Volume 8, Issue 12, December – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Sentimental Analysis using NLP

Terisri Paladugula1 Hiranmayee Nandyala2 S V V S S C Ekantha3
B.Tech III Year Sr. Asst Prof B. Tech III Year,
Department of AIM Department of CSE, Department of AIM,
Sri Vasavi Engg. College Sri Vasavi Engg. College, Sri Vasavi Engg. College,
Tadepalligudem Tadepalligudem Tadepalligudem

Puthin Dungala4 Karteek Kishor Ambati5 Jyothi Tanmai Ramisetti6

B. Tech III Year, B. Tech III Year, B. Tech III Year,
Department of AIM Department of AIM Department of AIM,
Sri Vasavi Engg. College, Sri Vasavi Engg. College, Sri Vasavi Engg. College,
Tadepalligudem Tadepalligudem. Tadepalligudem.

Abstract:- Sentiment analysis is a subset of text analysis language processing that monitors public opinion toward a
techniques that uses automatic text polarity detection. specific good or subject. Its main responsibility is to
One of the main responsibilities of NLP (Natural recognize and extract sentiment from a given string. It
Language Processing) is sentiment analysis, often known accepts a string as input and returns a sentiment rating
as opinion mining. In recent years, sentiment analysis between very negative and very positive ([-1 t0 1]). It entails
has gained a lot of popularity. It is meant for people to developing a system to gather and analyze product-related
build a system that can recognize and categorize feedback expressed in tweets, blog posts, reviews, and
sentiment or opinion as it is expressed in an electronic comments. Sentiment analysis has various applications. For
text. Nowadays, people who wish to purchase consumer instance, in marketing, it assists in determining whether an
goods prefer to read user reviews and participate in advertisement campaign or traditional text mining focuses
public online forums where others discuss the product. on the examination of facts, while sentiment analysis
This is because consumers frequently have to make focuses on attitudes. Sentiment analysis research primarily
trade-offs when making purchases. Before making a focuses on three primary areas: opinion summarization,
purchase, a lot of customers read other people's reviews. feature-based sentiment classification, and sentiment
Individuals frequently voice their opinions about several classification. Sentiment classification is the process of
things. Opinion mining has grown in significance as a categorizing entire writings based on the thoughts of the
result. Sentiment analysis is the process of determining if authors regarding particular things. On the other hand,
the expressed opinion about the subject is favorable or feature-based sentiment classification takes into account
negative. Customers must choose which portion of the users' judgments about certain object features. The task of
available data to utilize. Sentiment analysis is the summarizing opinions differs from standard text
technique of locating and removing subjective summarization in that it focuses solely on the product
information from unprocessed data. If we could attributes that have prompted consumer evaluations. Unlike
accurately forecast sentiments, we could be able to the standard text summary, opinion summarization does not
gather online opinions and anticipate the preferences of select a selection of the reviews or rework some of the
online customers. This information could be useful for original lines from the reviews to capture the essential
study in marketing or economics. As of right now, points. Sentiment analysis determines the tone of a given
sentiment classification, feature-based classification, and text by applying a variety of classification algorithms. It
handling negations are the three main issues facing this makes clear if the text is neutral, negative, or favorable.
research community. Large data sets can be used with this technique to produce
information that is useful in a variety of situations. For
Keywords:- Numpy, Pandas, TF-IDF, Tfidf Vectorizer, instance, the judgment expressed in the line "The life of the
Linear SVC, Train-Test Split, Accuracy Score, Classification battery of this mobile is too compressed" is negative and
Report, Confusion Matrix, user Input, Vectorization, focuses on the "life of the battery" of the mobile object
Prediction, Preprocessing, Text Classification, Supervised (target). This level of in-depth analysis is necessary for
Learning, Machine Learning Model, Scikit-Learn. many everyday applications because it is necessary to know
which aspects of a product are liked and disliked by
I. INTRODUCTION customers to upgrade it. Subjectivity and sentiment have not
been used to categorize such information. Linguistics is the
Sentiment analysis, sometimes referred to as opinion study of human (natural) language interaction and computer
mining, is the process of locating and extracting subjective science, artificial intelligence, and natural language
information from source materials using computational processing (NLP). For example, it identifies the product
linguistics, natural language processing, and text analysis. features that the reviewer has commented on and indicates
Put another way, sentiment analysis is a kind of natural whether the comments are neutral, positive, or negative.

IJISRT23DEC522 www.ijisrt.com 494

Volume 8, Issue 12, December – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Phrase-based SA involves classifying the phrases in a Confusion matrices represent counts from anticipated
sentence based on their polarity. Determining whether a text and actual values. The output "TN" stands for proper poor
expresses positive or negative weather views is part of the which indicates the variety of bad examples categorized
emotion analysis task. appropriately. Further, "TP" stands for actual effectiveness
which indicates the number of high-quality examples labeled
 Objective: accurately. The period "FP" suggests a fake fantastic price,
The goal of our task is to create a model that predicts i.e., the wide variety of actual bad examples classified as
the polarity of the given text i.e. whether the given text is tremendous; and "FN" approaches a false bad fee which is
positive, negative, or neutral. the variety of actual effective examples classified as terrible.
Performance metrics of a set of rules are accuracy, precision,
II. RELATED WORK recollect, and F1 rating, which are calculated based on the
above-stated TP, TN, FP, and FN.
 TF-IDF Vectorizer: It is a numeric that reflects the
common words that are used in Natural Language A. Accuracy:
Processing. The accuracy of an algorithm is represented as the ratio
of correctly classified patients to thetotal number of patients.
 Term Frequency (TF): It tells the no. of times a
particular term is repeated in the given document.  Accuracy=TN+TP / TN+FP+FN+TP

 It is Calculated as below:  Precision:

The precision of an algorithm is represented as the ratio
of correctly classified patients with the disease to the total
patients predicted to have the disease.

 Inverse Document Frequency (IDF): It measures the  Precision=TP / TP+FP

importance of a word in the document.
 Recall:
 It is Calculated as below: Recall metric is defined as the ratio of correctly
classified diseased patients (TP) divided by the total number
 IDF(t, D) = log (total no. of document in corpus D/ no. of patients who have the disease. The perception behind
of documents containing terms t+1 ) recalls how many patients have been classified as having the
disease. The recall is also called sensitivity.
 TF-IDF Score: It is the multiplication of TF and IDF.
The higher the value higher the importance of the word.  Recall=TP / TP+FN

 It is Calculated as below:  F1 Score:

The F1 score is also known as the F-measure. The F1
TF-IDF(t,d,D)=TF(t,d)×IDF(t,D) score states the equilibrium between the precision and the
recall.
 Advantages of TF-IDF:
 F1 score=2*precision*recall / precision + recall
 Term Importance
 Common Word Penalization IV. PROPOSED SYSTEM
 Dimensionality Reduction
Sentiment analysis is a natural language processing
 Document Similarity
technique that helps identify the neutral, positive, and
 Language Independence
negative aspects of data. It is frequently used on textual data
 Information Retrieval to assist companies in tracking customer sentiment toward
their brands and products in feedback from customers and in
III. CONFUSION MATRIX figuring out what those needs are.
Confusion matrix is a totally popular measure used There are various ways to determine a text's sentiment
while fixing type problems. It could be applied to binary score. Using positive and negative word counts with
classification as well as to multiclass-type problems. normalization is one such technique. Utilizing Support
Vector Machines (SVM) is an additional technique. The
SVM model predicts text emotions and ascertains if the text
is neutral, positive, or negative.

Fig 1 Confusion Matrix

IJISRT23DEC522 www.ijisrt.com 495

Volume 8, Issue 12, December – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 The following Steps could be Included in a Suggested
Sentiment Analysis System:

 Data Collection:
Gathering information from a range of sources,
including customer feedback forms and social media
platforms.

 Data Preprocessing:
The removal of stop words, punctuation, special Fig 2 System Archiecture
characters, etc. from the collected data.
VI. RESULTS AND DISCUSSIONS
 Feature Extraction:
Taking characteristics like word frequency and n-  Get the Dataset:
grams out of the preprocessed data. The dataset we took contains two columns sentence
and sentiment. The sentence contains the text and the
sentiment contains the polarity of the sentence.
 Model Training:
Using the features that were extracted, train a machine
 Importing Libraries:
learning model to categorize the text into positive, negative,
To perform data preprocessing using Python, we need
or neutral categories.
to import some predefined Python libraries. These libraries
are used to perform some specific jobs. There are three
 Model Evaluation:
specific libraries that we will use for data preprocessing,
Assessing the trained model's effectiveness with
which are: They are two main important libraries which are
metrics like recall, accuracy, and precision, among others.
used, they are:
 Model Deployment:
 Pandas:
Introducing the learned model into a real-world setting
The last library is the Pandas library, which is one of the
so it can be applied to the classification of fresh data.
most famous Python libraries and is used for importing and
managing the datasets. It is an open-source data manipulation
V. DESIGN andanalysis library. It will be imported as below:
The process of classifying sentiment using machine
 Matplotlib:
learning algorithms is illustrated in the flowchart. Six steps
The second library is matplotlib, which is a Python 2D
make up the process: gathering data, preprocessing the data,
plotting library, and with this library, we need to import a
machine learning algorithm, results presentation, sentiment
sub-library pyplot. This library is used to plot any type of
classification, and algorithm.
charts in Python for the code. It will be imported as below
To explain the procedure in more detail, first data is
 Import Pandas as pd
gathered from multiple sources. The gathered data is then
 Import Matplotlib.Pyplot as plt
preprocessed to eliminate any unnecessary information and
 From Sklearn Import *
transform it into a format that machine learning algorithms
can understand. Subsequently, a machine learning algorithm
 Importing Data:
is employed on the preprocessed data to train a model
The downloaded data is imported into the Python code
capable of categorizing text data sentiment. Next, the
file as a Data frame using pandas module.
sentiment of fresh text data is classified using the trained
model. The format in which the results are presented makes
them simple to read and analyze. Using the trained model,
the SVM algorithm is then used to categorize the sentiment
of fresh text data.

Fig 3 Importing File

 Feature Extraction:
In this step, we convert text data to numerical features using TF-IDF vectorization.

Fig 4 Converting Text to Numericals

IJISRT23DEC522 www.ijisrt.com 496

Volume 8, Issue 12, December – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Split Data:
In this step, we are going to split the data into x_train, x_test, y_train, and y_test.

Fig 5 Spliting Data

 Train the Linear SVC Model:

In this step, we train the model.

Fig 6 Training

 Make Predictions:

Fig 7 Predictions

 Evaluate the Model:

Fig 8 Evaluation

 Print the Confusion Matrix and Accuracy:

Fig 9 Accuracy

The result of the sentimental analysis tells the polarity

of the text given. It ultimately takes text as an input and tells
the polarity of it. It helps the business people to know how
the customers are feeling about their products so that they
can make changes accordingly as per the customer’s
interest. It helps in making changes easily and helps in
improving the quality of the product. It not only products
but also helps movies understand the pulse of the audience.
It makes the directors understand the willingness and type of
genre the audience wants and make films accordingly. Not
only movies and products sentimental analysis helps in
many sectors.

Fig 10 Positive Response

IJISRT23DEC522 www.ijisrt.com 497

Volume 8, Issue 12, December – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
REFERENCES

[1]. In 2019, IEEE Access published a paper titled

"Twitter Sentiment Analysis Based on Ordinal
Regression," written by S. E. Saad and J. Yang.
[2]. Multi-Strategy Sentiment Analysis of Consumer
Reviews Based on Semantic Fuzziness, Y. Fang, H.
Tan, and J. Zhang, IEEE Access, vol. 6, pp. 20625-
20631, 2018.
[3]. The article "Tourism Mobile App With Aspect-Based
Sentiment Classification Framework for Tourist
Reviews" was published in May 2019 in IEEE
Transactions on Consumer Electronics, volume 65,
issue 2, pages 233-242, by M. Afzaal, M. Usman,
and A. Fong.
Fig 11 Average Response [4]. "Halal Products on Twitter: Data Extraction and
Sentiment Analysis Using Stack of Deep Learning
Algorithms," A. Feizollah, S. Ainin, N. B. Anuar, N.
A. B. Abdullah, and M. Hazim, IEEE Access, vol. 7,
pp. 83354–83362, 2019.
[5]. A study titled "An efficient hybrid filter and
evolutionary wrapper approach for sentiment analysis
of various topics on Twitter" was published in
Knowledge-Based Systems on March 15, 2020. It
was completed by Mohammad A. Hassonah, Ibrahim
Aljarah, Hossam Faris, Rizik Al-Sayyed, Ali Radan,
Ala' M. Al-Zoubi, and Almaraz.

Fig 12 Negative Response

VII. FUTURE SCOPE

 Multimodal Sentiment Analysis Identifying sentiments

across various media formats by analyzing emotions in
text, photos, audio, and video.
 The Aspect-Based Sentiment Analysis method identifies
sentiments associated with discrete elements in text data
by concentrating on particular aspects or features.
 Sentiment expressions within textual content can be used
to identify a broad spectrum of emotions and minute
differences in sentiment.
 Sentiment analysis in healthcare refers to the process of
examining patient feedback and healthcare-related
content to ascertain attitudes regarding mental health and
healthcare services.
 Sentiment Analysis in Conversational AI: Adding
sentiment analysis skills to chatbots and virtual assistants
will help them comprehend user emotions and react to
them sympathetically.

IJISRT23DEC522 www.ijisrt.com 498