Techniques and Applications for

Sentiment Analysis
Communications of the ACM | April 2013 | Vol. 56 | nº 4
DOI:10.1145/2436256.2436274

Ronen Feldman
Hebrew University
ronen.feldman@huji.ac.il

Airton Bordin Junior

airtonbjunior@gmail.com
Prof. Dr. Nádia Félix Felipe da Silva
nadia@inf.ufg.br
Agenda
▪ Introduction
▪ Context
▪ Problem
▪ Objective
▪ Sentiment Analysis
▪ Sentiment Lexicon
▪ Applications
▪ Research Issues
▪ Conclusion
Context

3
Objective
▪ Present the main research problems
related to Sentiment Analysis (SA) and
some of the techniques used to solve
them
▪ Review some of the major application
areas where sentiment analysis is
being used today

4
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

5
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

6
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Document Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Assumed that the document contains

an opinion on one main object
expressed by the author of the
document
▪ Two main approaches
▪ Supervised learning
▪ Unsupervised learning

7
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Document Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Supervised learning
▪ Finite set of classes
▪ Training data is available
▪ Classes
▪ Positive/negative/neutral
▪ Numeric scale (stars)
▪ SVM, KNN, Naïve Bayes, Logistic
Regression, Genetic programming

8
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Document Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Important representations
▪ Bag of words
▪ TFIDF
▪ Part of Speech (PoS)
▪ Sentiment Lexicons

9
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Document Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Unsupervised learning
▪ Semantic orientation (SO) specific phrases
▪ PMI (Pointwise Mutual Information) of
the phrase with two sentiment words
▪ PMI(P, W)
▪ Statistical dependence between phrase P and
word W based on their co-occurrence in a
corpus

10
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Document Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Unsupervised learning
▪ The two words used in Turney (2002)
are ‘excellent’ and ‘poor’
▪ The SO measures whether P is closer in
meaning to the positive word
(‘excellent’) or the negative word
(‘poor’)

11
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

12
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Sentence Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Assume
▪ We know the identity of the entity
discussed in the sentence
▪ There is a single opinion in each
sentence

13
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Sentence Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Most approaches are based on

supervised learning
▪ Unsupervised approach is similar of
Turney (2002)

14
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Sentence Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Handle different types of sentences by

different strategies
▪ Sentences that need unique strategies
▪ Conditional
▪ Question
▪ Sarcastic

15
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

16
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Aspect Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ In many cases, people talk about

entities that have many aspects
(attributes) and they have a different
opinion about each of the aspects
▪ Often happens in reviews about
products or in discussion forums

17
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Aspect Level
Focus
Comparative

Sentiment Lexicon Acquisition

“As a long-time Kindle fan I was eager to get my hands on a Fire. There are
some great aspects; the device is quick and for the most part dead-simple to
use. The screen is fantastic with good brightness and excellent color, and a
very wide viewing angle. But there are some downsides too; the small bezel
size makes holding it without inadvertent page-turns difficult, the lack of
buttons makes controls harder, the accessible storage memory is limited to
just 5GB.”

18
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Aspect Level
Focus
Comparative

Sentiment Lexicon Acquisition

“As a long-time Kindle fan I was eager to get my hands on a Fire. There are
some great aspects; the device is quick and for the most part dead-simple to
use. The screen is fantastic with good brightness and excellent color, and a
very wide viewing angle. But there are some downsides too; the small bezel
size makes holding it without inadvertent page-turns difficult, the lack of
buttons makes controls harder, the accessible storage memory is limited to
just 5GB.”

19
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Aspect Level
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Approach the problem of aspect

identification as an information
extraction problem
▪ Use a tagged corpus to train a sequence
classifier such as a Conditional Random
Field (CRF)
▪ Implicit aspects

20
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

21
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Comparative
Focus
Comparative

Sentiment Lexicon Acquisition

▪ In many cases users do not provide a direct

opinion about one product but instead
provide comparable opinions
▪ The goal of SA in this case is to identify the
sentences that contain comparative
opinions, and to extract the preferred
entity(-ies) in each opinion.

22
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

SA – Comparative
Focus
Comparative

Sentiment Lexicon Acquisition

▪ Comparative adjectives adverbs

▪ ‘more,’ ‘less,’ and words ending with –er (for
example, ‘lighter’)
▪ Superlative adjectives and adverbs
▪ ‘most,’ ‘least,’ and words ending with –est (for
example, ‘finest’)
▪ Additional phrases
▪ ‘favor,’ ‘exceed,’ ‘outperform,’ ‘prefer,’ ‘than,’
‘superior,’ ‘inferior,’ ‘number one’

23
 Document Level
Focus
Sentence Level
Sentiment
Analysis
Aspect Level

Focus
Comparative

Sentiment Lexicon Acquisition

24
 Document Level

Sentence Level
Sentiment Analysis
Aspect based

Focus

Sentiment Lexicon
Comparative

Sentiment Lexicon Acquisition

▪ Most crucial resource for most

sentiment analysis algorithms
▪ Acquisition
▪ Manual
▪ Dictionary
▪ Sentiwordnet
▪ Corpus
▪ Sentiment consistency

25
Applications

26
Research Issues

▪ Automatic entity resolution

▪ Sarcasm
▪ Noisy texts
▪ Sentiment to objective statements

27
Conclusion
▪ Reviewed some of the main research
problems within the field of SA
▪ Discussed ways to solve each of these
problems
▪ Described some of the major
applications
▪ Provided a few major open challenges

28
Techniques and Applications for
Sentiment Analysis
Communications of the ACM | April 2013 | Vol. 56 | nº 4
DOI:10.1145/2436256.2436274

Ronen Feldman
Hebrew University
ronen.feldman@huji.ac.il

Airton Bordin Junior

airtonbjunior@gmail.com
Prof. Dr. Nádia Félix Felipe da Silva
nadia@inf.ufg.br
 Introduction

★★★★☆
“The king suite was spacious, clean, and well appointed. The reception staff,
bellmen, and housekeeping were very helpful. Requests for extras from the
maid were always provided. The heating and air conditioning functioned
well; this was good as the weather was variable. The sofa bed was the best
I’ve ever experienced. The king size bed was very comfortable. The building
and rooms are very well soundproofed. The neighborhood is the best for
shopping, restaurants, and access to subway. Only “complaint” has to do
with high-speed Internet access. It’s only available on floors 8–12.”
Introduction