Sentiment Analysis

Nakul Dave
Assistant Professor
Computer Engineering Department
Vishwakarma Government Engineering College - Ahmedabad

Nakul Dave Sentiment Analysis 1 / 76

Sentiment Analysis
Defined in simple terms

Sentiment Analysis is the process of tagging a label of positive,

negative or neutral to the short piece of text.

Figure: Sentiment Analysis1

1
Source of image - https://www.kdnuggets.com/images/sentiment-fig-1-689.jpg
Nakul Dave Sentiment Analysis 2 / 76
Introduction

Sentiment Analysis, also known as Opinion Mining, is the process

of identifying and categorizing opinions, emotions, and sentiments
expressed in text data.
It plays a crucial role in understanding public opinions, customer
feedback, social media sentiments, and more.
Sentiment Analysis finds applications in marketing, customer
service, brand monitoring, product development, and public
sentiment analysis.

Nakul Dave Sentiment Analysis 3 / 76

Sentiment Analysis
Basic Task

It just categorize the text to Positive or Negative

I am happy as I got the first class. ⇒ + V E
I am extremely happy as I am top of the list. ⇒ + V E
He was sad due to his failure. ⇒ -V E
He was deeply sad due to his father death ⇒ -VE.

Figure: Sentiment Analysis - The Basic Technique

Nakul Dave Sentiment Analysis 4 / 76

Where sentiment Analysis is used

Movie - Is this review positive or negative?

Products - What do people think about the new iPhone?

Public Sentiment - How is consumer confidence? Is despair

increasing?

Politics - What do people think about this candidate or issue?

Prediction - Predict election outcomes or marked trends from

sentiment

Nakul Dave Sentiment Analysis 5 / 76

Applications of Sentiment Analysis

Customer Feedback Analysis: Understand customer opinions

about products and services.
Brand Monitoring: Track brand reputation and sentiment across
social media and the web.
Market Research: Analyze public sentiments to assess market
trends and consumer preferences.
Social Media Analysis: Extract insights from social media
discussions and posts.

Nakul Dave Sentiment Analysis 6 / 76

Example of Movie Reviews

“Heartfelt and emotional, This is a must-watch.”

“Disappointing entry in the franchise, lacking the thrill and

intrigue of its predecessors. The plot is convoluted, and the
dialogues are uninspiring.”

“This movie is a weak romantic comedy that fails to make a

lasting impression.”

“This is the greatest comedy movie ever filmed.”

Nakul Dave Sentiment Analysis 7 / 76

Affective States Typology

Emotion: angry, sad, joyful, fearful, ashamed,

proud, elated
Mood: cheerful, gloomy, irritable, listless, depressed, buoyant
Interpersonal stances: friendly, flirtatious, distant, cold, warm,
supportive, contemptuous
Attitudes: liking, loving, hating, valuing, desiring
Personality Traits: nervous, anxious, reckless, morose, hostile, jealous

Nakul Dave Sentiment Analysis 8 / 76

Elements of Sentiment Analysis

Holder (source) of attitude

Target (aspect) of attitude

Type of attitude
From a set of types: like, love, hate, value, desire
Or simple weighted polarity: positive, negative, neutral, together
with strength

Text containing the attitude

Nakul Dave Sentiment Analysis 9 / 76

Tokenization Issue

Capitalization - Preserve for word in all capital letters

word lengthening e.g Cooool
slang words
lol - laughing out loud, brb - be right back, btw - by the way
lmk - let me know, ttyl - talk to you later
Handling emoticons
Handling negation

Nakul Dave Sentiment Analysis 10 / 76

Sentiment Analysis

Simple Task
Is the attitude of this text positive or negative?

More complex
Rank the attitudes of this text from 1 to 5

advanced
Detect the target, source, or complex attitude types

Nakul Dave Sentiment Analysis 11 / 76

Sentiment Analysis - More Complex

Figure: Sentiment Analysis - Advanced

Version2

2
source of image - https://encrypted-tbn0.gstatic.com
Nakul Dave Sentiment Analysis 12 / 76
Sentiment Analysis - More Complex

Table: Emotions with Numerical

Values

Emotion Enumeration
Extreme Negative -3
Very Negative -2
Negative -1
Neutral 0
Figure: Sentiment Analysis - Advanced
Positive 1
Version2
Very Positive 2
Super Positive 3

2
source of image - https://encrypted-tbn0.gstatic.com
Nakul Dave Sentiment Analysis 12
13 / 76
Steps of Sentiment Analysis

Data Collection: Gather text data from various sources such as

social media, reviews, surveys, etc.
Tokenization
Text Preprocessing: Clean and tokenize the text, remove stop
words, and perform stemming/lemmatization.
Feature Extraction
Classification using different classifiers
Rule-based approaches
Naive Bayes
Support Vector Machine
M L / D L Approaches

Nakul Dave Sentiment Analysis 13 / 76

Naive Bayes - A Reminder

Bayes’ theorem
P (d|C) ·P ( C )
P (C|d) =
P (d)

For Naive Bayes classification, the argmax notation can be expressed

as:
Na¨ıve Bayes Classifier

cˆ = arg max P (c|d)

c∈C
= arg max P (d|c)P (c)
c∈C
= arg max P (x 1 , x 2 , . . . , x n |c)P (c)
c∈C

Nakul Dave Sentiment Analysis 14 / 76

Why sentiment Analysis is a challenging task?

This laptop is a great deal.

A great deal of media attention surrounded the release of the new
laptop.
This laptop is a great deal ... and I’ve got a nice bridge you might
be interested in.
This film should be brilliant. It sounds like a great plot, the actors
are first grade, and the supporting cast is good as well, and
Stallone is attempting to deliver a good performance. However, it
can’t hold up.

Nakul Dave Sentiment Analysis 15 / 76

An Experiment - Peng et al. (2002)

Proposed word lists Accuracy Ties

Positive: dazzling, brilliant,
phenomenal, excellent, fantastic
Human 1 58% 75%
Negative: suck, terrible,
awful, unwatchable, hideous
Positive: gripping, mesmerizing,
riveting, spectacular, cool,
awesome, thrilling, badass,
Human 2 excellent, moving, exciting 64% 39%
Negative: bad, cliched, sucks,
boring, stupid, slow
Positive: love, wonderful, best,
great, superb, still, beautiful
Human 3 + Stats 69% 16%
Negative: bad, worst, stupid,
waste, boring, ?, !

Nakul Dave Sentiment Analysis 16 / 76

 Sentiment Analysis - Affective Lexicons

Nakul Dave
Assistant Professor
Computer Engineering Department
Vishwakarma Government Engineering College - Ahmedabad

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 17 / 76

Semantic Lexicon

The general inquirer

M P Q A Subjectivity Cues Lexicon
SentiWordnet
L I W C (Linguistic Inquiry and Word Count)

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 18 / 76

Bing Liu’s Opinion Lexicon

Table: Bing Liu’s Opinion Lexicon

Positive Words Nagetive Words

abound prodigious wow bad grief soreness
abounds prodigiously wowed badly grievance sorrow
abundance prodigy wowing baffle grievances sorrowful
abundant productive wows baffled grieve sorrowfully
accessable productively yay bafflement grieving sorry
accessible proficient youthful baffling grievous sour
acclaim proficiently zeal bait grievously sourly
acclaimed profound zenith balk grim spade
acclamation profoundly zest banal grimace spank
accolade profuse zippy banalize grind spendy

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 19 / 76

The Sentiment lexicon

The General Inquire

Positive (1915 words) and Negative (2291) words Strong vs weak,
active vs passive, overstated vs understated
pleasure, pain, virtue, vice, motivation, cognitive orientation etc.

sentiWordNet
Home page: h t t p : / / s e n t i w o r d n e t . i s t i . c n r . i t /
All Wordnet synsets automatically annotated for degrees of
positivity, negativity, and neutrality/objectiveness

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 20 / 76

Other lexicons

M P Q A Subjectivity Cues Lexicon

Home page: http://www.cs.pitt.edu/mpqa/subj_lexicon.html
6885 words from 8221 lemmas: 2718 positive, 4912 negative
Each word annotated for intensity (strong, weak)

L I W C (Linguistic Inquiry and Word Count 3 )

Home page: http://www.liwc.net/

2300 words, > 70 classes
Negative emotion (bad, weird, hate, problem, tough)
Positive emotion (love, nice, sweet)

3
https://www.liwc.app/
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 21 / 76
General Inquirer for Sentiment Analysis

Figure: General Inquirer for Sentiment Analysis4

4
http://sentiment.christopherpo tts.net/lexicons.html#o pinionlexicon
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 22
23 / 76
M Q PA for Sentiment Analysis

Figure: MQPA for Sentiment Analysis5

5
http://sentiment.christopherpo tts.net/lexicons.html#o pinionlexicon
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 23
24 / 76
SentiWordnet for Sentiment Analysis

Figure: SentiWordnet for Sentiment Analysis6

6
http://sentiment.christopherpo tts.net/lexicons.html#o pinionlexicon
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 24
25 / 76
L I W C Lexicon for Sentiment Analysis

Figure: L I W C Lexicon for Sentiment Analysis7

7
http://sentiment.christopherpotts.net/lexicons.html#op inionlexicon
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 25 / 76
V A D E R (Valence Aware Dictionary and sEntiment
Reasoner)

V A D E R is a lexicon and rule-based sentiment analysis tool

specifically designed for analyzing sentiments in text data,
especially social media content.
V A D E R sentiment analysis assigns a sentiment score to a piece of
text, indicating the text’s sentiment as positive, negative, or
neutral.
V A D E R is trained on a sentiment lexicon that includes words and
their associated sentiment scores.
It also accounts for various language nuances, including
punctuation, capitalization, and context.

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 26 / 76

Sentiment Analysis

Table: V E D A R Sentiment Score for 10 words of each category

P ositiv e Wor ds S entim ent Score Neg a tiv e Wor ds S entim ent Score Neutr a l Wor ds S en tim en t Score
awesome 0.6249 painful -0.4404 ordinary 0
delightful 0.5859 horrible -0.5423 good 0.4404
exciting 0.4939 failure -0.5106 okay 0.2263
happy 0.5719 hate -0.5719 neutral 0
amazing 0.5859 unhappy -0.4215 normal 0
excellent 0.5719 disgusting -0.5267 pleasant 0.5106
love 0.6369 sad -0.4767 fine 0.2023
victory 0 terrible -0.4767 average 0
beautiful 0.5994 dislike -0.3818 indifferent -0.2023
fantastic 0.5574 awful -0.4588 boring -0.3182

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 27 / 76

Valence

Valence represents the emotional positivity or negativity of an

emotion or sentiment.
It reflects how positive or negative an emotion or sentiment is
perceived.
Valence is typically measured on a numerical scale, with negative
values representing negativity and positive values representing
positivity.
Examples of valence:
Positive valence: Happiness, love, excitement
Neutral valence: Indifference, calmness
Negative valence: Sadness, anger, fear

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 28 / 76

Arousal

Arousal represents the intensity or activation level of an emotion

or sentiment.
It reflects how emotionally stimulating or arousing an emotion or
sentiment is perceived.
Arousal is typically measured on a numerical scale, ranging from
low arousal (calm) to high arousal (excited).
Examples of arousal:
Low arousal: Calm, relaxation
Moderate arousal: Interest, surprise
High arousal: Excitement, fear, anger

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 29 / 76

Lexicon of valence and Arousal

valence (the pleasantness of the stimulus)

9: happy, pleased, satisfied, contented, hopeful
1: unhappy, annoyed, unsatisfied, melancholic, despaired, or bored

arousal (the intensity of emotion provoked by the stimulus)

9: stimulated, excited, frenzied, jittery, wide-awake, or aroused
1: relaxed, calm, sluggish, dull, sleepy, or unaroused

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 30 / 76

Valence-Arousal Examples

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 31 / 76

Usage in Sentiment Analysis

In sentiment analysis, valence helps us determine whether a

sentiment is positive, negative, or neutral.
Arousal helps us understand the intensity or emotional charge of
sentiments.
Both dimensions provide a more comprehensive view of emotions
and sentiments in text data.

Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 32 / 76

Arousal and Valence

Figure: Arousal and Valence8

8
Warriner, Amy Beth, Victor Kuperman, and Marc Brysbaert. “Norms of
valence, arousal, and dominance for 13,915 English lemmas”.Behavior Research
Methods 45.4 (2013): 1191-1207.
Nakul Dave S e n t i m e n t A n a l y s i s - Affe ct iv e L e x i c ons 33 / 76
 Learning Affective Lexicons

Nakul Dave
Assistant Professor
Computer Engineering Department
Vishwakarma Government Engineering College - Ahmedabad

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 34 / 76

Learning Sentiment Lexicons

Basic Idea
Adjectives conjoined by “and” have the same polarity
Fair and legitimate, corrupt and brutal
Adjectives conjoined by “but”
Fair but brutal

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 35 / 76

Learning Sentiment Lexicons

Step 1: Label seed set of adjectives

Positive cases: adequate, central, clever, famous, intelligent,
remarkable, reputed, sensitive, slender, thriving ...
Negative cases: contagious, drunken, ignorant, lanky, listless,
primitive, strident, troublesome, unresolved, unsuspecting ...

Step 2: Expand seed set to conjoined adjectives

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 36 / 76

Learning Sentiment Lexicons

Step 3: Construct a graph

Polarity similarity is assigned to each word pair:

Figure: Sentiment Polarity9

9
Hatzivassloglou and Mckeown, 1997
Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 37 / 76
Learning Sentiment Lexicons
Clustering for partitioning the graph into two
Polarity similarity is assigned to each word pair:

Figure: Sentiment Polarity Partitioning10

10
Hatzivassloglou and Mckeown, 1997
Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 38 / 76
Output Polarity Lexicon

Positive
bold decisive disturbing generous good honest important large mature
patient peaceful positive proud sound stimulating straightforward
strange talented vigorous witty ...

Negative
ambiguous cautious cynical evasive harmful hypocritical inefficient
insecure irrational irresponsible minor outspoken pleasant reckless risky
selfish tedious unsupported vulnerable wasteful ...

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 39 / 76

Output Polarity Lexicon

Positive
bold decisive disturbing generous good honest important large
mature patient peaceful positive proud sound stimulating
straightforward strange talented vigorous witty ...

Negative
ambiguous cautious cynical evasive harmful hypocritical inefficient
insecure irrational irresponsible minor outspoken pleasant reckless
risky selfish tedious unsupported vulnerable wasteful ...

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 40 / 76

Sentiment Aspect - Assignment Question

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 41 / 76

Sentiment Aspect - Assignment Question

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 42 / 76

Turney Algorithm

Pseudocode for Turney Algorithm11

Extract a phrasal lexicon from reviews

Learn the polarity of each phrase

Rate a review by the average polarity of its phrases

11
Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised
Classification of Reviews, Peter D. Turney, 2002
Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 43 / 76
Extract two-word phrases with adjectives

J J - Adjectives
NN - Nouns
R B - Adverbs
V B - Verbs

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 44 / 76

Measuring the polarity of the phrases

Positive phrases co-occur more with “excellent”

Negative phrases co-occur more with “poor”
How to measure the co-occurrence?

Pointwise Mutual Information

P (word 1 ,word 2 )
P M I(wor d 1 , word2) = log P (word 1 )P (word 2 )

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 45 / 76

How to estimate PMI?

Query search engine (Altavista)

P(word) estimated by hits(word)\N
P(word1,word2) estimated by hits(word1 N E A R word2)\N

Polarity(phrase) = PMI(phrase, excellent) - PMI(phrase, poor)

hits(Phrase N E A R “excellent”)hits(“poor”)
log 2
hits(Phrase N E A R “poor”)hits(“excellent”)

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 46 / 76

Turney Assignment Question

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 47 / 76

Turney Assignment Question

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 48 / 76

Let’s Calculate Answer

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 49 / 76

Example: A thumbs-up Review

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 50 / 76

Example: A thumbs-down Review

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 51 / 76

Using WordNet to learn polarity

WordNet: online thesaurus

Create positive (“good”) and negative seed-words (“terrible”)
Find Synonyms and Antonyms
Positive Set: Add synonyms of positive words (“well”) and
antonyms of negative words
Negative Set: Add synonyms of negative words (“awful”) and
antonyms of positive words (“evil”)
Repeat, following chains of synonyms
Filter

Nakul Dave L e a r n i n g Affe ct iv e L e x i c o n s 52 / 76

 Computing with Affective Lexicons

Nakul Dave
Assistant Professor
Computer Engineering Department
Vishwakarma Government Engineering College - Ahmedabad

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 53 / 76

Learn word sentiment supervised by online review scores

Review datasets: IM D B, Goodreads, Amazon, Trip Advisor

Each review has a score (1-5, 1-10 etc)
Just count how many times each word occurs with each score (and
normalize).

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 54 / 76

Analyzing polarity of each word in IM D B
How likely is each word to appear in each sentiment class?
Let’s take count(“bad”) in 1-star, 2-star, 3-star etc.

We should use likelihood instead of counts

Make them comparable between words Scaled likelihood:
Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 55 / 76
I M D B Review Polarity Analysis

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 56 / 76

Logical Negation

Is logical negation (no, not) associated with negative sentiment?

Potts experiment:
I Count negation (not, n’t, no, never) in online reviews
I Regress against the review rating

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 57 / 76

More negation in negative sentiment

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 58 / 76

Using Linguistic Intuitions

Using a sentiment lexicon also works.

Some linguistic intuitions on top of that tend to give better results.

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 59 / 76

Handling negation

Example words
Excellent +5
good +3
terrible -5
bad -3

Reversing the polarity

Not Excellent -5
Not good -3
Not terrible +5
Not bad + 3

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 60 / 76

Handling negation

Example words
Excellent +5
good +3
terrible -5
bad -3

Instead, a polarity shift works better

Excellent (5-4) +1
Not good (3-4) -1
Not terrible (-5+4) -1
Not bad (-3+4) 1

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 61 / 76

Handling intensifiers

Intensifiers can be classified into two major categories

Amplifiers (e.g., very) increase the semantic intensity
Downtoners (e.g., slightly) decrease it
Values for intensifiers
Slightly -50
Somewhat -30
Pretty -10
Really +15
Very +25
Extraordinary +50
Most +100

An Example
sleazy: -3, Somewhat sleazy -3 (100% − 30%) = -2.1

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 62 / 76

Intensifiers - Assignment Question

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 63 / 76

Intensifiers - Assignment Question

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 64 / 76

Irrealis moods: where the words may not be reliable

I thought this movie would be as good as the Grinch, but

unfortunately, it wasn’t.
This should have been a great movie.

What are the indicators?

conditional markers (if)
negative polarity items like ‘any’ and ‘anything’
certain (mostly intentional) verbs (expect, doubt),
questions
words enclosed in quotes (which may be factual, but not
necessarily reflective of the author’s opinion)

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 65 / 76

Finding Aspects or attributes

Opinion
The food was great but the service was awful.

Aspects involved
food, service

Nakul Dave C o m p u t i n g w i t h Affe ct iv e L e x i c o n s 66 / 76

 Aspect-based Sentiment Analysis

Nakul Dave
Assistant Professor
Computer Engineering Department
Vishwakarma Government Engineering College - Ahmedabad

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 67 / 76

Frequent phrases + rules

Find all highly frequent phrases across reviews (“fish tacos”)

Filter by rules like “occurs right after sentiment word”
“... great fish tacos” means “fish tacos” a likely aspect

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 68 / 76

Same Phrase but different sentiment

Large – positive or negative

Large screen vs. Large battery

Long – positive or negative

Long battery life vs. Long loading time

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 69 / 76

Explicit vs. Implicit Aspect Expressions

The picture quality of this camera is great – ‘picture quality’ is an

explicit aspect.
This camera is expensive – ‘expensive’ is an implicit aspect
expression describing ‘price’.
Implicit aspect expressions can be very complex as well, e.g., This
camera will not fit in a pocket – “fit in a pocket’ indicates the
aspect ‘size’.

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 70 / 76

Aspect-based Opinion Summarization - An Interesting
Application

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 71 / 76

Aspect-based Product Comparison - An Interesting
Application

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 72 / 76

Sentiment Aspect - Assignment Question

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 73 / 76

Challenges in Sentiment Analysis

Ambiguity: Sentiment expression can be context-dependent and

ambiguous.
Sarcasm and Irony: Detecting sarcastic or ironic statements can
be challenging.
Domain-Specific Language: Sentiment analysis models need to
adapt to different domains and languages.
Negation and Intensifiers: Understanding negation and intensifiers
can affect sentiment polarity.

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 74 / 76

Conclusion

Sentiment Analysis is a powerful tool to extract valuable insights

from text data.
It aids in understanding public opinion, customer feedback, and
brand perception.
Sentiment Analysis has a wide range of applications in various
industries, including marketing, customer service, and market
research.
The field of Sentiment Analysis continues to evolve with
advancements in natural language processing and machine
learning techniques.

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 75 / 76

Questions?

Thank you for your attention!

Any questions?

Nakul Dave A s p e c t - bas e d S e n t i m e n t A n a l y s i s 76 / 76