WNAffect

WNAffect is a python package to get the emotion of a word using WordNet resources.

Requirements

This package requires NLTK, WordNet 1.6 (UNIX-like) and WordNet-Domains 3.2.

WARNING: In the wn-domains-3.2/wn-affect-1.1/a-hierarchy.xml file, you should correct simpathy by sympathy.

Introduction

WordNet is a lexical resource database for the English language. WordNet Domains 3.2 is a XML resource providing a set of emotional words organized in a tree:

 ┌physical-state
├behaviour
├trait
├sensation
├situation┐
│ └emotion-eliciting-situation
├signal┐
│ └edonic-signal
root┤
│ ┌cognitive-state
│ ├cognitive-affective-state
└mental-state┤
│ ┌mood
└affective-state┤
│ ┌emotionlessness
│ ┌apathy┤
│ │ └neutral-languor
│ ┌neutral-emotion┤
│ │ └neutral-unconcern┐
│ │ │ ┌distance
│ │ └indifference┤
│ │ └withdrawal
│ │ ┌thing
│ │ ├pensiveness
│ │ ├gravity┐
│ │ │ └earnestness
│ │ ├ambiguous-fear┐
│ │ │ └reverence
│ │ │ ┌ambiguous-hope
│ │ ├ambiguous-expectation┤
│ │ │ └fever┐
│ │ │ └buck-fever
│ ├ambiguous-emotion┤
│ │ │ ┌unrest
│ │ │ ├tumult
│ │ ├ambiguous-agitation┤
│ │ │ │ ┌electricity
│ │ │ └stir┤
│ │ │ └sensation
│ │ └surprise┐
│ │ │ ┌surprise
│ │ │ ├stupefaction
│ │ └astonishment┤
│ │ └wonder┐
│ │ └awe
│ │ ┌gratitude┐
│ │ │ └gratefulness
│ │ ├levity┐
│ │ │ └playfulness
│ │ ├positive-fear┐
│ │ │ └frisson
│ │ ├fearlessness┐
│ │ │ └security┐
│ │ │ └confidence
│ │ ├positive-expectation┐
│ │ │ └anticipation┐
│ │ │ └positive-suspense
│ │ │ ┌self-esteem
│ │ ├self-pride┤
│ │ │ ├amour-propre
│ │ │ └ego
│ │ │ ┌attachment
│ │ │ ├protectiveness
│ │ ├affection┤
│ │ │ ├soft-spot
│ │ │ └regard
│ │ │ ┌gusto
│ │ │ ├exuberance
│ │ ├enthusiasm┤
│ │ │ └eagerness┐
│ │ │ └enthusiasm-ardor
│ │ │ ┌hopefulness
│ │ │ ├encouragement
│ │ ├positive-hope┤
│ │ │ └optimism┐
│ │ │ └sanguinity
│ │ │ ┌placidity
│ │ │ ├coolness
│ │ ├calmness┤
│ │ │ │ ┌peace
│ │ │ └tranquillity┤
│ │ │ └easiness┐
│ │ │ └positive-languor
│ │ │ ┌worship
│ │ │ ├love-ardor
│ │ │ ├amorousness
│ │ │ ├puppy-love
│ │ │ ├devotion
│ │ ├love┤
│ │ │ ├lovingness┐
│ │ │ │ └warmheartedness
│ │ │ ├benevolence┐
│ │ │ │ └beneficence
│ │ │ └loyalty
│ ├positive-emotion┤
│ │ │ ┌amusement
│ │ │ ├exuberance
│ │ │ ├happiness
│ │ │ ├bonheur
│ │ │ ├gladness
│ │ │ ├rejoicing
│ │ │ ├elation┐
│ │ │ │ └euphoria
│ │ │ ├exultation┐
│ │ │ │ └triumph
│ │ │ │ ┌bang
│ │ │ ├exhilaration┤
│ │ │ │ └titillation
│ │ ├joy┤
│ │ │ ├contentment┐
│ │ │ │ │ ┌satisfaction-pride
│ │ │ │ │ ├fulfillment
│ │ │ │ └satisfaction┤
│ │ │ │ ├complacency┐
│ │ │ │ │ └smugness
│ │ │ │ └gloat
│ │ │ │ ┌comfortableness
│ │ │ ├belonging┤
│ │ │ │ └closeness┐
│ │ │ │ └togetherness
│ │ │ │ ┌hilarity
│ │ │ ├merriment┤
│ │ │ │ ├jollity
│ │ │ │ └jocundity
│ │ │ │ ┌buoyancy
│ │ │ └cheerfulness┤
│ │ │ └carefreeness
│ │ │ ┌fondness
│ │ │ ├captivation
│ │ │ ├preference┐
│ │ │ │ └weakness
│ │ │ ├approval┐
│ │ │ │ └favor
│ │ │ ├admiration┐
│ │ │ │ └hero-worship
│ │ └liking┤
│ │ │ ┌kindheartedness
│ │ │ ├compatibility
│ │ ├sympathy┤
│ │ │ ├empathy┐
│ │ │ │ └identification
│ │ │ └positive-concern┐
│ │ │ └softheartedness
│ │ │ ┌amicability
│ │ └friendliness┤
│ │ ├brotherhood
│ │ └good-will
└emotion┤
│ ┌ingratitude
│ ├daze
│ │ ┌self-depreciation
│ ├humility┤
│ │ └meekness
│ │ ┌commiseration
│ │ ├tenderness
│ ├compassion┤
│ │ └mercifulness┐
│ │ └forgiveness
│ │ ┌hopelessness
│ │ ├resignation┐
│ │ │ └defeatism
│ ├despair┤
│ │ ├pessimism┐
│ │ │ └cynicism
│ │ └discouragement┐
│ │ └despair-intimidation
│ │ ┌conscience
│ │ ├self-disgust
│ ├shame┤
│ │ │ ┌self-consciousness
│ │ │ ├shamefacedness
│ │ │ ├chagrin
│ │ └embarrassment┤
│ │ ├discomfiture
│ │ ├abashment
│ │ └confusion
│ │ ┌discomfiture
│ │ ├distress
│ │ ├negative-concern
│ │ ├anxiousness
│ │ ├insecurity
│ │ ├edginess
│ │ ├sinking
│ │ ├scruple
│ ├anxiety┤
│ │ │ ┌stewing
│ │ │ ├tumult
│ │ ├negative-agitation┤
│ │ │ └fidget┐
│ │ │ └impatience
│ │ ├solicitude
│ │ ├anxiousness
│ │ ├angst
│ │ └jitteriness
│ │ ┌alarm
│ │ ├creeps
│ │ ├horror
│ │ ├hysteria
│ │ ├panic
│ │ ├scare
│ │ ├stage-fright
│ │ ├fear-intimidation
│ │ ├negative-unconcern┐
│ │ │ └heartlessness┐
│ │ │ └cruelty
│ ├negative-fear┤
│ │ │ ┌trepidation
│ │ │ ├negative-suspense
│ │ │ ├chill
│ │ ├apprehension┤
│ │ │ │ ┌shadow
│ │ │ └foreboding┤
│ │ │ └presage
│ │ │ ┌shyness
│ │ └timidity┤
│ │ │ ┌hesitance
│ │ └diffidence┤
│ │ └unassertiveness
└negative-emotion┤
│ ┌disinclination
│ ├unfriendliness
│ ├antipathy
│ ├disapproval
│ ├contempt
│ ┌dislike┤
│ │ │ ┌repugnance
│ │ ├disgust┤
│ │ │ └nausea
│ │ └alienation┐
│ │ └isolation
├general-dislike┤
│ │ ┌abhorrence
│ │ ├misanthropy
│ │ ├misogamy
│ │ ├misogyny
│ │ ├misology
│ │ ├misopedia
│ │ ├murderousness
│ │ ├despisal
│ │ ├misoneism┐
│ │ │ └misocainea
│ │ │ ┌maleficence
│ │ ├malevolence┤
│ │ │ ├vindictiveness
│ │ │ └malice
│ ├hate┤
│ │ │ ┌animosity
│ │ │ ├class-feeling
│ │ │ ├antagonism
│ │ │ ├aggression
│ │ │ ├belligerence┐
│ │ │ │ └warpath
│ │ └hostility┤
│ │ │ ┌heartburning
│ │ │ ├sulkiness
│ │ │ ├grudge
│ │ └resentment┤
│ │ │ ┌covetousness
│ │ └envy┤
│ │ └jealousy
│ │ ┌infuriation
│ │ ├umbrage
│ │ ├huffiness
│ │ ├dander
│ │ ├indignation┐
│ │ │ └dudgeon
│ │ │ ┌wrath
│ │ ├fury┤
│ │ │ └lividity
│ └anger┤
│ │ ┌pique
│ │ ├frustration
│ ├annoyance┤
│ │ ├displeasure
│ │ ├harassment
│ │ └aggravation
│ │ ┌irascibility
│ └bad-temper┤
│ └fit
│ ┌dolefulness
│ ├misery
│ ├forlornness
│ ├weepiness
│ ├downheartedness
│ ├cheerlessness┐
│ │ └joylessness
│ │ ┌gloom
│ ├melancholy┤
│ │ ├world-weariness
│ │ └heavyheartedness
└sadness┤
│ ┌attrition
│ ┌regret-sorrow┤
│ │ │ ┌guilt
│ │ └compunction┤
│ │ └repentance
├sorrow┤
│ │ ┌self-pity
│ │ ├grief┐
│ │ │ └dolor
│ └lost-sorrow┤
│ │ ┌woe
│ └mournfulness┤
│ └plaintiveness
│ ┌demoralization
│ ├helplessness
│ ├dysphoria
└depression┤
├oppression┐
│ └weight
└despondency┐
└blue-devils

(printed using pptree)

The WNAffect python package allows to find the emotion of a given word and to navigate in the tree. This package contains two classes: WNAffect and Emotion. These two classes are described below.

Documentation

WNAffect

The WNAffect object allows to load WordNet 1.6 and Word Domains 3.2 resources.

WNAffect(wordnet16_dir, wn_domains_dir)

This is the constructor of a WNAffect object. It requires two arguments:

• wordnet16_dir the wordnet-1.6 folder
• wn_domains_dir the wn-domains-3.2 folder

Example:

wna=WNAffect('wordnet-1.6/', 'wn-domains-3.2/')

get_emotion(word, pos)

This function returns the emotion of a given word.

• word the word
• pos the part-of-speech tag of the word. You can get the part-of-speech tag using a third party library such as NLTK. The tag should be one of the Pen Treebank tag set.
• Returns an Emotion object

emo=wna.get_emotion('angry', 'JJ')

get_emotion_synset(offset)

If you use WordNet in your own project, you can directly get the emotion of a synset using this function.

• offset WordNet offset of the synset
• Returns: an Emotion object

Example:

emo=wna.get_emotion_synset(5574157)

Emotion

The Emotion object is a node in the tree of emotions. The leafs are emotions and the other nodes are categories of emotions.

• name the name of the Emotion
• parent the parent Emotion in the tree
• level the level of the node in the tree
• children the array of children Emotion nodes

The Emotion class also contains a static dict Emotion.emotions containing the emotion names as keys and the emotion objects as values.

Notice that Emotion implements __str__. Indeed, print(emo) is equivalent to print(emo.name).

get_level(level)

Searches in the parent nodes the emotion at the given level.

• level the level to look for
• Returns the Emotion object at the given level

Example:

parent=emo.get_level(emo.level-1)

nb_children()

Count the number of all the children from the current Emotion to the leaves.

• Returns the number

Example:

n=emo.nb_children()

Emotion.printTree(emotion=None)

You should instantiate WNAffect before calling this static function.

Prints the tree of emotions from the given node to the leaves.

• emotion: the emotion to print from. If not provided, the method prints the whole tree.

Example:

Emotion.printTree(Emotion.emotions["annoyance"])

Output:

 ┌pique
├frustration
annoyance┤
├displeasure
├harassment
└aggravation

(printed using pptree)

Example

To import WNAffect package:

>>>fromwnaffectimportWNAffect>>>fromemotionimportEmotion# if needed

The first step is to instantiate a WNAffect object to load the resources

>>>wna=WNAffect('wordnet-1.6/', 'wn-domains-3.2/')

Then, you can get the emotion of a word using the get_emotion function. This function requires the part-of-speech tag. You can get the part-of-speech tag using a third party library such as NLTK.

>>>emo=wna.get_emotion('angry', 'JJ')
>>>print(emo)
anger

What is the parent emotion?

>>>parent=emo.get_level(emo.level-1)
>>>print(parent)
general-dislike

And the root emotion?

>>>root=emo.get_level(0)
>>>print(root)
root

Let's print all the parent emotions:

>>>print(' -> '.join([emo.get_level(i).nameforiinrange(emo.level+1)]))
root->mental-state->affective-state->emotion->negative-emotion->general-dislike->anger

How many children has the annoyance emotion?

>>>annoyance=Emotion.emotions["annoyance"]
>>>annoyance.nb_children()
5

Let's print them:

>>>Emotion.printTree(annoyance) # WNAffect should have been instantiated previously
┌pique
├frustrationannoyance┤
├displeasure
├harassment
└aggravation