Nature-Inspired-Based Approach for Automated Cyberbullying Classification on Multimedia Social Networking

Table 1

Selected attributes to classify the Tweets.

Attributes	Class	Format
Noun	CB/non-CB	Text
Pronoun	CB/non-CB	Text
Adjective	CB/non-CB	Text
Local features	The basic features extracted from a tweet	Text
Contextual features	Professional, religious, family, legal, and financial factors specific to CB	Text
Sentiment features	Positive or negative (foul words specific to CB) or direct or indirect CB	Text
Emotion features	Polite words, modal words, unknown words, number of insults and hateful blacklisted words, harming with detailed description, power differential, any form of aggression, targeting a person, targeting two or more persons, intent, repetition, one-time CB, harm, perception, reasonable person/witness, and racist sentiments	Text
Gender-specific language	Male/female	Text
User feature	Network information, user information, his/her activity information, tweet content, account creation time, and verified account time	Text/numeric
Twitter basic features	Number of followers, number of mentions, and number of following, favorite count, popularity, number of hash tags, and status count	Numeric
Linguistic features	Other languages words, punctuation marks, and abbreviated words rather than abusive sentence judgments	Text