IRJET- Multimedia Chatbot using Classification

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 09 | Sep 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 1264
Multimedia Chatbot using Classification
Priya Sharma 1, Shivani Sabnis 2, Yash Mane3, Abhinay Chauhan4
1,2,3,4B.E. Student, Computer Science, Ramrao Adik Institute of Technology, Maharashtra, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - With the ever-proliferating need for information,
the office staff often find themselves burdened with additional
responsibilities of repeatedly answering the common queries.
Chat bots in the digital world have made every organization
curious to use them as a major tool to interact with their
customers. The system, Multimedia chatbot, for answering
various queries of different committees in the collegehasbeen
developed by classifying the questions into various committee
category questions. The proposed system aims at alleviating
the setbacks faced by the students by developing a chatbot
system as an application that aids students in getting updated
with college activities. The chatbot is built using voice
recognition techniques and Part-Of-Speech (POS) tagging
which is a natural languageprocessingtechniquethatanalyze
user's queries and understand user's input. The chatbot
includes voice assistance, which is accomplished using Google
Speech API; and responds with the appropriate available
format like audio, images, text. The chatbot has been made by
using AIML (Artificial Intelligence Markup Language), JAVA
and NLP (Natural Language Processing). Our chatbot
effectively answers college related queries so the students get
the desired information quickly without having to go through
the unnecessary hassle. Students can provide input through
voice or by text and can use this application for enquiring
anytime in the college premises.
Key Words: POS tagging, NLP, AIML, JAVA, Voice
assistance.
1. INTRODUCTION
With burgeoning domains of cutting-edge technology and
information, prospects are that chatbots can bridge the gap
between the reservoir of data and the latest generation
programs which perform human conversation intelligently.
The Multimedia chatbot is a system designed for students to
fire queries pertaining to academic information as well as
general information about the college. In behestoftheuser’s
query, the system interprets the question and delivers the
result. It is not required for the users to follow some
sequential format for firing the queries. NLP (Natural
Language Processing) is used that primarily programs the
computer to surveys and process large natural language
data. The knowledge base contains question and answers.
For every question a keyword or series of keywords are
present which acts like classification words for that
particular question. As the data-mining and machine-
learning techniques have improved, chat bots have begun to
be more practical in daily life applications. One of the
examples being automatic telephone answering systems.
This project, titled ‘Multimedia Chatbot’ streamlines a
pathway towards creating a chatbot application for our
college which can accept user queries as text input aswell as
audio input, cognize and fathom thecontextandconnotation
of the queries, look up the pertinent data in the knowledge
base and administer a relevant answer as output-all through
an interactive and easy to understand chat interface. The
chatbot encompasses an admin panel which has complete
control over the system, enabling them to manage the
knowledge base, keep the records of chat logs etc. Chatbot
can pull data database and presentitinthemostappropriate
format. The project overcomes the disadvantage of getting
all the data on a single page without any complications of
visiting multiple pages by providinga user-friendlyinterface
that uses natural language processing to interact and solve
queries of college students. The system analyses user’s
queries which are answered using AIML and NLP. Users can
directly enter the chat screen and can ask anycollegerelated
queries and in response get the answer from the chatbot
with an attractive graphical user interface whereas the
admin will have its user id and password for security
purpose to manage database and chatlogs. Admin can know
about the questions that were unansweredbythechatbot on
viewing the chatlogs and can get the knowledge base
updated. The system helps the users to be updated about
college activities. The system analyses student’s queries by
using NLP’s Grammatical Tagging algorithm.
2. LITERATURE SURVEY
COLLEGE ENQUIRY SYSTEM [1]: This System is a web
application which provides answer to the query of the
student very effectively. Studentscaninputtheirqueryinthe
bot to receive the answers. If the answer is found invalid,
then system declares the answer as invalid. These invalid
answers are modified or else removed from the database by
the admin of the system. Chatbot helpsthestudentstogetthe
answers to their queries. This keeps the students notified
with the college activities. The system has an online notice
board. On this notice board, any Text notices or PDF
documents can be displayed. This helps the user to be
updated with the important notices. Not much time will be
wasted by the user to search for the important notices. The
answer to the query is answered on the basis of the users
queriesandtheknowledgebase.Theimportantkeywordsare
fetched from the keywords and the answer to those
keywords are searched in theknowledgebase. Ifthematchis
found, the relevant answer is provided to the user or the
default message is shown to the user that Answer to this
query is not available at the moment, please revertbackafter
some time. The Keyword Matching algorithm will be used to
match the keywords from the knowledge base [1].

CHATBOT FOR COLLEGE MANAGEMENT SYSTEM USING AI
[2]: The User can ask any college-related queries to the
chatbot with the advantage of not personally going to the
college for enquiry. The System examines the question and
then provides the answer to the user. With the help of AI,
system can provide answers to any queries asked by the
students. This is done using an effective GUI as if a real
person is talking to the user to use the system, user first
requires to register and thenlogin to it. This system uses two
algorithms. First, Porter Stemmer Algorithm, which is a
process for removing suffixes from words in English Second,
Word Order Similarity between Sentences. Sentenceshaving
exactly the same words but differing in ordering may mean
different. It is simple for humans to process word order
information. Due to this, many preexisting methods do not
handle this type of information. So, this system establisheda
method which stores the word order information into
account when calculating sentence similarity [2].
ENQUIRYCHATBOTUSINGKNOWLEDGEINDATABASE[3]–
This system answers to student questions that is related to
college. In this system, students are first expected to select
the category regarding which they want to question the bot.
Once, the category is selected,studentinputsthequerywhich
is then checked in the database foritspresence.Ifthequeryis
matched in database, the correspondinganswerisforwarded
to the student as an output. These queries can be related to
admission, faculty details, etc. Students won’t have to
personally go to the college for enquiring. The chatbot's
developmentisdoneusingMicrosoftbotframework,whichis
using Microsoft cognitive service i.e. LUIS.ai for training
language model which is used to identify intent of the user
and fetch an appropriate response. The System makes the
use of built in Artificial Intelligence to provide the answer go
the query provided by LUIS.ai. The response fromthesystem
is given in an effective GUI manner such that it appears as if a
real person is communicating. First, bot will transfer the
query to the LUIS.ai. which further forwards its reply to the
system after which it fetches for the value from the
knowledge base supplying the answer to student question
[3]?
Summary of related work:
Table -1:
System Year Advantages Disadvanta
ges
College Enquiry
Chatbot
March
2107
1. Open Source
2. Effectively
Categorize
Products
1. Text-
Only
2. No Self-
Learning
Chatbot For
College
Management
System
2017
1. Open Source
2. Use of
Worldnet
Dictionary
3. Automatically
Prioritize
1. Text-
Only
2. Time
Efficiency
Query1
College Enquiry
ChatBot Using
Knowledge In
Database
April
2018
1. Open Source
2. Uses LUIS.ai
for training
language model
No Self
Learning
New
Language
Delay
The systems studied here serves a stepping stone in
developing our own comprehensive chatbot solution. From
the case studies referred, a complete study of the existing
chatbot systemisundertaken.Here,theproblemwith‘college
enquiry chatbot’ is that if the query is not found in the
database, the query is declared invalid. In oursystem,wewill
try to overcome it-whenever user submits a complaint, the
query is detected. Then it is checked if the query resides in
the database. If answer found in database, it is displayed on
user’s screen. If a particular query is not found in the
database, a pop up asking for the user's email id comes. Such
situation is handled by admin. Once he answered the
question the answer is provided to that user. And that
question -answer pair is stored in database so thatwhenever
that question will be asked, the correspondinganswerwillbe
found in database and answer will be given as an output.
Therefore, answers need not be entered manually by the
admin anymore.
In case of ‘chatbot for college management system’, it is not
integrated with text-to-speechandspeech-to-textsoftware’s.
Our system is a speech-recognizedchatbot.Voicerecognition
will be accomplished using a library called Google Speech.
Speech recognition can be broken into two groups with five
subsets in total; speakers and speech styles. Speakers make
up singlespeakingandspeakingindependentlywherespeech
styles include isolated word recognition, connected word
recognition and continuous word recognition.
In addition to all this, oursystemwillalsocompriseofchatlog
section where the admin can see all the questions asked by
the user. The questions unanswered can be viewed from this
chatlog and the database is updated accordingly. The
literaturesurvey gave abriefoverviewofthesystemsalready
set in place. Comparisons between these systems gave a
clarity and betterunderstandingofourproject.Advancement
in technology has brought in many such systems which are
already in place today, as briefed in literature survey. But all
of these systems have theirownsetoflimitations.Oursystem
multimedia chatbot aims to provide a comprehensive
solution to these limitations by yielding better accuracy in
terms of performance and computational time compared to
the preexistingsystemsanddevelopingascalable,robustand
user-friendly web application.
3. DESIGN
The design of the system is represented in the pictorial
format by the figure 3.1.

The Flowchart shows how the user will interact with this
desktop application by entering their query in the text box
provided on the front end of the system. Once they submit
this query, the query will be checked in the knowledge base
if the exactly same question is present.Ifpresent,theanswer
will be displayed on the screen. Otherwise, the query will be
sent for pre-processing using NLP’ POS-tagger technique.
This pre-processed query is again checkedinthedatabaseto
provide the answer to the user. If the answer is not found,
the query will be checked in the AIML.
There are situations where chatbot don't have a response.
For such circumstances, a log file is created where all the
queries asked by the user are added including the
unanswered queries. Admin handles such situation. Admin
will check for these unanswered queries and update the
knowledge base with the answers to these queries so that
whenever user enters the same query next time, he will get
the answer. Also, when the answer to any query is
unavailable, a window pop-up asking for the user’s email id
and phone number. When the admin updatestheknowledge
base, the response is sent to the user who asked the certain
question to their entered email id. In this way, the chat bot
accuracy is improved.
Fig -1: Flow chart depicting Navigation of the working of
chatbot
4. IMPLEMENTATION
Multimedia chatbot converses with students.Thequeryfired
by the student is processed using NLP. The grammatical
tagging algorithm is implemented or working of this
proposed system is as follows:
Grammatical Tagging Algorithm:
INPUT: Sentence String (str)
OUTPUT: Response String (res)
STEP 1: Take query as input from user.
STEP 2: Check for response using rule-based approach.
STEP 3: If response string found.
STEP 3.1: Return response string.
STEP 3.2: STOP
STEP 4: Else
STEP 4.1: str.trim(); //removes extra spaces.
STEP 4.2: Sring g1[]=str.split(); //tokenizes the query
STEP 4.3: Repeat the following steps till g1.length();
STEP 4.3.1:Stringres=MaxenTagger.tagString(g1[]);//labels
each token as nouns, verbs, adverbs, adjectives, etc.
STEP 4.4: If res.endsWith(NN,NNS,NNP) or If
res.endsWith(VBG,VBD,VBN). // tags for noun and verb.
STEP 4.4.1: Match res with keywords in database.
STEP 5: If match found, return res string as an output to the
user.
STEP 6: STOP
The front end for the chatbot is developed using jQuery. Java
Swing is used forbuildingtheUIcomponents.Theknowledge
base for the chatbot is developed using MySQL. The
responses for the normal greetings are generated using the
custom AIML files. The POS tagger is used for processing the
queries. The appropriateanswers are provided to the userin
the form of text or voice.
Natural language processing (NLP) is a subfield of artificial
intelligence, computer science which is concerned with the
interaction between computer and human (natural
languages).
In our system, as user input will be in English, to let machine
understand this language we used Natural Language
Processing.

The technique used in NLP is pos tagger. A Part-Of- Speech
Tagger (POS Tagger) reads the text and allots parts of speech
to every word, i.e., noun, verb, adjective, etc. Pos incudes 36
tags. A part-of-speech tagger, or POS-tagger operatesaseries
of words, and attaches a part of speech tag to each word. The
given text is divided into tokens. Example: text = word
tokenize (” When is Udaan conducted?”) postag(text)
[(’When’, ’WRB’), (’is’, ’VB’) (’Udaan’, ’NN’), (’conducted’, JJ)]
Once this is done, the words categorized are checked in
database to match with the keywords. If the match is found,
the corresponding answer is displayed. [4]
Artificial intelligence Markup Language is one of the first
tools used to design ALICE. An XML based markup language,
AIML is designed to create artificial intelligent application
like chatbot. AIML has a data store which is used for
answering questions which are not related to college. For
example, "What is your name?". Basically, AIML includes
responses to normal greetings.UsingAIMLwewillcreateour
custom AIML file. This is done by invoking the bot by using
the writeAIMLFiles () function. A java program called
AddAIML.java is created for this purpose. The program
execution adds the custom entries to the bot’s brain.
In java we will be using jQuery for creating user interface.
Java Swing is used to design window-based applications. It is
entirely written in java. Java Swing supplies platform-
independent and lightweight components. The javax.swing
package is used for providing classes for java swing API such
as JButton, JTextField, JTextArea, JRadioButton, JCheckbox,
and JMenu.
5. RESULT
The system has setup a chat-based application which would
serve the users by answering any queries they might have
regarding various daily college activities. Be it a simple
question such as "Where is CSIroom” or a slightly complex
one such as “Steps for issuing bonafide certificate".
The administrator panel is controlled by the admin who has
total control of the system. Admin is able to update the
knowledge with the latest details from time to time.
Fig 5.1 CHAT SCREEN
Fig 5.1 shows the chat screen which is used by the users to
interact with the chatbot and ask anycollegerelatedqueries.
Fig 5.2 ADMIN LOGIN PAGE
Figure 5.2 shows the LoginPage where the admin logs in
using credentials. By logging in, the admin can view the
chatlogs, update the database, etc.
Fig 5.3. ADMIN PANEL FOR UPDATINGQnA
Figure 5.3 shows the admin panel. once the admin is
logged in, this panel is used to add, update, view the
database and answered/unanswered chatlogs.
Fig 5.4 QnA
Fig 5.4 shows the Q n A section which shows all the
questions and corresponding responses stored in database.

Fig 5.5 UNANSWERED CHAT LOGS
Figure 5.5 shows Unanswered chatlogs. This is the tab in
admin panel where the admin can view the unanswered
chatlogs.
Fig 5.6(a). CHAT INTERFACE
Fig 5.6(b). ANSWER IN VIDEO FORMAT
Figure 5.6(a) shows the query asked by the student and the
answer displayed in the form of link which when clicked
shows the video for answering the query in 5.6(b). Also, it is
the same for image answer.
6. FUTURE SCOPE
The system involvesanswering queriesrelatedtocollege.
It holds a combination of decision tree and NLP approaches
to better understand the user's intent to offer smoother
interactions. A future approach for this chatbot could be
including recognition of language and also, it's translation
facilities. Further, the system could also be designed for text
analytics. Additionally, Content moderation could also be
added so as to ensure that the system does not get misused.
7. CONCLUSION
Multimedia Chatbotisdesignedasanultimatevirtual
assistant, helping to execute all kinds of interactionsthatare
currently being performed by the college staff,
circumventing all the unnecessary hassles faced by the
students. Pos tagging is used to classify and solve the
problem. A dynamic database is created using SQL which is
capable of updating the knowledge base when the answer is
not found. A more user accessiblechatsystemusingArtificial
Intelligence is created by the introduction of a simpler input
method using not only text but also voice, creating and
catering for a more personal and convenient experience. The
system’s dialogue interface is being constantly perfected, so
as to correctly interpret user’s natural language and also for
bettering their answering capabilities.
REFERENCES
[1] College Enquiry Chatbot
http://ijritcc.org/download/browse/Volume_5_Issues/
March_17_Volume_5_Issue_3/1492062497_13-04-
2017.pdf
[2] Chatbot for College Management System
https://www.researchgate.net/publication/327667155
_Chat-Bot_For_College_Management_System_Using_AI
[3] College Enquiry Chatbot Using Knowledge in Database
https://www.ijraset.com/fileserve.php?FID=15814
[4] POS tagging
https://www.ijarcsse.com/docs/papers/Volume_7/4_A
pril2017/V7I4-0 140.pdf

IRJET- Multimedia Chatbot using Classification

More Related Content

What's hot (19)

Similar to IRJET- Multimedia Chatbot using Classification (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Multimedia Chatbot using Classification