SlideShare a Scribd company logo

MTech Seminar Presentation [IIT-Bombay]

Sagar Ahire, profile picture
Sagar Ahire

The document presents a seminar by Sagar Ahire from IIT Bombay, focusing on various lexical resources for sentiment analysis. It covers key tools like SentiWordNet, So-Cal, and WordNet-Affect, detailing their methodologies, structures, and how sentiment is represented. The discussion also highlights the limitations of classifier-based approaches and the advantages of lexicon-based methods in determining sentiment in text.

TechnologyEducation
Related topics:
Natural Language Processing • IIT Bombay • Sentiment Analysis Methods
1 of 108
Downloaded 363 times
1
2
3
4
5
Most read
6
7
8
9
Most read
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Most read
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
Resources for Sentiment Analysis Seminar Presentation Sagar Ahire 133050073 IIT Bombay 02 May, 2014 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 1 / 48
Roadmap 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 2 / 48
Introduction Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 3 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Wordnet-Affect, created semi-automatically, with affect information for each synset Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Wordnet-Affect, created semi-automatically, with affect information for each synset Indian-Language Sentiwordnet, created by projecting the English Sentiwordnet Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Classifier-based Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Classifier-based Lexicon-based Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Domain Specificity (Example: Movie reviews mentioning ‘writer’, ‘plot’, etc.) [Bro01] Lack of Context (Example: ‘good’ vs ‘not good’ vs ‘not very good’) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Domain Specificity (Example: Movie reviews mentioning ‘writer’, ‘plot’, etc.) [Bro01] Lack of Context (Example: ‘good’ vs ‘not good’ vs ‘not very good’) The lexicon-based approach aims at solving these problems. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Choices for lexical unit: Word Word sense Phrase, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Choices for lexical unit: Word Word sense Phrase, etc. Choices for sentiment: Fixed categorization into ‘positive’ and ‘negative’ Graded sets like ‘strongly positive’, ‘mildly positive’, ‘neutral’, ‘mildly negative’, ‘strongly negative’ Score in an interval like [0, 1] or [−1, +1] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
Introduction Sentiment Lexicons Approaches for Creation Manual Automatic Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 8 / 48
Sentiwordnet Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 9 / 48
Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Support for graded sentiment labels Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Support for graded sentiment labels Support for both sentiment classification and subjectivity detection Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Each synset s is given three sentiment scores: Positive score Pos(s) Negative score Neg(s) Objective score Obj(s) Pos(s) + Neg(s) + Obj(s) = 1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Each synset s is given three sentiment scores: Positive score Pos(s) Negative score Neg(s) Objective score Obj(s) Pos(s) + Neg(s) + Obj(s) = 1 Example Synset beautifula: Pos = 0.75, Neg = 0.00, Obj = 0.25 a URL: http://sentiwordnet.isti.cnr.it/search.php?q=beautiful Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations 3 Training of a team of ternary classifiers Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations 3 Training of a team of ternary classifiers 4 Classification of each Wordnet synset using the classifiers Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
SO-CAL Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 13 / 48
SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Graded sentiment label Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Graded sentiment label Low coverage, but high accuracy Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Negation Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Negation Irrealis Blocking Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
SO-CAL Structure Structure of SO-CAL Sentiment scoring: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
SO-CAL Structure Structure of SO-CAL Sentiment scoring: Words are scored in [−5, +5] Intensifiers and negation further act upon these scores Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
SO-CAL Structure Structure of SO-CAL Sentiment scoring: Words are scored in [−5, +5] Intensifiers and negation further act upon these scores Examples good: +3 monstrosity: −5 masterpiece: +5 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
Wordnet-Affect Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 17 / 48
Wordnet-Affect Introduction to Wordnet-Affect Wordnet-Affect [SV04] is a semi-automatically generated sentiment lexicon made using Wordnet. It associates affective information with each synset. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 18 / 48
Wordnet-Affect Introduction to Wordnet-Affect Wordnet-Affect [SV04] is a semi-automatically generated sentiment lexicon made using Wordnet. It associates affective information with each synset. Its salient features are: Highly detailed Ability to handle sentiment differently depending on emotion Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 18 / 48
Wordnet-Affect Structure Structure of Wordnet-Affect Wordnet-Affect = Wordnet + Affect Information. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 19 / 48
Wordnet-Affect Structure Structure of Wordnet-Affect Wordnet-Affect = Wordnet + Affect Information. Affect is represented using the following: An a-label which represents the emotion, The valency which indicates the sentiment. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 19 / 48
Wordnet-Affect Structure Structure of Wordnet-Affect The a-label is a tree of emotions starting at a root node with each leaf node corresponding to a synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 20 / 48
Wordnet-Affect Structure Structure of Wordnet-Affect The a-label is a tree of emotions starting at a root node with each leaf node corresponding to a synset. The valency can be any of positive, negative, neutral or ambiguous. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 20 / 48
Wordnet-Affect Structure root mental-state cognitive-state affective-state mood emotion positive-emotion joy elation love worship negative-emotion sadness melancholy shame embarrassment . . . . . . physical-state . . . Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 21 / 48
Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Manual creation of initial resource Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Manual creation of initial resource Automatic expansion using Wordnet relations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
Indian-Language Sentiwordnets Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 23 / 48
Indian-Language Sentiwordnets Introduction to Indian-Language Sentiwordnets Indian-language Sentiwordnets can be created using Wordnet projection [JRB10]. This approach has the following salient features: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 24 / 48
Indian-Language Sentiwordnets Introduction to Indian-Language Sentiwordnets Indian-language Sentiwordnets can be created using Wordnet projection [JRB10]. This approach has the following salient features: Easy to create once backing resources are available No reduplication of effort Use of tried-and-tested representations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 24 / 48
Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Find the corresponding Hindi synset using Indowordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Find the corresponding Hindi synset using Indowordnet. Assign sentiment scores from English synset to Hindi synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
Conclusions Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 26 / 48
Conclusions A Comparison of the Resources Criterion SWN SO-CAL WN-Affect IL-SWN Sentiment 3 x [0, 1] [−5, +5] Affect 3 x [0, 1] Lexical Unit Synset Word Synset Synset Backing Resource Wordnet None Wordnet SWN + In- dowordnet Creation Automatic Manual Automatic Projection No of Entries 117,000 5,000 900 16,000 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 27 / 48
Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Lexical Unit: Word, Synset or Higher Representations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Lexical Unit: Word, Synset or Higher Representations Sentiment: Labels, Graded Scores or Affect Information Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
Conclusions Concluding Remarks: Creation Approach Manual Approach Automatic Approach High annotation accuracy Low annotation accuracy High time investment Low time investment More details supported Less details supported Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 29 / 48
Conclusions Concluding Remarks: Lexical Unit Word Synset Unreliable for polysemous words Reliable for polysemous words No pre-processing required Requires WSD Projection is comparatively difficult Projection is comparatively easier Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 30 / 48
Conclusions Concluding Remarks: Sentiment Graded scores have been shown to be better than mere labels in general. Moreover, a graded score resource can always be converted to a label-based resource. Affect information can help in specialized circumstances. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 31 / 48
Conclusions Future Work Possible directions in the future: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 32 / 48
Conclusions Future Work Possible directions in the future: Automatic resources for higher-level lexical units like phrases, trees, etc. Manual resources for synsets Manual lexicons for Indian languages Techniques for building dynamic resources to incorporate ‘netspeak’ and other slang Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 32 / 48
Conclusions References I Julian Brooke, A semantic approach to automatic text sentiment analysis, M.A. thesis, Stanford University, 2001. Andrea Esuli and Fabrizio Sebastiani, SentiWordNet: A publicly available lexical resource for opinion mining, Proceedings of the 5th Conference on Language Resources and Evaluation (LREC-06), 2006, pp. 417–422. Andrea Esuli, Automatic generation of lexical resources for opinion mining: Models, algorithms and applications, Ph.D. thesis, Universita di Pisa, 2008. Christiane Fellbaum, Wordnet: An electronic lexical database, A Bradford Book, 1998. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 33 / 48
Conclusions References II Vasileios Hatzivassiloglou and Kathleen R. McKeown, Predicting the semantic orientation of adjectives, Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics and Eighth Conference of the European Chapter of the Association for Computational Linguistics, Association for Computational Linguistics, 1997, pp. 174–181. Aditya Joshi, Balamurali A R, and Pushpak Bhattacharyya, A fall-back strategy for sentiment analysis in hindi: a case study, Proceedings of ICON 2010: 8th International Conference on Natural Language Processing, Macmillan Publishers, India, 2010. Jaap Kamps, Maarten Marx, Robert J. Mokken, and Maarten de Rijke, Using wordnet to measure semantic orientations of adjectives, Proceedings of LREC-04, 4th International Conference on Language Resources and Evaluation, 2004, pp. 1115–1118. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 34 / 48
Conclusions References III Ellen Riloff and Janyce Wiebe, Learning extraction patterns for subjective expressions, Proceedings of the 2003 Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics, 2003, pp. 105–112. Carlo Strapparava and Alessandro Valitutti, WordNet-Affect: an affective extension of WordNet, Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC-04), 2004, pp. 1083–1086. Peter D. Turney and Michael L. Littman, Measuring praise and criticism: Inference of semantic orientation from association, ACM Transactions on Information Systems 21 (2003), no. 4, 315–346. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 35 / 48
Additional Slides Wordnet Wordnet Wordnet [Fel98] is a lexical database organized by word sense. The fundamental unit of storage is called a synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 36 / 48
Additional Slides Wordnet Wordnet Wordnet [Fel98] is a lexical database organized by word sense. The fundamental unit of storage is called a synset. An Example Synset brilliant, superba: of surpassing excellence “a brilliant performance”; “a superb actor” a URL: http://wordnetweb.princeton.edu/perl/webwn?s=brilliant Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 36 / 48
Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Antonymy Meronymy Hypernymy Hyponymy Similar to, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Antonymy Meronymy Hypernymy Hyponymy Similar to, etc. These relations are helpful in creating the training set for classifying synsets to create Sentiwordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Graph Expansion using Wordnet [KMMdR04] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Graph Expansion using Wordnet [KMMdR04] Classification using Wordnet Glosses [Esu08] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Adapting Wordnet Glosses to Subjectivity Detection [Esu08] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Adapting Wordnet Glosses to Subjectivity Detection [Esu08] Bootstrapping Subjective Expressions from a Corpus [RW03] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
Additional Slides Structure of SO-CAL Adjectives Adjectives were collected from a 500-document corpus and annotated with a sentiment score from −5 to +5. Examples good: +3 sleazy: −3 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 40 / 48
Additional Slides Structure of SO-CAL Nouns, Verbs, Adverbs, Multiwords This was extended to other parts of speech and multiword expressions, for a total of about 5,000 words. Examples monstrosity: −5 masterpiece: +5 inspire: +2 funny: +2 vs. act funny: −1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 41 / 48
Additional Slides Structure of SO-CAL Intensifiers and Downtoners Intensifiers are words that increase sentiment intensity while downtoners are words that reduce sentiment intensity. For example extraordinarily and somewhat. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 42 / 48
Additional Slides Structure of SO-CAL Intensifiers and Downtoners Intensifiers are words that increase sentiment intensity while downtoners are words that reduce sentiment intensity. For example extraordinarily and somewhat. Intensifiers and downtoners are modeled as percentage modifiers. Examples slightly: −50% extraordinarily: +50% Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 42 / 48
Additional Slides Structure of SO-CAL Negation Negation is modeled as a numeric shift of value 4 towards the opposite sentiment. Examples good: +3 ⇒ not good: −1 atrocious: −5 ⇒ not atrocious: −1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 43 / 48
Additional Slides Structure of SO-CAL Irrealis Blocking An irrealis marker is a word that indicates that the sentiment may not be reliable because the event hasn’t actually happened. For example, ‘would’, ‘expect’, ‘if’, quotation marks, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 44 / 48
Additional Slides Structure of SO-CAL Irrealis Blocking An irrealis marker is a word that indicates that the sentiment may not be reliable because the event hasn’t actually happened. For example, ‘would’, ‘expect’, ‘if’, quotation marks, etc. Sentences with irrealis markers are ignored for sentiment analysis. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 44 / 48
Additional Slides Sentiwordnet Creation Seed Set Two seed sets are created: Lp for positive synsets Ln for negative synsets Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 45 / 48
Additional Slides Sentiwordnet Creation Seed Set Two seed sets are created: Lp for positive synsets Ln for negative synsets Each synset representation consists of: The terms The defninition The sample phrases Explicit indication of negation Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 45 / 48
Additional Slides Sentiwordnet Creation Wordnet Expansion Relations of Wordnet used for expansion: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 46 / 48
Additional Slides Sentiwordnet Creation Wordnet Expansion Relations of Wordnet used for expansion: Direct antonymy Similarity Derived from Pertains to Attribute Also see Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 46 / 48
Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: No of iterations of expansion (0, 2, 4, 6) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: No of iterations of expansion (0, 2, 4, 6) Learning algorithm (SVM, Rocchio) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48
Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Positive vs. Not Positive Negative vs. Not Negative Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48
Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Positive vs. Not Positive Negative vs. Not Negative The results are combined as: Positive Not Positive Negative Objective Negative Not Negative Positive Objective Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48

More Related Content

PDF
M.Tech Final Seminar
santoshborate2
 
PPTX
Distance Measurement by Ultrasonic Sensor
Edgefxkits & Solutions
 
PDF
Speech recognition project report
Sarang Afle
 
PPTX
Street light controlling using Microcontroller
9989476539
 
PDF
Difference between uat testing and sit testing
TestingXperts
 
PDF
IRJET - Sign Language to Speech Conversion Gloves using Arduino and Flex Sens...
IRJET Journal
 
PDF
Real time conversion of sign language to text and speech presentation
HarshKSatheesh
 
DOCX
VTU final year project report
athiathi3
 
M.Tech Final Seminar
santoshborate2
 
Distance Measurement by Ultrasonic Sensor
Edgefxkits & Solutions
 
Speech recognition project report
Sarang Afle
 
Street light controlling using Microcontroller
9989476539
 
Difference between uat testing and sit testing
TestingXperts
 
IRJET - Sign Language to Speech Conversion Gloves using Arduino and Flex Sens...
IRJET Journal
 
Real time conversion of sign language to text and speech presentation
HarshKSatheesh
 
VTU final year project report
athiathi3
 

What's hot (20)

PPTX
Smart Door locking system using arduino
BhawnaSingh351973
 
PDF
Esp32 cam arduino-123
Victor Sue
 
PPTX
Smart water tank
Suraksha Sanghavi
 
DOC
Audio Steganography java project
Tutorial Learners
 
PDF
Distance Measurement Using Ultrasonic Sensor and Nodemcu
IRJET Journal
 
PPTX
Smart door lock
aswin5432
 
PPTX
Performance trends and alerts with ThingSpeak IoT
Anoush Najarian
 
PDF
Alcohol Detection System in Vehicle using Arduino
IRJET Journal
 
PDF
Project Report Distance measurement system
kurkute1994
 
PDF
Sonar Project Report
Sumit Sapra
 
PPTX
IoT based Garbage Monitoting System
Ranjan Gupta
 
PPTX
Accelerometer sensor
INTAKHAB KHAN
 
PDF
Fiber Optic Sensors, Fiber Optical Temperature Sensor - Rugged Monitoring
rugged_monitoring
 
PPTX
Vechicle accident prevention using eye bilnk sensor ppt
satish 486
 
PPTX
Air and sound pollution monitoring system
AbhinavRajput49
 
PPTX
IoT enabled automatic aquaponics system
Nishmi Suresh
 
PPTX
Automatic Light Control using LDR and PIR Sensor
Ziauddin Ridoy
 
PPTX
Temperature Sensors – Types
elprocus
 
PPTX
Pre-Defense CSE Thesis Presentation in BAUST
Naiyan Noor
 
ODP
Sensor Technology ppt
mansimore1
 
Smart Door locking system using arduino
BhawnaSingh351973
 
Esp32 cam arduino-123
Victor Sue
 
Smart water tank
Suraksha Sanghavi
 
Audio Steganography java project
Tutorial Learners
 
Distance Measurement Using Ultrasonic Sensor and Nodemcu
IRJET Journal
 
Smart door lock
aswin5432
 
Performance trends and alerts with ThingSpeak IoT
Anoush Najarian
 
Alcohol Detection System in Vehicle using Arduino
IRJET Journal
 
Project Report Distance measurement system
kurkute1994
 
Sonar Project Report
Sumit Sapra
 
IoT based Garbage Monitoting System
Ranjan Gupta
 
Accelerometer sensor
INTAKHAB KHAN
 
Fiber Optic Sensors, Fiber Optical Temperature Sensor - Rugged Monitoring
rugged_monitoring
 
Vechicle accident prevention using eye bilnk sensor ppt
satish 486
 
Air and sound pollution monitoring system
AbhinavRajput49
 
IoT enabled automatic aquaponics system
Nishmi Suresh
 
Automatic Light Control using LDR and PIR Sensor
Ziauddin Ridoy
 
Temperature Sensors – Types
elprocus
 
Pre-Defense CSE Thesis Presentation in BAUST
Naiyan Noor
 
Sensor Technology ppt
mansimore1
 
Ad

Similar to MTech Seminar Presentation [IIT-Bombay] (20)

PDF
Sentiwordnet [IIT-Bombay]
Sagar Ahire
 
PPT
ppt on sentiment analysis using various techniques
NiharikaKhanna19
 
PPT
sa.ppt
INyomanSwitrayana
 
PPT
sa-mincut-aditya.ppt
aashnareddy1
 
PPT
An overview of concepts of Sentiment Analysis
Ravi Kumar Lanke
 
PPT
Sentiment+Analysis.ppt
visheshs4
 
PPTX
Tutorial of Sentiment Analysis
Fabio Benedetti
 
PDF
#like or #fail - How Can Computers Tell the Difference?
Mark Cieliebak
 
PDF
A SURVEY OF S ENTIMENT CLASSIFICATION TECHNIQUES USED FOR I NDIAN REGIONA...
ijcsa
 
PPTX
Lac presentation
Roseline Antai
 
PDF
A SURVEY OF SENTIMENT CLASSSIFICTION TECHNIQUES
Journal For Research
 
PDF
The SentiME System at the SSA Challenge Task 1
Giuseppe Rizzo
 
PPTX
To Label or Not? Advances and Open Challenges in SE-specific Sentiment Analysis
Nicole Novielli
 
PDF
Arabic SentiWordNet in Relation to SentiWordNet 3.0
Waqas Tariq
 
PDF
A survey on approaches for performing sentiment analysis ijrset october15
International Journal of Advance Research and Innovative Ideas in Education
 
PDF
SemEval - Aspect Based Sentiment Analysis
Aditya Joshi
 
PPTX
Lecture 3: Structuring Unstructured Texts Through Sentiment Analysis
Marina Santini
 
PDF
A Survey On Sentiment Analysis And Opinion Mining Techniques
Sabrina Green
 
PDF
A Survey on Sentiment Analysis and Opinion Mining.pdf
Mandy Brown
 
PDF
Kishaloy Haldar and Wenqiang Lei - WESST - Sentiment Analysis of Social Media
NUS Institute of Applied Learning Sciences and Educational Technology
 
Sentiwordnet [IIT-Bombay]
Sagar Ahire
 
ppt on sentiment analysis using various techniques
NiharikaKhanna19
 
sa.ppt
INyomanSwitrayana
 
sa-mincut-aditya.ppt
aashnareddy1
 
An overview of concepts of Sentiment Analysis
Ravi Kumar Lanke
 
Sentiment+Analysis.ppt
visheshs4
 
Tutorial of Sentiment Analysis
Fabio Benedetti
 
#like or #fail - How Can Computers Tell the Difference?
Mark Cieliebak
 
A SURVEY OF S ENTIMENT CLASSIFICATION TECHNIQUES USED FOR I NDIAN REGIONA...
ijcsa
 
Lac presentation
Roseline Antai
 
A SURVEY OF SENTIMENT CLASSSIFICTION TECHNIQUES
Journal For Research
 
The SentiME System at the SSA Challenge Task 1
Giuseppe Rizzo
 
To Label or Not? Advances and Open Challenges in SE-specific Sentiment Analysis
Nicole Novielli
 
Arabic SentiWordNet in Relation to SentiWordNet 3.0
Waqas Tariq
 
A survey on approaches for performing sentiment analysis ijrset october15
International Journal of Advance Research and Innovative Ideas in Education
 
SemEval - Aspect Based Sentiment Analysis
Aditya Joshi
 
Lecture 3: Structuring Unstructured Texts Through Sentiment Analysis
Marina Santini
 
A Survey On Sentiment Analysis And Opinion Mining Techniques
Sabrina Green
 
A Survey on Sentiment Analysis and Opinion Mining.pdf
Mandy Brown
 
Kishaloy Haldar and Wenqiang Lei - WESST - Sentiment Analysis of Social Media
NUS Institute of Applied Learning Sciences and Educational Technology
 
Ad

More from Sagar Ahire (7)

PDF
Wordnet-Affect [IIT-Bombay]
Sagar Ahire
 
PDF
Paper Presentation: HMM-based Alignment
Sagar Ahire
 
PDF
Paper Presentation: A Pendulum Swung Too Far
Sagar Ahire
 
PDF
NLP Asignment Final Presentation [IIT-Bombay]
Sagar Ahire
 
PDF
Sarcasm & Thwarting in Sentiment Analysis [IIT-Bombay]
Sagar Ahire
 
PPTX
Sentiment Analysis
Sagar Ahire
 
PPTX
Neuro-fuzzy systems
Sagar Ahire
 
Wordnet-Affect [IIT-Bombay]
Sagar Ahire
 
Paper Presentation: HMM-based Alignment
Sagar Ahire
 
Paper Presentation: A Pendulum Swung Too Far
Sagar Ahire
 
NLP Asignment Final Presentation [IIT-Bombay]
Sagar Ahire
 
Sarcasm & Thwarting in Sentiment Analysis [IIT-Bombay]
Sagar Ahire
 
Sentiment Analysis
Sagar Ahire
 
Neuro-fuzzy systems
Sagar Ahire
 

Recently uploaded (20)

PPTX
cloud_computing_Infrastucture_as_cloud_p
mainakbhattacharya20
 
PPTX
TLE Review Electricity (Electricity).pptx
JayPolicarpio2
 
PPTX
Tartificialntelligence_presentation.pptx
ry7r52mzb4
 
PDF
A comparative analysis of optical character recognition models for extracting...
IAESIJAI
 
PDF
Network Security Unit 5.pdf for BCA BBA.
Serpent6
 
PPTX
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
muhammadmuneebnaeem7
 
PDF
Architecting across the Boundaries of two Complex Domains - Healthcare & Tech...
Peter Tan
 
PPTX
SOPHOS-XG Firewall Administrator PPT.pptx
AgnesMunisi
 
PDF
7 ChatGPT Prompts to Help You Define Your Ideal Customer Profile.pdf
SOFTTECHHUB
 
PDF
Empathic Computing: Creating Shared Understanding
Mark Billinghurst
 
PDF
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
IAESIJAI
 
PDF
NewMind AI Weekly Chronicles - August'25-Week II
NewMind AI
 
PDF
Profit Center Accounting in SAP S/4HANA, S4F28 Col11
Libreria ERP
 
PDF
Encapsulation_ Review paper, used for researhc scholars
gurumoop
 
PPTX
OMC Textile Division Presentation 2021.pptx
JahangirAlam816069
 
PDF
August Patch Tuesday
Ivanti
 
PDF
Heart disease approach using modified random forest and particle swarm optimi...
IAESIJAI
 
PDF
Assigned Numbers - 2025 - Bluetooth® Document
Bloombase
 
PDF
Univ-Connecticut-ChatGPT-Presentaion.pdf
ry7r52mzb4
 
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
SandeepSingh286037
 
cloud_computing_Infrastucture_as_cloud_p
mainakbhattacharya20
 
TLE Review Electricity (Electricity).pptx
JayPolicarpio2
 
Tartificialntelligence_presentation.pptx
ry7r52mzb4
 
A comparative analysis of optical character recognition models for extracting...
IAESIJAI
 
Network Security Unit 5.pdf for BCA BBA.
Serpent6
 
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
muhammadmuneebnaeem7
 
Architecting across the Boundaries of two Complex Domains - Healthcare & Tech...
Peter Tan
 
SOPHOS-XG Firewall Administrator PPT.pptx
AgnesMunisi
 
7 ChatGPT Prompts to Help You Define Your Ideal Customer Profile.pdf
SOFTTECHHUB
 
Empathic Computing: Creating Shared Understanding
Mark Billinghurst
 
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
IAESIJAI
 
NewMind AI Weekly Chronicles - August'25-Week II
NewMind AI
 
Profit Center Accounting in SAP S/4HANA, S4F28 Col11
Libreria ERP
 
Encapsulation_ Review paper, used for researhc scholars
gurumoop
 
OMC Textile Division Presentation 2021.pptx
JahangirAlam816069
 
August Patch Tuesday
Ivanti
 
Heart disease approach using modified random forest and particle swarm optimi...
IAESIJAI
 
Assigned Numbers - 2025 - Bluetooth® Document
Bloombase
 
Univ-Connecticut-ChatGPT-Presentaion.pdf
ry7r52mzb4
 
Building Integrated photovoltaic BIPV_UPV.pdf
SandeepSingh286037
 

MTech Seminar Presentation [IIT-Bombay]

  • 1. Resources for Sentiment Analysis Seminar Presentation Sagar Ahire 133050073 IIT Bombay 02 May, 2014 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 1 / 48
  • 2. Roadmap 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 2 / 48
  • 3. Introduction Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 3 / 48
  • 4. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 5. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 6. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 7. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 8. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Wordnet-Affect, created semi-automatically, with affect information for each synset Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 9. Introduction Overview Overview An overview of today’s presentation: This presentation covers lexical resources for sentiment analysis. Four resources are covered, each using a different approach for representation and creation: Sentiwordnet, created automatically, with 3 graded scores per synset SO-CAL, created manually, with a graded score per word Wordnet-Affect, created semi-automatically, with affect information for each synset Indian-Language Sentiwordnet, created by projecting the English Sentiwordnet Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 4 / 48
  • 10. Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
  • 11. Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
  • 12. Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Classifier-based Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
  • 13. Introduction Sentiment Analysis Sentiment Analysis Sentiment Analysis: Determining the opinion expressed in a text Approaches: Classifier-based Lexicon-based Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 5 / 48
  • 14. Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
  • 15. Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Domain Specificity (Example: Movie reviews mentioning ‘writer’, ‘plot’, etc.) [Bro01] Lack of Context (Example: ‘good’ vs ‘not good’ vs ‘not very good’) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
  • 16. Introduction Sentiment Analysis Why Lexicon-based Approach? The classifier-based approach has the following drawbacks: Domain Specificity (Example: Movie reviews mentioning ‘writer’, ‘plot’, etc.) [Bro01] Lack of Context (Example: ‘good’ vs ‘not good’ vs ‘not very good’) The lexicon-based approach aims at solving these problems. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 6 / 48
  • 17. Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
  • 18. Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Choices for lexical unit: Word Word sense Phrase, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
  • 19. Introduction Sentiment Lexicons Sentiment Lexicons A sentiment lexicon is a sentiment database for language units of the form (lexical unit, sentiment). Choices for lexical unit: Word Word sense Phrase, etc. Choices for sentiment: Fixed categorization into ‘positive’ and ‘negative’ Graded sets like ‘strongly positive’, ‘mildly positive’, ‘neutral’, ‘mildly negative’, ‘strongly negative’ Score in an interval like [0, 1] or [−1, +1] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 7 / 48
  • 20. Introduction Sentiment Lexicons Approaches for Creation Manual Automatic Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 8 / 48
  • 21. Sentiwordnet Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 9 / 48
  • 22. Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
  • 23. Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
  • 24. Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Support for graded sentiment labels Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
  • 25. Sentiwordnet Introduction to Sentiwordnet Sentiwordnet [ES06] is an automatically generated sentiment lexicon made using Wordnet. Its salient features are: High coverage Support for graded sentiment labels Support for both sentiment classification and subjectivity detection Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 10 / 48
  • 26. Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
  • 27. Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Each synset s is given three sentiment scores: Positive score Pos(s) Negative score Neg(s) Objective score Obj(s) Pos(s) + Neg(s) + Obj(s) = 1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
  • 28. Sentiwordnet Structure Structure of Sentiwordnet Sentiwordnet = Wordnet + Sentiment Information. Each synset s is given three sentiment scores: Positive score Pos(s) Negative score Neg(s) Objective score Obj(s) Pos(s) + Neg(s) + Obj(s) = 1 Example Synset beautifula: Pos = 0.75, Neg = 0.00, Obj = 0.25 a URL: http://sentiwordnet.isti.cnr.it/search.php?q=beautiful Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 11 / 48
  • 29. Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
  • 30. Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
  • 31. Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
  • 32. Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations 3 Training of a team of ternary classifiers Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
  • 33. Sentiwordnet Creation Creation Steps The top-level steps in the algorithm to create Sentiwordnet are as follows: 1 Selection of seed set 2 Expansion using Wordnet’s semantic relations 3 Training of a team of ternary classifiers 4 Classification of each Wordnet synset using the classifiers Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 12 / 48
  • 34. SO-CAL Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 13 / 48
  • 35. SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
  • 36. SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
  • 37. SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Graded sentiment label Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
  • 38. SO-CAL Introduction to SO-CAL SO-CAL is a system that uses a manually-constructed lexicon. Its salient features are: Highly detailed lexicon Graded sentiment label Low coverage, but high accuracy Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 14 / 48
  • 39. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 40. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 41. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 42. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 43. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Negation Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 44. SO-CAL Structure Features Used SO-CAL classifies words into various features and treats each feature differently in the lexicon. They are: Adjectives Nouns, Verbs, Adverbs and Multiwords Intensifiers and Downtoners Negation Irrealis Blocking Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 15 / 48
  • 45. SO-CAL Structure Structure of SO-CAL Sentiment scoring: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
  • 46. SO-CAL Structure Structure of SO-CAL Sentiment scoring: Words are scored in [−5, +5] Intensifiers and negation further act upon these scores Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
  • 47. SO-CAL Structure Structure of SO-CAL Sentiment scoring: Words are scored in [−5, +5] Intensifiers and negation further act upon these scores Examples good: +3 monstrosity: −5 masterpiece: +5 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 16 / 48
  • 48. Wordnet-Affect Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 17 / 48
  • 49. Wordnet-Affect Introduction to Wordnet-Affect Wordnet-Affect [SV04] is a semi-automatically generated sentiment lexicon made using Wordnet. It associates affective information with each synset. Its salient features are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 18 / 48
  • 50. Wordnet-Affect Introduction to Wordnet-Affect Wordnet-Affect [SV04] is a semi-automatically generated sentiment lexicon made using Wordnet. It associates affective information with each synset. Its salient features are: Highly detailed Ability to handle sentiment differently depending on emotion Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 18 / 48
  • 51. Wordnet-Affect Structure Structure of Wordnet-Affect Wordnet-Affect = Wordnet + Affect Information. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 19 / 48
  • 52. Wordnet-Affect Structure Structure of Wordnet-Affect Wordnet-Affect = Wordnet + Affect Information. Affect is represented using the following: An a-label which represents the emotion, The valency which indicates the sentiment. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 19 / 48
  • 53. Wordnet-Affect Structure Structure of Wordnet-Affect The a-label is a tree of emotions starting at a root node with each leaf node corresponding to a synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 20 / 48
  • 54. Wordnet-Affect Structure Structure of Wordnet-Affect The a-label is a tree of emotions starting at a root node with each leaf node corresponding to a synset. The valency can be any of positive, negative, neutral or ambiguous. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 20 / 48
  • 55. Wordnet-Affect Structure root mental-state cognitive-state affective-state mood emotion positive-emotion joy elation love worship negative-emotion sadness melancholy shame embarrassment . . . . . . physical-state . . . Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 21 / 48
  • 56. Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
  • 57. Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Manual creation of initial resource Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
  • 58. Wordnet-Affect Creation Creation Steps Wordnet-Affect was created using the following steps: Manual creation of initial resource Automatic expansion using Wordnet relations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 22 / 48
  • 59. Indian-Language Sentiwordnets Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 23 / 48
  • 60. Indian-Language Sentiwordnets Introduction to Indian-Language Sentiwordnets Indian-language Sentiwordnets can be created using Wordnet projection [JRB10]. This approach has the following salient features: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 24 / 48
  • 61. Indian-Language Sentiwordnets Introduction to Indian-Language Sentiwordnets Indian-language Sentiwordnets can be created using Wordnet projection [JRB10]. This approach has the following salient features: Easy to create once backing resources are available No reduplication of effort Use of tried-and-tested representations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 24 / 48
  • 62. Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
  • 63. Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Find the corresponding Hindi synset using Indowordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
  • 64. Indian-Language Sentiwordnets Creation Creation Steps The process of projecting a Sentiwordnet has the following steps: Fetch a synset from the English Sentiwordnet. Find the corresponding Hindi synset using Indowordnet. Assign sentiment scores from English synset to Hindi synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 25 / 48
  • 65. Conclusions Roadmap: We Are Here 1 Introduction 2 Sentiwordnet 3 SO-CAL 4 Wordnet-Affect 5 Indian-Language Sentiwordnets 6 Conclusions Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 26 / 48
  • 66. Conclusions A Comparison of the Resources Criterion SWN SO-CAL WN-Affect IL-SWN Sentiment 3 x [0, 1] [−5, +5] Affect 3 x [0, 1] Lexical Unit Synset Word Synset Synset Backing Resource Wordnet None Wordnet SWN + In- dowordnet Creation Automatic Manual Automatic Projection No of Entries 117,000 5,000 900 16,000 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 27 / 48
  • 67. Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
  • 68. Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
  • 69. Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Lexical Unit: Word, Synset or Higher Representations Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
  • 70. Conclusions Concluding Remarks To conclude, there are three choices in making a sentiment lexicon: Creation Approach: Manual, Automatic, Semi-Automatic or Projection Lexical Unit: Word, Synset or Higher Representations Sentiment: Labels, Graded Scores or Affect Information Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 28 / 48
  • 71. Conclusions Concluding Remarks: Creation Approach Manual Approach Automatic Approach High annotation accuracy Low annotation accuracy High time investment Low time investment More details supported Less details supported Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 29 / 48
  • 72. Conclusions Concluding Remarks: Lexical Unit Word Synset Unreliable for polysemous words Reliable for polysemous words No pre-processing required Requires WSD Projection is comparatively difficult Projection is comparatively easier Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 30 / 48
  • 73. Conclusions Concluding Remarks: Sentiment Graded scores have been shown to be better than mere labels in general. Moreover, a graded score resource can always be converted to a label-based resource. Affect information can help in specialized circumstances. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 31 / 48
  • 74. Conclusions Future Work Possible directions in the future: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 32 / 48
  • 75. Conclusions Future Work Possible directions in the future: Automatic resources for higher-level lexical units like phrases, trees, etc. Manual resources for synsets Manual lexicons for Indian languages Techniques for building dynamic resources to incorporate ‘netspeak’ and other slang Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 32 / 48
  • 76. Conclusions References I Julian Brooke, A semantic approach to automatic text sentiment analysis, M.A. thesis, Stanford University, 2001. Andrea Esuli and Fabrizio Sebastiani, SentiWordNet: A publicly available lexical resource for opinion mining, Proceedings of the 5th Conference on Language Resources and Evaluation (LREC-06), 2006, pp. 417–422. Andrea Esuli, Automatic generation of lexical resources for opinion mining: Models, algorithms and applications, Ph.D. thesis, Universita di Pisa, 2008. Christiane Fellbaum, Wordnet: An electronic lexical database, A Bradford Book, 1998. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 33 / 48
  • 77. Conclusions References II Vasileios Hatzivassiloglou and Kathleen R. McKeown, Predicting the semantic orientation of adjectives, Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics and Eighth Conference of the European Chapter of the Association for Computational Linguistics, Association for Computational Linguistics, 1997, pp. 174–181. Aditya Joshi, Balamurali A R, and Pushpak Bhattacharyya, A fall-back strategy for sentiment analysis in hindi: a case study, Proceedings of ICON 2010: 8th International Conference on Natural Language Processing, Macmillan Publishers, India, 2010. Jaap Kamps, Maarten Marx, Robert J. Mokken, and Maarten de Rijke, Using wordnet to measure semantic orientations of adjectives, Proceedings of LREC-04, 4th International Conference on Language Resources and Evaluation, 2004, pp. 1115–1118. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 34 / 48
  • 78. Conclusions References III Ellen Riloff and Janyce Wiebe, Learning extraction patterns for subjective expressions, Proceedings of the 2003 Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics, 2003, pp. 105–112. Carlo Strapparava and Alessandro Valitutti, WordNet-Affect: an affective extension of WordNet, Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC-04), 2004, pp. 1083–1086. Peter D. Turney and Michael L. Littman, Measuring praise and criticism: Inference of semantic orientation from association, ACM Transactions on Information Systems 21 (2003), no. 4, 315–346. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 35 / 48
  • 79. Additional Slides Wordnet Wordnet Wordnet [Fel98] is a lexical database organized by word sense. The fundamental unit of storage is called a synset. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 36 / 48
  • 80. Additional Slides Wordnet Wordnet Wordnet [Fel98] is a lexical database organized by word sense. The fundamental unit of storage is called a synset. An Example Synset brilliant, superba: of surpassing excellence “a brilliant performance”; “a superb actor” a URL: http://wordnetweb.princeton.edu/perl/webwn?s=brilliant Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 36 / 48
  • 81. Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
  • 82. Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Antonymy Meronymy Hypernymy Hyponymy Similar to, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
  • 83. Additional Slides Wordnet Semantic Relations in Wordnet Wordnet synsets are linked to each other by relations called semantic relations. Some of them are: Antonymy Meronymy Hypernymy Hyponymy Similar to, etc. These relations are helpful in creating the training set for classifying synsets to create Sentiwordnet. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 37 / 48
  • 84. Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
  • 85. Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
  • 86. Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
  • 87. Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Graph Expansion using Wordnet [KMMdR04] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
  • 88. Additional Slides Background Sentiment Classification Initial work that automatically detected the sentiment of a word led to today’s modern lexicons. This included: Use of conjunction-separated adjectives [HM97] PMI-based Extraction using Web Queries [TL03] Graph Expansion using Wordnet [KMMdR04] Classification using Wordnet Glosses [Esu08] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 38 / 48
  • 89. Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
  • 90. Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Adapting Wordnet Glosses to Subjectivity Detection [Esu08] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
  • 91. Additional Slides Background Subjectivity Detection Work that identifies whether a term is indeed subjective is necessary to filter out objective words from sentiment classification. This includes: Adapting Wordnet Glosses to Subjectivity Detection [Esu08] Bootstrapping Subjective Expressions from a Corpus [RW03] Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 39 / 48
  • 92. Additional Slides Structure of SO-CAL Adjectives Adjectives were collected from a 500-document corpus and annotated with a sentiment score from −5 to +5. Examples good: +3 sleazy: −3 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 40 / 48
  • 93. Additional Slides Structure of SO-CAL Nouns, Verbs, Adverbs, Multiwords This was extended to other parts of speech and multiword expressions, for a total of about 5,000 words. Examples monstrosity: −5 masterpiece: +5 inspire: +2 funny: +2 vs. act funny: −1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 41 / 48
  • 94. Additional Slides Structure of SO-CAL Intensifiers and Downtoners Intensifiers are words that increase sentiment intensity while downtoners are words that reduce sentiment intensity. For example extraordinarily and somewhat. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 42 / 48
  • 95. Additional Slides Structure of SO-CAL Intensifiers and Downtoners Intensifiers are words that increase sentiment intensity while downtoners are words that reduce sentiment intensity. For example extraordinarily and somewhat. Intensifiers and downtoners are modeled as percentage modifiers. Examples slightly: −50% extraordinarily: +50% Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 42 / 48
  • 96. Additional Slides Structure of SO-CAL Negation Negation is modeled as a numeric shift of value 4 towards the opposite sentiment. Examples good: +3 ⇒ not good: −1 atrocious: −5 ⇒ not atrocious: −1 Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 43 / 48
  • 97. Additional Slides Structure of SO-CAL Irrealis Blocking An irrealis marker is a word that indicates that the sentiment may not be reliable because the event hasn’t actually happened. For example, ‘would’, ‘expect’, ‘if’, quotation marks, etc. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 44 / 48
  • 98. Additional Slides Structure of SO-CAL Irrealis Blocking An irrealis marker is a word that indicates that the sentiment may not be reliable because the event hasn’t actually happened. For example, ‘would’, ‘expect’, ‘if’, quotation marks, etc. Sentences with irrealis markers are ignored for sentiment analysis. Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 44 / 48
  • 99. Additional Slides Sentiwordnet Creation Seed Set Two seed sets are created: Lp for positive synsets Ln for negative synsets Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 45 / 48
  • 100. Additional Slides Sentiwordnet Creation Seed Set Two seed sets are created: Lp for positive synsets Ln for negative synsets Each synset representation consists of: The terms The defninition The sample phrases Explicit indication of negation Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 45 / 48
  • 101. Additional Slides Sentiwordnet Creation Wordnet Expansion Relations of Wordnet used for expansion: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 46 / 48
  • 102. Additional Slides Sentiwordnet Creation Wordnet Expansion Relations of Wordnet used for expansion: Direct antonymy Similarity Derived from Pertains to Attribute Also see Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 46 / 48
  • 103. Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
  • 104. Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: No of iterations of expansion (0, 2, 4, 6) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
  • 105. Additional Slides Sentiwordnet Creation Classifiers 8 classifiers were created differing in: No of iterations of expansion (0, 2, 4, 6) Learning algorithm (SVM, Rocchio) Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 47 / 48
  • 106. Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48
  • 107. Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Positive vs. Not Positive Negative vs. Not Negative Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48
  • 108. Additional Slides Sentiwordnet Creation Classifiers Each ternary classifier is a sum of 2 binary classifiers: Positive vs. Not Positive Negative vs. Not Negative The results are combined as: Positive Not Positive Negative Objective Negative Not Negative Positive Objective Sagar Ahire (IIT Bombay) Sentiment Resources 02 May, 2014 48 / 48