Димитрина

Златкова

(ИИ)

Даниел

Копев

(ИИ)

Атанас

Атанасов

(ИИ)

SemEval-2018

Emoji Prediction

@

So this happened :) with my girls @user and erinblonshine. ️ #29 #sacredhearttattoo @ Sacred…

Data:

Label:

red_heart

two_hearts

blue_heart

purple_heart

camera_with_flash

camera

Classes

Our Plan

Data
Text Preprocessing
Feature Engineering
Classification
Evaluation

Text Processing

So this happened :) with my girls @user and erinblonshine. ️ #29 #sacredhearttattoo @ Sacred…

Original:

Text Processing (2)

So this happened __smile__ with my girls @user and erinblonshine. ️ #29 #sacredhearttattoo @ Sacred…

Pattern replace:

['so', 'this', 'happened', '__smile__', 'with', 'my', 'girls', 'and', 'erinblonshine', 'sacred', 'heart', 'tat', 'too', 'sacred']

Tokenize:

So this happened __smile__ with my girls and erinblonshine#sacredhearttattoo Sacred

Char filter:

Text Processing (3)

['happened', 'smile', 'girls', 'erinblonshine', 'sacred', 'heart', 'tat', 'sacred']

Stop words:

['happen', 'smile', 'girl', 'erinblonshine', 'sacred', 'heart', 'tat', 'sacred']

Lemmatize:

[('so','RB'),('this','DT'),('happened','VBD'),('smile','NN'), ('with','IN'),('my','PRP$'), ('girls','NNS'),('and','CC'), ('erinblonshine','NN'),('sacred','JJ'),('heart','NN'),('tat','VB'),('too', 'RB'),('sacred','VBD')]

POS Tagger:

Vectorization

tf-idf
GloVe
word2vec
FB research (StarSpace)

Feature Engineering

Text Features

words
unique words
stopwords
@user
Words Title
WORDS UPPER
mean word length

#
a b c
1 2 3
$
%
!!!
???

Emotions

anger
sadness
disgust
fear

joy
anticipation
surprise
trust

- fun, sun

- sun

Colors

black
blue
brown
green
grey

orange
pink
purple
red
white
yellow

- sun

- park

Sentiment

Positive:

Negative:

"pos_0", "pos_.15", "pos_.20", "pos_.27", "pos_.4", "pos_above"

"neg_0", "neg_.15", "neg_.25", "neg_.35", "neg_.6", "neg_above"

Hierarchical Twitter Clusters

^010011000

got qot gott g0t gotz qott gottt gawt ghot gotcho goht ggot

^111010100010

lmao lmfao lmaoo lmaooo lool rofl loool lmfaoo lmfaooo lmaoooo

^111010100011

haha hahaha hehe hahahaha hahah aha hehehe ahaha hah hahahah hahaa ahah

Experimental Results

	Precision	Recall	F1 Macro
Multinomial Naive Bayes	0.05	0.21	1.763
Logistic Regression with L-BFGS	0.22	0.28	13.16
MLP, 2 hidden layers, ReLU	0.26	0.26	17.898
Random Forest (50 estimators)	0.20	0.26	16.167
SVM, tf-idf	0.23	0.27	19.554
SVM, Twitter embeddings	0.16	0.18	8.522
AdaBoost, Extra Tree base	0.15	0.19	7.825
SVM+AdaBoost+Random Forest	0.25	0.24	13.764
SVM+AdaBoost+MLP	0.25	0.28	20.106