EPRA International Journal of

Published By :EPRA Publishing

CC License



ISSN (Online) : 2455 – 3662

SJIF Impact Factor :5.148

Monthly Peer Reviewed & Indexed

International Online Journal

Volume: 5 Issue: 6 June 2019

www.eprajournals.com Volume: 5| I ssue: 6 | June 2 01 9

97 EPRA International Journal of

Multidisciplinary Research (IJMR) Peer Reviewed Journal

Volume: 5 | Issue: 6 | June 2019 || SJIF Impact Factor: 5.148 ISSN (Online): 2455 – 3662




Patil Kiran Sanjay 1

P.G. Studen t

Department of Comp Engineering

Sharadchandra Pawar co llege of Engineering

Otur, Pune

Prof. Kurhade N.V 2


Department of Comp Engineering,

Sharadchand ra Pawar college of Engineering

Otur, Pune


In a world that routinely conveys dynamically literary data. It is fundame ntal errand to managing that printed data. There

are various substance examination systems are open to directing and envisioning that data, anyway various strategies may

give less exactness because of the vulnerability of common language. To give the ?ne – g rained examination, in this paper

present e? cient AI calculations for order content data.

To upgrade the accuracy, in proposed system I familiar NLTK python

library with perform common language preparing. The rule purpose of proposed system is to whole up the model for

continuous application by using e

? cient content classi?cation just as clustering calculations and ?nd the exactness of model

using execution measure.

KEYWORDS : Text analytics, TF – IDF, Text classi?cation, Text categorization.


With the fast development of on line data, text

categorization has turned out to be one of the key

procedures for taking care of and sorting out text

information. Text categorization procedures are

utilized to group news stories, to ?nd intriguing data on

the WWW, and to direct a client’s hunt through

hypertext. Since structure text classi?ers by hand is

di ? cult and tedious. In this paper I will investigate and

recognize the bene?ts of di ? erent sort of procedures

like classi?cation and clustering for text c ategorization.

Here I have marked just as non named information for

investigation by utilizing managed just as unsupervised

AI calculations I can classified the information

e ? ciently and after text categorization I will think

about all methods and envision ed which is better for

constant applications. The primary motivation behind proposed

framework is that make summed up model according

to client’s prerequisites, since when we apply AI

calculations on dataset then they gives diverse


Before goin g to arrange the dataset we need to

apply preprocessing on that information and afterward

pass that information preprocessing yield to

classi?cation or clustering calculations as an info. For

information preprocessing here I have utilized natural

language processing (NLP).


A According to Divyansh Khanna, Rohan Sahu,

Veeky Baths, and Bharat Deshpande[2] This study

provides a benchmark to the present research in the

?eld of heart disease prediction. The dataset used is the

Cleveland Heart Disease Dataset, which is to an extent

www.eprajournals.com Volume: 5| I ssue: 6 | June 2 01 9

98 EPRA International Journal of Multidisciplinary Research (IJMR) | ISSN (Online): 2455 – 3662 | SJIF Impact Factor: 5.148

curated, but is a valid standard for research. This paper

has provided details on the comparison of classi?ers

for the detection of heart disease. We have

implemented logistic regression, support vector

machines and neural networks for classi?cation. The

results suggest SVM methodologies as a very good

technique for accurate predicti on of heart disease,

especially considering classi?cation accuracy as a

performance measure. Generalized Regression Neural

Network gives remarkable results, considering its

novelty and unorthodox approach as compared to

classical models.

From this I had ta ken the idea of SVM algorithm for


According to Krunoslav Zubrinic, Mario

Milicevic and Ivona Zakarija[3] In this research we

tested the ability of classi?cation of CMs using simple

classi?ers and bag of words approach that is commonly

used in document classi?cation. In two experiments we

compared the results of classi?cation randomly

selected CMs using three classi?ers. The best results

are achieved using multinomial NB classi?er. On

reduced set of attributes and instances that classi?er

co rrectly classi?ed 79.44 of instances. We believe that

the results are promising, and that with further data

preprocessing and adjustment of the classi?ers they can

be improved.

From this this I had introduced NB classifiers

algorithm in my system for mappi ng the different


According to Thorsten Joachims This [4]paper

introduces support vector machines for text

categorization. It provides both theoretical and

empirical evidence that SVMs are very well suited for

text categorization. The theoretical analysis concludes

that SVMs acknowledge the particular properties of


1. high dimensional feature spaces

2. few irrelevant features (dense concept vector)

3. sparse instance vectors.

The experimental results show that SVMs consistently

achieve good performance on text categorization tasks,

outperforming existing methods substantially and

signi?cantly. With their ability to generalize well in

high dimensional feature spaces, SVMs eliminate the

need for feature selection, making the application of

text categorization considerably easier. Another

advantage of SVMs over the conventional methods is

their robustness. SVMs show good performance in all

experiments, avoiding catastrophic failure, as observed

with the conventional methods on some tasks.

Further more, SVMs do not require any parameter tuning, since they can ?nd good parameter settings

automatically. All this makes SVMs a very promising

and easy – to – use method for learning text classi?ers

from examples.

According to Payal R. Undhad,Dharmesh J.

Bha lodiya[5] Text classi?cation is a data mining

technique used to predict categorical label. Aim of

research on text classi?cation is to improve the quality

of text representation and develop high quality

classi?ers. Text classi?cation process includes

follo wing steps i.e. collection of data documents, data

preprocessing, Indexing, term weighing methods,

classi?cation algorithms and performance measure.

Machine learning techniques have been actively

explored for text classi?cation. Machine learning

algorithm for text classi?cation are Naive Bayes

classi?er, K – nearest neighbor classi?ers, support vector

machine. Text classi?cation is very helpful in the ?eld

of text mining, The volume of electronic information is

increase Day by Day and its extracting knowledge

from these large volumes of data. The classi?cation

problem is the most essential problems in the machine

learning along with data mining literature. This paper

survey on text classi?cation. This survey focused on

the existing literature and explored the documents

representation and an analysis classi?cation algorithms

Term weighting is one of the most vital parts for

construct a text classi?er. The existing classi?cation

methods are compared based on pros and cons. From

the above discussion it is understo od that no single

representation scheme and classi?er can be mentioned

as a general model for any application Di ? erent

algorithms perform di ? erently depending on data

collection.TF – IDF word embedding concept is taken

from this paper for vectorization.

Acc ording to Deokgun Park, Seungyeon Kim,

Jurim Lee, Jaegul Choo, Nicholas Diakopoulos,

and Niklas Elmqvist[1] Current text analytics

methods are either based on manually crafted human –

generated dictionaries or require the user to interpret a

complex, confusi ng, and sometimes nonsensical topic

model generated by the computer. In this paper we

proposed Concept Vector, a novel text analytics system

that takes an visual analytics approach to document

analysis by allowing the user to iteratively de?ned

concepts wi th the aid of automatic recommendations

provided using word embedding. The resulting

concepts can be used for concept – based document

analysis, where each document is scored depending on

how many words related to these concepts it contains.

We crystallized the generalizable lessons as design

guidelines about how visual analytics can help concept

www.eprajournals.com Volume: 5| I ssue: 6 | June 2 01 9

99 EPRA International Journal of Multidisciplinary Research (IJMR) | ISSN (Online): 2455 – 3662 | SJIF Impact Factor: 5.148

based document analysis. We compared our interface

for generating lexica with existing databases and found

that Concept Vector enabled users to generate concepts

mor e e ? ectively using the new system than when using

existing databases. We proposed an advanced model

for concept generation that can incorporate irrelevant

words input and negative words input for bipolar

concepts. We also evaluated our model by comparing

i ts performance with a crowd sourced dictionary for

validity. Finally, we compared Concept Vector to

Empath in an expert review. The text analysis provided

by Concept Vector enables several novel concept –

based document analysis, such as richer sentiment

ana lysis than previous approaches, and such

capabilities can be useful for data journalism or social

media analysis. There are many limitations that

Concept Vector does not solve. Among these, the

selection/integration of multiple heterogeneous training

data according to the target corpus and the automatic

disambiguation of multiple meanings of words

according to the context are promising avenues of

future research.

In proposed system I introduced text categorization on

labeled and non labeled data to create g eneralized

model for real time applications.


The proposed work is on textual dataset, using

classification and clustering machine learning

algorithms perform text categorization . If data is

labeled then text categorization is using classification

otherwise using clustering ML algorithm and find the best algorithm for input dataset by using performance


The ma in purpose of this system is to provide

generalized model for real time applications .

Objectives of System

? To provides generalized model for real time


? To categorized large labeled as well as non

labeled textual dataset efficiently.

? To applying di ? erent ML algorithm for

di ? erent dataset and ?nd accuracy of model

using pe rformance measure.

Scope of System

? To provides efficient text categorization.

? To provide great user experience to users in

their day to day activity this text

categorization to be analyzed.


In today’s world, most of work is doing on

textual data. Huge textual data is very critical to

handle, for maintaining that textual data here used

some machine learning algorithms. If data is labeled

then it can handle using classification ML algorithms

li ke SVM, Naive Bayes.

If data is not labeled then this type of textual

data is group by using clustering ML algorithms like

K – means, Gaussian Mixture Model.

After applying algorithms the main aim of

proposed system is to find the efficient ML algorithm

for particular input dataset using performance measure.

www.eprajournals.com Volume: 5| I ssue: 6 | June 2 01 9

100 EPRA International Journal of Multidisciplinary Research (IJMR) | ISSN (Online): 2455 – 3662 | SJIF Impact Factor: 5.148

Figure 1: Proposed System Architecture


In this research work, the principle spotlight is

on the text categorization, at whatever point data is

labeled or unlabeled by utilizing AI calculations group

free text e ? ciently. Bolster vector machine (SVM)

and guileless Bayes classi?cation calculation for

labeled data and K – means and Gaussian mixture model

(GMM) clustering calculation for non – labeled data.

The principle m otivation behind this

undertaking is to delineate continuous text arranged

issue to fitting AI calculation and ?nd precise

con?dence likelihood of data thing. E ? ciency of AI

calculation is differing with each dataset. By utilizing

execution measure ascerta in the precision model for

classi?cation. After that I will envisioned that outcome

utilizing python libraries.


Using MD5 alg orithm we can calculate more accuracy

of SVM algorithm.


1. Deokgun Park, et al. “Concept Vector: Text Visual

Analytics via Interactive Lexicon Building using Word

Embedding”, IEEE Transactions on Visualization and

Computer Graphics, Vol. 24, NO. 1,2018

2. Divyansh Khanna, et al. “Comparative Study of

Classi ? cation Techn iques (SVM, Logistic Regression and Neural Networks) to Predict the Prevalence of Heart

Disease” International Journal of Machine Learning and

Computing, Vol. 5, No. 5, October 2015.

3. Krunoslav Zubrinic,et al “Comparison of Naive Bayes

and SVM Classi ? er s in Categorization of Concept Maps”

International Journal of computers Issue 3, Volume 7,


4. Thorsten Joachims “Text Categorization with Support

Vector Machines :Learning with Many Relevant Features”

5. Payal R. Undhad, Dharmesh J. Bhalodiya , “Text

Classi ? cation and Classi ? ers: A Comparative Study ”

2017 IJEDR,Volume 5, Issue 2,ISSN: 2321 – 9939

6. M. Berger, K. McDonough, and L.M. Seversky. “cite2vec:

Citation driven document exploration via word

embeddings.” IEEE Transactions on Visualization and

Co mputer Graphics, 23(1):691700, Jan 2017.


8. Lkit:A Toolkit for Natuaral Language Interface


Cite this page

publishpaper. (2019, Nov 23). Retrieved from http://paperap.com/publishpaper-best-essay/

Let’s chat?  We're online 24/7