AI and Web-Based Interactive College Enquiry Chatbot

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1235
AI and Web-Based Interactive College Enquiry Chatbot
Akshada Phalle1, Saniya Kadam1, Sakshi Sonphule1, Ila Savant2
1Department of Computer Engineering, Marathwada Mitra Mandal’s College of Engineering, Maharashtra, India
2Assistant Professor, Department of Computer Engineering, Marathwada Mitra Mandal’s College of Engineering,
Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – A chatbot is often described as the most
advanced and promising expression of interaction
between humans and machines. These digital assistants
streamline the interactions between people and services
with less human interaction. This is proved as to be a boon
during the COVID-19 pandemic when people could no
longer visit information desks to get their queries
answered. Admission season is arguably the busiest time
for a college office. Concerned parents flock the enquiry
desks with inquiries about the college. These questions are
repetitive and one can use a virtual agent for these doubts
which would free the office personnel to handle other
issues. This paper proposes a college enquiry chatbot
based on Python's chatbot framework, Rasa. Rasa is an
open source chatbot framework based on machine
learning. With its help, one can easily create highly
accurate chatbots and integrate them with websites and
messaging applications such as WhatsApp, Facebook,
Telegram, etc. The proposed chatbot could handle
manually typed as well as rich responses generated by
clicking on a payload. It could be integrated with the
college website by adding a widget and deployed as on the
messaging application, Telegram.
Keywords: Conversational AI, Natural Language
Processing, Human Computer Interaction, Chatbot, College
Enquiry Chatbot, Rasa, Rasa X, Python, Telegram Chatbot
1. INTRODUCTION
Most websites today be it real-estate, educational
institutions, financial firms have a dedicated page for
contact information in case a visitor has questions. The
problem with this traditional form of online
communication is that visitors have to contact help desks
only within a stipulated time even if the query is trivial. In
rush hours, most people have to be kept on hold in case of
limited human resources. Although this is unavoidable, it
may leave clients dissatisfied and sully a company's image.
This is where conversational AI comes into the picture in
the form of chatbots. An artificial conversation entity
called chatbot system uses conversational Artificial
Intelligence (AI) to simulate a dialogue. It uses rule-based
language applications to perform live chat functions in
response to real –time user interactions. It removes
barriers to customer support that can occur when demand
outpaces resources. Instead of waiting on hold, an
enquirer can get answers to his/her questions in real time.
This virtual assistant could provide valuable help to
college institutions, especially during the admission
season. In most colleges, faculties have to devote their
time in the admission cells to answer the questions of
concerned parents about a college. These queries, though
important, are mostly redundant. Thus, we have proposed
a web-based college enquiry chatbot that could be
integrated with the college's website that could handle all
these questions from various departments to placement
cells to hostels and so on at a single click from a user. It
allows the users to type their query manually or to click on
a button of their choice in case they are unsure about their
actual doubt. The system would contain information from
the official website, except the user won't have to navigate
to multiple pages just to know the cut-off for a particular
department. Furthermore, for those who do not have the
time to actually visit the college website, we have also
proposed integration with the messaging application,
Telegram which provides users with the opportunity to
look back to their previous conversations with the bot, an
addition to the web version.
2. MATHEMATICAL MODEL
Rasa NLU internally uses the following two algorithms for
entity extraction:
1: Bag of words (BoW) is one of the famous methods for
extracting the characteristics from conversations so that it
can be used in building models for a machine learning
algorithm. It is a representation of words which includes
two main things, a document of known words and the
frequency of known words. It can simply be simple or
complex depending upon the complexity of creating the
vocabulary of known words and the circumstance of
known words. It follows the succeeding steps
a): Collecting the data which could be any input data taken
from the user, we can treat each and every line as a
different document.
b): Designing the vocabulary where we gather the list of all
the unique words and discarding case sensitive and
punctuation and put that into the model vocabulary.

c): Creating the document vector where we check the
frequency of words in each document. The main thing of
this step is to make sure that it turns each word in the
document into the vector form like a matrix so that it can
be easily used as an input to the other machine learning
algorithms. The easiest way of scoring is to mark the
presence of words as a Boolean value, 0 for absent and
1for present and then convert it into a binary vector.
2: Conditional Random Fields (CRFs) are a
discriminative model, used for predicting sequences. They
use contextual information from previous labels, thereby
increasing the amount of information the model has, to
make a good prediction.
We used the following parameters to evaluate the
performance of our chatbot at each stage.
1. Intent accuracy=
2. Entity extract accuracy =
3. Integrity of an entity =
4. Integrity of a sentence =
3. METHODOLOGY
A Rasa NLU pipeline usually incorporates three main
parts:
1. Tokenization: Tokenization involves breaking the
raw text into small chunks called tokens. These tokens
help in comprehending the context or developing a model
for the NLP. It helps in explicating the meaning of the text
by analyzing the progression of the words. There are two
tokenizers which we have used in our chatbot.
● Whitespace Tokenizer: A Whitespace Tokenizer is
a tokenizer that splits on and discards only whitespace
characters. This implementation can return Word, Core
Label or other LexedToken objects. It also has a method
for whether to make EOL a token or whether to treat EOL
characters or as a whitespace.
● SpacyTokenizer: It processes the text from left to
right. First, the tokenizer splits the text on whitespace
similar to the split() function in Python. Then it checks
whether the substring matches the tokenizer exception
rules.
2. Featurization: Featurization is the process to
transfigure varied forms of data into numerical data. This
data can be in the form of text, images, videos, graphs,
various database tables, time-series, categorical features,
etc. The featurizers used in this project are as follows:
● LanguageModelFeaturizer: This featurizer that
uses transformer-based language models.It uses a pre-
trained language model to compute vector representations
of input text.
● RegexFeaturizer:It creates a vector
representation of a user message using regular
expressions. During training, the RegexFeaturizer will
make a list of regular expressions as per the training data
format. For each regex, a parameter will be set marking
whether this expression was found in the user message or
not.
3. Intent Classification: Intent classification uses
machine learning and natural language processing to
automatically equate words or expressions with a
particular intent. The classifier used in this project is as
follows:
● DIETClassifier : DIET (Dual Intent and Entity
Transformer) is a multi-task architecture for intent
classification and entity recognition. The system
architecture is based on a transformer which is shared for
both tasks. A sequence of entity labels is predicted through
a Conditional Random Field (CRF) tagging layer on top of
the transformer output sequence corresponding to the
input sequence of tokens. For the intent labels the
transformer output for the complete utterance and intent
labels are embedded into a single semantic vector space.
4. Entity Extraction : Entity extraction is a text
analysis technique that uses Natural Language Processing
(NLP) to automatically pull out specific data from
unstructured text, and classifies it according to predefined
categories. In this project, we have used the DIETClassifier
for entity extraction as well.
4. Hardware & Software Requirements
Hardware Requirements:
1. RAM: 8 GB
2. Solid State Drive
3. External Storage Capacity: 64 GB
Software Requirements:
1. Operating System: Windows 10 or above
2. Code Editor: Visual Studio Code 1.65.2

3. Programming Language: Python 3.7.6
4. Bot Framework:
● Rasa 2.2.0
● Rasa SDK 2.8.4
● Rasa X 0.34.0
5. ngrok 5.3.1
5. RESULTS
Fig -1: Web Version Output
Fig -2: Telegram Version Output
6. CONCLUSION
The project proposes an AI-based College Enquiry Chatbot
that could be integrated with the college website as well as
the popular messaging application, Telegram to interact
with visitors and answer their questions about the college.
The chatbot will work with users who input their queries
in both normal textual as well as custom payloads which
allow users to click on their desired query instead of
typing them manually.
REFERENCES
[1] Pi, S.N.M.S. and Majid, M.A., 2020, December.
“Components of Smart Chatbot Academic Model for a
University Website”. In 2020 Emerging Technology in
Computing, Communication, and Electronics (ETCCE) (pp.
1-6). IEEE.
[2] Windiatmoko, Y., Rahmadi, R. and Hidayatullah, A.F.,
2021, February. “Developing Facebook Chatbot Based on
Deep Learning Using RASA Framework for University
Enquiries”. In IOP Conference Series: Materials Science
and Engineering (Vol. 1077, No. 1, p. 012060). IOP
Publishing.
[3] Gbenga, O., Okedigba, T. and Oluwatobi, H., 2020. “An
Improved Rapid Response Model for University Admission
Enquiry System Using Chatbot”. Int. J. Comput., 38(1),
pp.123-131.

[4] Kumari, S., Naikwadi, Z., Akole, A. and Darshankar, P.,
2020, July. “Enhancing College Chat Bot Assistant with the
Help of Richer Human-Computer Interaction and Speech
Recognition”. In 2020 International Conference on
Electronics and Sustainable Communication Systems
(ICESC) (pp. 427-433). IEEE.
[5] Koundinya, H., Palakurthi, A.K., Putnala, V. and Kumar,
A., 2020, July. “Smart College Chatbot using ML and
Python”. In 2020 International Conference on System,
Computation, Automation, and Networking (ICSCAN) (pp.
1-5). IEEE.
[6] Patel, N.P., Parikh, D.R., Patel, D.A. and Patel, R.R., 2019,
June. “AI and web-based human-like interactive university
chatbot(UNIBOT)”.In 2019 3rdInternational Conference
on Electronics, Communication and Aerospace Technology
(ICECA) (pp. 148-150). IEEE.
[7] Kulkarni, P., Mahabaleshwarkar, A., Kulkarni, M.,
Sirsikar, N. and Gadgil, K., 2019, September.
“Conversational AI: An Overview of Methodologies,
Applications & Future Scope”. In 2019 5th International
Conference On Computing, Communication, Control An
Automation (ICCUBEA) (pp. 1-7). IEEE.
[8] Ralston, K., Chen, Y., Isah, H. and Zulkernine, F., 2019,
December. “A voice interactive multilingual student
support system using IBM Watson”. In 2019 18th IEEE
International Conference On Machine Learning And
Applications (ICMLA) (pp. 1924-1929). IEEE.

AI and Web-Based Interactive College Enquiry Chatbot

More Related Content

Similar to AI and Web-Based Interactive College Enquiry Chatbot (20)

More from IRJET Journal (20)

Recently uploaded (20)

AI and Web-Based Interactive College Enquiry Chatbot