript revised: 27 April 2022, Accepted for publication: 9 June 2022
1403
1404
time pople began to develop ways to interact with computers. The first chatbot
introduced was Eliza, developed in 1966 by MIT Artificial Intelligence Laboratory
by Joseph Weizenbaum. Eliza is used to examine the keywords obtained as input
and then activate the output by a defined set of rules. The next chatbot was
Parry, developed by psychiatrist Kenneth Colby in 1972 at the Stanford
University. Parry tried to model the behavior of a Paranoid Schizophrenia. The
next chatbot was A.L.I.C.E (Artificial Linguistic Internet Computer Entity),
developed in 1995 by Richard Wallace. ALICE was designed to steer in a
conversation by reacting to human input and responding immediately. After
developing chatbots, there were various virtual assistants launched. The first
conversational assistant was Siri by Apple. The next assistant was Google
Assistant launched by Google. Microsoft launched Cortana.
After this, smart speakers were introduced which made voice conversation
possible between humans and the chatbots. One of the best examples of smart
speakers are Amazon Alexa and Google Home[2]. Chatbots are appearing at a
rapid pace throughout the communication industry. Over time, chatbots have
become more sophisticated and adopted new AI (Artificial Intelligence) features,
NLP (Natural Language Processing) features, ML (Machine Learning) features. By
adopting these features, chatbot improves user experience. Chatbot has the
ability to communicate and understand human language through different
platforms. There are three types of chatbot – Rule-based, AI-based and Hybrid.
1. Rule Based Chatbot: This type of chatbot is just like decision tree. Therefore,
it is known as Decision treebots. This chatbot stored a set of predefined
responses from a database for a particular question based on the keywords
appeared in the question. Then, it gives a response based on the keywords.
2. AI Based Chatbot: This type of chatbot uses ML, NLP features. This chatbot is
smart enough to provide responses based on previous interaction with the users.
3. Hybrid Chatbot: This type of chatbot is a combination of Rule based and AI
based chatbots. Hybrid chatbot interact with users and gives a reply. In this
chatbot, when the conversation between human and bot gets deeper chatbot is
replaced by a human.
In chatbot, there are some steps for designing the bot. The first step is to identify
scope and requirement of chatbot. The second step is to gather inputs from users
by asking questions. The third step is to recognize User Interface (UI) elements.
The next step is to design first interaction with the users and build a
conversation. The last step is testing of the chatbot.
Figure 1. shows architecture of chatbot. Architecture of chatbot is the main part
of chatbot development. Based on client requirements, we need to design it; but
the basic flow of architecture remains the same. Architecture also depends on the
domain of chatbot. Architecture of chatbot includes user input message, intent
classification, entity recognition, intents, entities, response generator and
response.
1405
Figure 1. Architecture of Chatbot
In the above figure, input of user message is given to an entity recognition and
intent classification. User input message is the text message recorded by chatbot.
Intent specifies the aim of a user’s input. For example- What are some good
South Indian restaurants? The intent of an example would be to find a
restaurant. Entity represents an object which provides a particular context for
intent. Response generator uses different algorithms for processing user’s
request. Then the result which is calculated by response generator is the
response.
This paper presents different types of tools used for designing chatbot. Here, we
will discuss about many frameworks available for developing the chatbot such as
Google dialogflow, Amazon Lex, Microsoft bot, Facebook Wit.AI, RASA, IBM
Watson, Botpress, Botkit, Botsify, Pandorabots, MobileMonkey, SAP
Conversational AI, CoRover, Chatfuel.
Related Work
It consists of many methods which have been used previously for designing
chatbot. Some of the previously published research papers are:
(Rohit Tamrakar, 2021), has focused on a new learning-cum assistance tool called
chatbot. First, the development history of chatbot is reviewed then architecture of
chatbot is explained. After that, different types of chatbot such as knowledge
domain, service provided, goal-based and response generation are presented. In
this paper, various design techniques and approaches are reviewed, followed by
various platforms to build chatbot (Google Dialogflow, IBM Watson, RASA NLU,
Manychat, Microsoft bot, Python and Tensorflow) are presented. Advancement in
chatbot technology, real life practical examples and application of chatbot are also
1406
presented. This review paper proposed the use of the chatbot tool for Computer
Aided Design (CAD) applications. Chatbot tool is also used to solve the CAD
problem. [6]
(Ashutosh Vishwakarma, 2021) This paper proposed an overview on the methods
that are used for designing chatbots. Taxonomy of chatbot is presented. In this
paper, different design techniques such as Elizabeth bot, Microsoft LUIS
(Language Understanding Information Service), Alicebot, Mitsuku, IBM Watson,
Cleverbot, Chatfuel, Amazon Lex are reviewed. After that, comparison is done
between different chatbots according to the technical methods involved in it.
Some limitations on different types of chatbots are explained for providing
successful and productive chatbot. [7]
(Sara Perez-Soler, 2021) This review paper represented chatbot development tools
(Google Dialogflow, IBM Watson, Amazon Lex, Microsoft bot, Botkit, Chatfuel,
Chatterbot, FlowXO, LUIS, Landbot.io, Pandorabots, Rasa, Xenioo, SmartLoop).
The paper emphasizes on managerial and technical features. Basic working
scheme of a chatbot with a procedural and structural diagram is also presented.
For choosing a chatbot development tool, technical factors included here are
input processing, dialogue, deployment, system integration, development &
testing, execution and security. Managerial factors such as organizational,
development-related and operational factors are also included. Finally,
comparison is done on chatbot libraries, frameworks, platforms and services. [8]
(Gkikas C Dimitris, 2021), has focused on how chatbots’ experts work
continuously to add value to these chatbot tools by adding features, plug-ins and
implementing natural language characteristics. This paper examined chatbot
capabilities, features and their utility. Here, a chatbot framework comparison is
also conducted to show similarities and differences of each tool to the
shareholders. Chatbot tools assessment were taken. In order to fulfill this task,
comparison of chatbot frameworks is done based on criteria such as pricing,
location, support of human languages and conversation channels. Six dimensions
are considered for chatbot assessment framework. The first dimension is “primary
applications”. This dimension is essential to know about dynamic concept of a
chatbot framework. The second dimension is “pricing”. Next, “communication
channels’ are integrated. It is considered to be crucial indicator. Fourth
dimension refers to “location”. Fifth essential dimension is “type” of chatbot
framework. There are three types such as rule-based, smart and hybrid. Last
useful dimension is “metrics”. With the help of six dimensions, chatbot
assessment framework is done. [9]
(Amit Patil, 2017), has focused on the chatbot concept and the various cloud
platforms to develop chatbot. Cloud platforms used here are Kore, Chatfuel,
Microsoft bot framework, Microsoft Azure bot, Heroku, AWS lambda and IBM
Watson. Alll the cloud platforms discussed in this paper has many specifications
and functionalities. After that, development of efficient and interactive chatbot is
done with the help of these cloud platforms. This paper represented the
comparison between all cloud platforms with some features such as built in AI,
completion time, setup time and complexity. Finally, this paper is useful to know
which cloud platform is systematic and apt for developing chatbot. After
1407
development and comparison, table is shown for depicting features, pros and cons
of Microsoft Azure, Heroku and Microsoft bot platform. [10]
Tool to build chatbot
Now a days, the necessity of chatbots has increased. A chatbot can communicate
with a human and understand human’s needs. In this section, we will present
many tools which are available for designing and developing the chatbot.
1. Google Dialogflow: Dialogflow is a conversational platform used for designing
and building chatbots and voice chatbots. Google has created this platform for
performing better services. We can create a chatbot with less coding. It supports
Natural Language Processing with more than 20 languages. Dialogflow supports
messaging platforms such as Slack, Kik, Line, Twitter, Viber, Facebook
Messenger, Skype, etc. REST API is provided by Google dialogflow for integrating
the chatbot to our own chatbot application. This tool has two packages - the
standard edition and a paid version.
2. Microsoft Bot: The Azure bot framework is used for building chatbots. This
tool allows you to publish chatbots on apps, Skype, Microsoft Teams, Slack,
Cortana, websites, etc. Microsoft bot easily integrates with existing Microsoft
services. It has two components such as Channel connectors and Bot Builder
SDK. Channel connector is used to connect the chatbot to different messaging
platforms. Bot Builder SDK is used to apply the logic while creating chatbot. This
SDK support Java, Python, C#, Javascript versions. Bot Builder SDK is easily
connected with Natural Language Understanding Services. Microsoft’s LUIS is a
cloud based artificial intelligence service that uses machine learning and natural
language processing technologies. This tool uses LUIS to understand
conversations and bring out important information.
3. Amazon Lex: This tool is used for creating colloquial assistant using text and
voice. Amazon Lex is a service offered by Amazon Web Services. This tool has
incorporated many technologies such as Amazon Cognito for the user
authentication process. For converting audio to text, Automatic Speech
Recognition is used. For converting text to speech, Amazon Lex uses Amazon Polly
services. These features help us to improve the functionality of chatbots. We can
easily test, deploy, and build chatbots with Amazon Lex. It also supports
messaging platforms like Facebook Messenger, Twilio SMS, Slack, Kik.
4. Facebook Wit.AI: Wit.AI is a free, open-source chatbot development tool and it
is owned by Facebook. Wit.AI is the most widely used tool for Natural Language
Processing (NLP) based Facebook Messenger bots because it has NLP features.
HTTP API is used for connecting wit.ai to chatbot application. It also provides SDK
for better purpose. Wit.ai SDK supports Node.js, Python, Ruby.
5. IBM Watson: The tool is used for building colloquial interfaces into any
channel, application, device. This tool uses technologies like Watson Artificial
Intelligence, Machine Learning and Natural Language Understanding. IBM
Watson Assistant supports 13 languages. Watson Assistant provides SDK to
create application in Java, Python, iOS. The framework allows everyone to keep
1408
data that flows through it. IBM uses private cloud to store the information
gathered by their chatbots.
6. RASA: Rasa is an open source tool used to develop conversational chatbot
using text and voice. Rasa is used in large companies. RASA has two fundamental
components, that is Rasa Core and Rasa NLU. Rasa Core is used for building
conversational assistant. Rasa core has a feature of dialogue management. Rasa
core support Facebook Messenger, Twilio, Telegram, Slack and it is written in
Python. The process of classification, entity extraction and retrieving of responses
is handled by Rasa NLU (Natural Language Understanding). Rasa has paid and
advanced version. Rasa framework has incorporated with APIs, enterprise grade
support and graphical user interface. We can deploy chatbot on our own server.
7. Botpress: Botpress is an open source chatbot development tool with dual
license. For developing chatbot, we require React, NodeJS, Javascript, Typescript
languages. Botpress offers an interface in which developers and non-technical
people can manage the chatbot’s content after implementation. Dialog Manager
and Flow Builder are the two features of Bopress. These features are used for
building and debugging conversational flow. We can deploy chatbot on different
messaging platforms such as Slack, Facebook, Telegram, Bot, Twilio, Web.
Botpress is mostly used by governments, finance companies, insurance
companies.
8. Botkit: This tool is an open source chatbot development platform. Botkit is
handled by Microsoft. It is used for developing custom integrations, chatbots and
custom apps for messaging platforms. Botkit SDK has NodeJS language. Botkit
supports messaging platforms such as Cisco, Cisco Jabber, Facebook Messenger,
Microsoft Teams, Microsoft Bot, Google Hangouts, Slack, Webex. With Botkit, we
can deploy the chatbot on our own server.
9. Botsify: Botsify is popular chatbot development tool. We can build the bots
using drag and drops UI (User interface). Botsify is not free platform. Botsify has
integrated with Google Sheet, Dialogflow, Dashbot, Shopify. The best feature of
Botsify is human agent handover. We can deploy the chatbot on Website,
Facebook Messenger, Slack using Botsify.
10. Pandorabots: Pandorabots is a web service used to develop and publish
chatbots. This tool has an open source, free libraries. It also offers premium
libraries available for an additional monthly fee. For scripting the chatbot
conversation, Aritificial Intelligence Markup Language (AIML) is used by
Pandorabots. Using AIML, we can build chatbot in any natural language. It also
offers SDKs in Java, NodeJS, PHP, Python, Go, Ruby. This tool has plenty of free
documentation, and training courses available online. We can deploy the chatbot
on messaging platforms such as Facebook Messenger, WeChat and Viber.
1409
Table 1. Pricing, Pros and Cons of various Tech-giant tool to build Chatbot
Tool Pricing Pros Cons
Standard version Offers Software
Google Dialogflow – free Development Kits Does not offer
Paid version - (SDKs) for more live customer
$0.002 per text than 14 platforms support
request and
$0.075 per
minute
Free version – Supports Framework is
1000 messages seamless complex.
Microsoft Bot per month integration with Need to write too
Paid version - existing Microsoft much code to
$0.50 per 1000 services implement a
messages Provides SDKs for basic function
various languages
Free version – Automatic scaling Framework is
Amazon Lex 10000 message capabilities not multilingual.
requests and Supports various It supports
5000 speech platforms and English only.
requests per deployment Data
month preparation
Paid version - process is
$0.004 per complicated
speech &
$0.00075 per
text request
Make it easy for
large developer The training of
Facebook Wit.AI Provides custom community to the NLP engine
pricing for their create in Wit.AI is
software applications complicated
Well-designed
developer user
interface
Lite plan– free, Promote data Framework is
IBM Watson allows 1000 privacy complex
users It allows seamless Does not allow
Plus plan - $14 phone integration end user to
per 100 users access chat
Enterprise plan history
– 50000 users
Develops highly Chatbots are
RASA Offers custom customizable resource-
pricing based on chatbots and run intensive on
the features as simple http server side
opted by users servers Does not provide
integrations
1410
Open-source Easily Framework has
Botpress package – free customizable limited features
Enterprise Documentation Framework uses
package – provided with a high learning
pricing is framework is easy curve
calculated based to use
on chosen
features
Free version – Botkit is easy to Lack of options
Botkit Build simple bot use and work for for developers
Paid version – different platform who don’t work
With Botkit in isolation in node.js
CMS, create a
bot & pay for
hosting bot
Free trial of 14 Pre-built User Interface is
Botsify days templates for not as intuitive
Personal - building a as other options
$40/m chatbot quickly
Professional - Integrates with
$124/m websites
Free version – Pandorabots Accuracy and
Pandorabots get unlimited AIML provides reliability for
messages in flexibility business is not
sandbox Predefined good
Developer content libraries Not advisable to
version - provide some host your
$19/month shortcuts chatbot
Pro version -
$199/month
Free plan – Free account for Slow scalability,
MobileMonkey perfect for small businesses, no built-in lead
business large number of gen integration,
Pro plan - integrations, good minimal
$14.25 per for Facebook reporting, uses
month marketing Facebook only
Uicorn plan -
$36.75 per
month
Team plan -
$299 per month
Free version Entity User interface is
SAP Paid version – enrichment, not simple, bot
Conversational AI EUR 300 for efficient training, usage analytics
price per block totally free for are bit limited
of 1000 individuals with
chats/month high rating
Free version Improve
CoRover Paid version – operational Costly to
pricing based on efficiency, easy to implement
1411
opted features implement and
integrate
Free version - Automate
Chatfuel $0/m responses Setting it up is
Pro version - Link to resources time consuming
$15/m Provide a phone
number
11. MobileMonkey: MobileMonkey is a tool used for developing chatbot without
any programming knowledge. MobileMonkey has many features which are used to
build and deploy chatbots for Websites, Facebook Ads, Bulk Messaging, etc.
MobileMonkey acquired ChattyPeople to extend chatbot building capabilities.
MobileMonkey can be integrated with Gmail, Google Forms, Asana, Mailchimp,
WordPress, Instagram.
12. SAP Conversational AI: SAP Conversational AI is an end- to-end chatbot
development platform for the enterprise. It is also known as Recast.AI. Through
this tool, we can build chatbots easily, in days and in full SAP integration. It is
easy to use and has fastest bot building platform. This tool presents SDK for
Javascript, Python, Java, Ruby, PHP. It has a limited free version available for
use. SAP Conversational AI supports messaging channels such as Cisco Spark,
Cortana, CoPilot, Messenger, Microsoft Teams, Slack, SAPJam Collaboration, Kik,
Line, Twilio, Telegram, Twitter.
13. CoRover: CoRover is a conversational AI platform provides managed chatbot
service with self onboarding support to improve customer experience and to help
enterprise grade. It is one of the most widely used chatbot across banking, travel,
finance, healthcare and manufacturing. CoRover has Omni channels such as
Web, WhatsApp, SMS, Social Media Bots. It works with 100 foreign and 12
vernacular languages. It has robust, secure, scalable architecture. Chatbots can
be deployed on public or private clouds.
14. Chatfuel: Chatfuel is a famous tool used for creating chatbots. With chatfuel,
we can create and publish chatbots for Facebook Messenger. Chatfuel powers
46% of all messaging bots. Chatbot can be created from the exiting templates or
using the drag and drop option. We don’t need any coding knowledge to create a
chatbot using chatfuel.
Table 1 shows pricing, pros and cons of various tech-giant platform to build
chatbot.
Proposed Work
Here, we have opted Amazon Lex platform for implementation purpose. Amazon
Lex is an AWS service used to create conversational interfaces i.e. chatbot using
voice and text. Amazon Lex bot works with Automatic Speech Recognition and
Natural Language Understanding capabilities. We are creating a chatbot such as
ordering a pizza. We will see some terminologies for creating a bot. First one is
intent. An intent is used to represent a task that user wants to perform. For our
chatbot, OrderPizza is the name of intent. Sample utterences are phrases that we
expect a user to speak or type to invoke the intent. For example, “I would like to
1412
order pizza”. Amazon Lex provides built-in intents to set up the chatbot. After
intent, slots are created. Slots include the information that a chatbot needs to
fulfill the intent. The OrderPizza intent requires slots such as pizza type, pizza
crust and appetizers. In intent configuration, we add these slots. Each slot has a
type. Slot type is a list of values used to capture values for a slot. For example, we
use the following slot types for the OrderPizza intent: crust – thick and thin.
Amazon Lex provides built-in slot types such as AMAZON.FirstName,
AMAZON.Number. After creating slot types, we can build, test, and deploy our
chatbot directly from the Amazon Lex console. Amazon Lex guides us through the
Amazon Lex console to create our own chatbot in minutes.
Results and Conclusion
Amazon Lex is a fully controlled artificial intelligence service with advanced
natural language models to design, build, test and deploy chatbot. We have
created a chatbot for ordering a pizza. With Amazon Lex console, we have tested
it. The final output from the console can be seen in the figure below.
Figure 2. Output of Pizza Order Chatbot
In this paper, we have presented 14 tools that are used for designing and
developing a chatbot. Among 14 tools, we have used Amazon Lex platform for
creating a pizza ordering chatbot. The purpose of Amazon Lex is to provide a
standard platform for developing a chatbot using text or voice. With pizza ordering
chatbot, we can easily order a pizza online. Thus, we have implemented a chatbot
using Amazon Lex.
1413
References
[1] What Is a Chatbot? [Online] Available:
https://www.oracle.com/in/chatbots/what-is-a-chatbot/
[2] The History and Evolution of Chatbots. [Online] Available:
https://insights.daffodilsw.com/blog/the-history-and-evolution-of-chatbots
[3] Some Basic Terms Of Chatbots. [Online] Available:
https://www.geeksforgeeks.org/some-basic-terms-of-chatbots/
[4] What is Chatbot: Design Process & Its Architecture. [Online] Available:
https://www.elprocus.com/chatbot-design-process-and-its-architecture/
[5] 5 Reasons Why Your Chatbot Needs Natural Language Processing. [Online]
[Available]: https://towardsdatascience.com/5-reasons-why-your-chatbot-
needs-natural-language-processing-
[6] Rohit Tamrakar, Niraj Wani, “Design and Development of CHATBOT: A
Review”, International Conference on Latest Trends in Civil, Mechanical and
Electrical Engineering, Bhopal, India, April 2021.
[7] Ashutosh Vishwakarma, Ankur Pandey, “A Review & Comparative Analysis
on Various Chatbots Design”, International Journal of Computer Science
and Mobile Computing, vol. 10, no. 2, pp. 72-78, February 2021.
[8] Sara Perez-Soler, Sandra Juarez-Puerta, Esther Guerra and Juan de Lara,
“Choosing a chatbot development tool”, in IEEE Software, vol. 38, no. 4, pp.
94-103, July-Aug. 2021.
[9] Gkikas C Dimitris, Theodoridis K Prokopis, Tzavella Georgia, Vlachopoulou
Georgia, Kondili Isidora and Tzioli Melpomeni, in 8th International
Conference on Contemporary Marketing Issues, January 2021.
[10] Amit Patil, K Marimuthu, Nagaraja Rao A and R Niranchana, “Comparative
Study of cloud platforms to develop a Chatbot”, International Journal of
Engineering and Technology, vol. 6, no. 3, pp. 57-61, June 2017.
[11] 15 Best AI Chatbot Development Framework & Platforms [Online] Available:
https://www.kommunicate.io/blog/chatbot-framework-platform/
[12] 18 Chatbot Development Frameworks and Chatbot Platforms [Online]
Available: https://cedextech.com/chatbot-development-frameworks/
[13] Building Bots with Botkit [Online] Available: https://discover.bot/bot-
talk/guide-to-bot-buiding-frameworks/botkit/
[14] Pandorabots Review of Pricing, Features, Pros & Cons, and Expert Chatbot
Software Advice [Online] Available: https://www.chatbots.org/pandorabots
[15] MobileMonkey Review [Online] Available:
https://supportyourapp.com/blog/mobilemonkey-review/
[16] SAP Conversational AI [Online] Available:
https://www.capterra.com/p/204599/SAP-Conversational-AI/
[17] SAP Conversational AI [Online] Available:
https://www.sapstore.com/solutions/42282/SAP-Conversational-AI
[18] CoRover – AI Virtual Assistant [Online] Available: https://corover.ai/
[19] What is Amazon Lex? [Online] Available:
https://docs.aws.amazon.com/lex/latest/dg/what-is.html
[20] Amazon Lex: How It Works [Online] Available:
https://docs.aws.amazon.com/lex/latest/dg/how-it-works.html
